优化

src/main/java/com/aps/common/util/FileHelper.java

@@ -10,6 +10,32 @@ public class FileHelper {
     private static final String LOG_FILE = "schedule_log.txt";
     private static final String LOG_FILE_PATH = "log/";
 
+    // 日志级别
+    private static final int LOG_LEVEL_DEBUG = 0;
+    private static final int LOG_LEVEL_INFO = 1;
+    private static final int LOG_LEVEL_WARN = 2;
+    private static int currentLogLevel = LOG_LEVEL_INFO;
+
+    // 局部搜索优化
+
+    public static void log(String message) {
+        log(message, LOG_LEVEL_INFO, true);
+    }
+
+    public static void log(String message, boolean enableLogging) {
+        log(message, LOG_LEVEL_INFO, enableLogging);
+    }
+
+    public static void log(String message, int level) {
+        log(message, level, false);
+    }
+
+    public static void log(String message, int level, boolean enableLogging) {
+        if (enableLogging && level >= currentLogLevel) {
+            writeLogFile(message);
+        }
+    }
+
     public static void writeLogFile(String message) {
         String date = LocalDateTime.now().format(DateTimeFormatter.ofPattern("yyyyMMdd"))+"-";
 

src/main/java/com/aps/entity/Algorithm/Chromosome.java

@@ -133,7 +133,7 @@ public class Chromosome {
 
     private TreeMap<String, Material> materials = new TreeMap<>();
 
-   // private List<Material> materials = new ArrayList<>();
+    // private List<Material> materials = new ArrayList<>();
 
     private  List<String> materialIds = new ArrayList<>();
     /*
@@ -142,6 +142,12 @@ public class Chromosome {
     private double[] Objectives = new double[0]; // 多目标值：[Makespan, TotalFlowTime, TotalChangeover, LoadStd, Delay]
     private double[] MaxObjectives = new double[0]; //
     private double[] MinObjectives = new double[0]; //
+    private double[] weights = new double[0];
+    /**
+     * 各目标维度的理论下界（用于计算 Gap = (current - lowerBound) / lowerBound）。
+     * 由 KpiLowerBoundCalculator 在 decode 后计算并填入。
+     */
+    private double[] LowerBoundObjectives = new double[0];
     private int Rank; // 非支配排序等级（1最优）
     private double CrowdingDistance =0; // 拥挤距离 越小越优
     /*
@@ -255,7 +261,7 @@ public class Chromosome {
      *
      * @return 不在 allOperations 中的 GAScheduleResult 列表；若均存在则返回空列表
      */
-    public List<GAScheduleResult> getResultsNotInAllOperations() {
+    public List<GAScheduleResult> checkResultsNotInAllOperations1() {
         if (Result == null || Result.isEmpty()) {
             return Collections.emptyList();
         }
@@ -265,7 +271,7 @@ public class Chromosome {
         Set<Integer> allOpIds = allOperations.stream()
                 .map(com.aps.entity.basic.Entry::getId)
                 .collect(Collectors.toSet());
-      List<GAScheduleResult>  NotInAllOperations=  Result.stream()
+        List<GAScheduleResult>  NotInAllOperations=  Result.stream()
                 .filter(r -> !allOpIds.contains(r.getOperationId()))
                 .collect(Collectors.toList());
 
@@ -281,8 +287,52 @@ public class Chromosome {
      *
      * @return 存在则返回 true，否则返回 false
      */
-    public boolean hasResultNotInAllOperations() {
-        return !getResultsNotInAllOperations().isEmpty();
+//    public boolean hasResultNotInAllOperations1() {
+//      /  return !getResultsNotInAllOperations1().isEmpty();
+//    }
+
+    /**
+     * LNS 需要：创建染色体的深拷贝（machineSelection 和 operationSequencing 独立副本）
+     * 大对象（globalOpList / allOperations / orders / materials）共享引用
+     */
+    public Chromosome deepCopy() {
+        Chromosome copy = new Chromosome();
+        if (machineSelection != null) {
+            copy.machineSelection = new CopyOnWriteArrayList<>(machineSelection);
+            copy.machineStrDirty = true;
+        }
+        if (operationSequencing != null) {
+            copy.operationSequencing = new CopyOnWriteArrayList<>(operationSequencing);
+            copy.operationStrDirty = true;
+        }
+        copy.geneStrDirty = true;
+        if (Objectives != null) copy.Objectives = Arrays.copyOf(Objectives, Objectives.length);
+        if (WeightedObjectives != null) copy.WeightedObjectives = Arrays.copyOf(WeightedObjectives, WeightedObjectives.length);
+        if (fitnessLevel != null) copy.fitnessLevel = Arrays.copyOf(fitnessLevel, fitnessLevel.length);
+        if (LowerBoundObjectives != null) copy.LowerBoundObjectives = Arrays.copyOf(LowerBoundObjectives, LowerBoundObjectives.length);
+        copy.WeightedObjective = WeightedObjective;
+        copy.Makespan = Makespan;
+        copy.TotalFlowTime = TotalFlowTime;
+        copy.TotalChangeoverTime = TotalChangeoverTime;
+        copy.MachineLoadStd = MachineLoadStd;
+        copy.Fitness = Fitness;
+        copy.Rank = Rank;
+        copy.CrowdingDistance = CrowdingDistance;
+        copy.gsOrls = gsOrls;
+        copy.generateType = generateType + "_copy";
+        copy.globalParamSnapshot = globalParamSnapshot;
+        copy.objectiveWeights = objectiveWeights;
+        copy.globalOpList = globalOpList;
+        copy.allOperations = allOperations;
+        copy.orders = orders;
+        copy.InitMachines = InitMachines;
+        copy.materials = materials;
+        copy.materialIds = materialIds;
+        copy.OperatRel = OperatRel;
+        copy.orderMaterials = orderMaterials;
+        copy.Machines = Machines;
+        if (Result != null) copy.Result = new CopyOnWriteArrayList<>(Result);
+        return copy;
     }
 
 }

src/main/java/com/aps/entity/basic/Entry.java

@@ -85,6 +85,9 @@ public class Entry {
      * 工序顺序
      */
     private int sequence;
+
+    private double minProcessingTime; // 加工时间 (秒)
+
     /**
      * 可选设备列表
      */

src/main/java/com/aps/service/Algorithm/AdaptiveLargeNeighborhoodSearch.java

+package com.aps.service.Algorithm;
+
+import com.aps.common.util.FileHelper;
+import com.aps.common.util.GlobalCacheUtil;
+import com.aps.common.util.ProductionDeepCopyUtil;
+import com.aps.entity.Algorithm.*;
+import com.aps.entity.Algorithm.IDAndChildID.GroupResult;
+import com.aps.entity.basic.*;
+
+import java.util.*;
+import java.util.concurrent.CopyOnWriteArrayList;
+import java.util.stream.Collectors;
+
+/**
+ * 自适应大邻域搜索（ALNS）算法。
+ *
+ * <p>核心思想：
+ * <ol>
+ *   <li>多个 Destroy 算子（破坏解的一部分） + 多个 Repair 算子（修复被破坏的解）</li>
+ *   <li>自适应权重：根据算子表现动态调整选择概率</li>
+ *   <li>模拟退火接受准则：以一定概率接受劣解，概率随迭代递减</li>
+ * </ol>
+ *
+ * <p>与 VNS 的区别：
+ * <ul>
+ *   <li>VNS 系统性地切换邻域结构（换设备 → 工序前移 → 工序交换）</li>
+ *   <li>ALNS 随机选择 destroy/repair 组合，按权重进行 roulette wheel 选择</li>
+ * </ul>
+ *
+ * <p>架构：
+ * <pre>
+ *   ALNS.search(chromosome, tabuSearch, vns, decoder, machines)
+ *       │
+ *       ├─ 每个迭代：
+ *       │   ├─ roulette select destroy operator
+ *       │   ├─ roulette select repair operator
+ *       │   ├─ destroy(chromosome) → partial solution
+ *       │   ├─ repair(partial solution) → candidate
+ *       │   ├─ tabu check
+ *       │   ├─ localSearch(candidate)
+ *       │   ├─ SA acceptance
+ *       │   └─ update operator weights
+ *       │
+ *       └─ return best
+ * </pre>
+ *
+ * 作者：佟礼
+ */
+public class AdaptiveLargeNeighborhoodSearch {
+
+    // ==================== 随机数 ====================
+    private final Random rnd = new Random();
+
+    // ==================== ALNS 核心参数 ====================
+    private static final int MAX_ITERATIONS = 120;                    // 最大迭代次数
+    private static final int MAX_NO_IMPROVE_ITERATIONS = 25;          // 最大连续无改进迭代
+    private static final int SEGMENT_SIZE = 10;                       // 权重更新段大小（每 SEGMENT_SIZE 迭代更新一次权重）
+    private static final double DESTROY_RATIO_MIN = 0.10;             // 最小破坏比例（提高以增强探索）
+    private static final double DESTROY_RATIO_MAX = 0.40;             // 最大破坏比例
+    private static final int MAX_RETRY_ATTEMPTS = 2;                  // 最多重试次数（首次失败后换组合）
+
+    // ==================== 自适应权重参数 ====================
+    private static final double INITIAL_WEIGHT = 1.0;                 // 初始权重
+    private static final double REACTION_FACTOR = 0.7;                // 反应因子（历史权重与新分的比例）
+    private static final double MIN_WEIGHT = 0.1;                     // 最小权重（防止算子完全熄灭）
+
+    // ==================== 分数增量（越大越好） ====================
+    private static final double SCORE_NEW_BEST = 3.0;                 // 发现新的全局最优解
+    private static final double SCORE_BETTER = 1.5;                   // 比当前解更好
+    private static final double SCORE_ACCEPTED = 0.8;                 // 劣解被接受
+    private static final double SCORE_REJECTED = 0.1;                 // 劣解被拒绝
+
+    // ==================== 模拟退火参数 ====================
+    private static final double INITIAL_TEMPERATURE = 0.5;            // 初始温度
+    private static final double COOLING_RATE = 0.98;                  // 冷却速率（加快冷却，使后期更聚焦）
+    private static final double FINAL_TEMPERATURE = 0.01;             // 最终温度
+
+    // ==================== 改进判断参数 ====================
+    private static final double SIGNIFICANT_IMPROVEMENT_THRESHOLD = 1e-11;
+    private static final double MINOR_IMPROVEMENT_THRESHOLD = 0.0;
+
+    // ==================== 多轮修复参数 ====================
+    private static final int MULTI_PASS_COUNT = 3;                    // 多轮修复尝试次数
+
+    // ==================== 时间预算 ====================
+    private static final long ALNS_TIME_BUDGET_MS = 15L * 60L * 1000L;
+    private static final long ALNS_PER_ITER_BUDGET_MS = 17L * 1000L;
+
+    // ==================== 依赖项 ====================
+    private final List<Entry> allOperations;
+    private final FitnessCalculator fitnessCalculator;
+    private final List<Order> orders;
+    private final TreeMap<String, Material> materials;
+    private final List<GroupResult> entryRel;
+
+    private final Map<Integer, Entry> entryByIds;
+
+    // ==================== 缓存解码数据 ====================
+    private List<Machine> cachedMachines;
+    private List<Order> cachedOrders;
+    private List<GroupResult> cachedEntryRel;
+    private TreeMap<String, Material> cachedMaterials;
+    private List<Entry> cachedAllOperations;
+
+    // ==================== 算子注册表 ====================
+    private final List<DestroyOperator> destroyOperators = new ArrayList<>();
+    private final List<RepairOperator> repairOperators = new ArrayList<>();
+    private double[] destroyWeights;
+    private double[] destroyScores;
+    private int[] destroyUseCount;
+    private double[] repairWeights;
+    private double[] repairScores;
+    private int[] repairUseCount;
+
+    // ==================== 构造函数 ====================
+
+    public AdaptiveLargeNeighborhoodSearch(List<Entry> allOperations, List<Order> orders,
+                                           TreeMap<String, Material> materials,
+                                           List<GroupResult> entryRel,
+                                           FitnessCalculator fitnessCalculator) {
+        this.allOperations = allOperations;
+        this.fitnessCalculator = fitnessCalculator;
+        this.orders = orders;
+        this.materials = materials;
+        this.entryRel = entryRel;
+
+        Map<Integer, Object> mp = buildEntryKey();
+
+        this.entryByIds = (Map<Integer, Entry>) mp.get(1);
+
+        // 预缓存解码数据
+        this.cachedAllOperations = ProductionDeepCopyUtil.deepCopyList(
+                new CopyOnWriteArrayList<>(allOperations), Entry.class);
+        this.cachedOrders = ProductionDeepCopyUtil.deepCopyList(
+                new CopyOnWriteArrayList<>(orders), Order.class);
+        this.cachedEntryRel = ProductionDeepCopyUtil.deepCopyList(
+                new CopyOnWriteArrayList<>(entryRel), GroupResult.class);
+        this.cachedMaterials = ProductionDeepCopyUtil.deepCopyTreeMap(
+                materials, String.class, Material.class);
+
+        // 注册 destroy 算子
+        registerDestroyOperators();
+        // 注册 repair 算子
+        registerRepairOperators();
+        // 初始化权重
+        initWeights();
+    }
+
+    // ==================== 算子注册 ====================
+
+    private void registerDestroyOperators() {
+        // 1. 随机换设备：随机选择部分工序更换机器
+        destroyOperators.add(new DestroyOperator("RandomMachineChange", this::destroyRandomMachineChange));
+        // 2. 瓶颈工序移除：移除瓶颈设备上的工序
+        destroyOperators.add(new DestroyOperator("BottleneckOpRemoval", this::destroyBottleneckOps));
+        // 3. 延迟工序打乱：打乱导致延迟的工序
+        destroyOperators.add(new DestroyOperator("DelayOpShuffle", this::destroyDelayOps));
+        // 4. 随机区间打乱：随机打乱一段连续的工序序列
+        destroyOperators.add(new DestroyOperator("RandomSegmentShuffle", this::destroyRandomSegment));
+        // 5. 负载均衡迁移：将超载机器上的工序迁移到低负载机器
+        destroyOperators.add(new DestroyOperator("LoadBalanceTransfer", this::destroyLoadBalanceTransfer));
+    }
+
+    private void registerRepairOperators() {
+        // 1. 多轮修复：多次调用 VNS 邻域生成，选最优结果
+        repairOperators.add(new RepairOperator("MultiPassRepair", this::repairMultiPass));
+        // 2. 贪婪换设备：对打乱的工序重新选择最优机器
+        repairOperators.add(new RepairOperator("GreedyMachineReassign", this::repairGreedyMachine));
+        // 3. 局部搜索修复：解码后通过 VNS 生成邻域
+        repairOperators.add(new RepairOperator("LocalSearchRepair", this::repairLocalSearch));
+    }
+
+    private void initWeights() {
+        destroyWeights = new double[destroyOperators.size()];
+        destroyScores = new double[destroyOperators.size()];
+        destroyUseCount = new int[destroyOperators.size()];
+        Arrays.fill(destroyWeights, INITIAL_WEIGHT);
+
+        repairWeights = new double[repairOperators.size()];
+        repairScores = new double[repairOperators.size()];
+        repairUseCount = new int[repairOperators.size()];
+        Arrays.fill(repairWeights, INITIAL_WEIGHT);
+    }
+
+    // ====================================================================
+    // 搜索主循环
+    // ====================================================================
+
+    /**
+     * ALNS 搜索主循环。
+     *
+     * @param chromosome 初始解
+     * @param tabuSearch 禁忌表（多算法共享）
+     * @param vns        提供邻域生成与局部搜索
+     * @param decoder    解码器
+     * @param machines   机器列表
+     * @return 优化后的最优解
+     */
+    public Chromosome search(Chromosome chromosome, TabuSearch tabuSearch,
+                             VariableNeighborhoodSearch vns,
+                             GeneticDecoder decoder, List<Machine> machines) {
+        FileHelper.writeLogFile("ALNS - 开始执行");
+
+        Chromosome current = ProductionDeepCopyUtil.deepCopy(chromosome, Chromosome.class);
+        Chromosome best = ProductionDeepCopyUtil.deepCopy(chromosome, Chromosome.class);
+        double currentBestFitness = best.getFitness();
+
+        int iterations = 0;
+        int improveCount = 0;
+        int noImprovementCount = 0;
+        double temperature = INITIAL_TEMPERATURE;
+
+        // 时间预算
+        long startTimeMs = System.currentTimeMillis();
+        long remainingBudgetMs = Math.max(5L * 60L * 1000L, ALNS_TIME_BUDGET_MS / 2);
+        int timeBasedMaxIter = (int) Math.max(20, remainingBudgetMs / ALNS_PER_ITER_BUDGET_MS);
+        int maxIterations = Math.min(MAX_ITERATIONS, Math.max(40, timeBasedMaxIter));
+
+        FileHelper.writeLogFile(String.format(
+                "ALNS - 参数: 最大迭代=%d, 初始温度=%.3f, 冷却率=%.3f, 破坏比例=%.0f%%-%.0f%%, 时间预算=%.1fmin",
+                maxIterations, INITIAL_TEMPERATURE, COOLING_RATE,
+                DESTROY_RATIO_MIN * 100, DESTROY_RATIO_MAX * 100,
+                (double) remainingBudgetMs / 60000.0));
+
+        for (int iter = 0; iter < maxIterations; iter++) {
+            iterations++;
+            decoder.DelOrder(current);
+
+            // ---- 1. 自适应选择 destroy + repair 算子 ----
+            int destroyIdx = rouletteSelect(destroyWeights);
+            int repairIdx = rouletteSelect(repairWeights);
+            DestroyOperator destroyOp = destroyOperators.get(destroyIdx);
+            RepairOperator repairOp = repairOperators.get(repairIdx);
+
+            Chromosome repaired = null;
+            int repairAttempt = 0;
+
+            // ---- 重试机制：最多尝试 MAX_RETRY_ATTEMPTS 次不同的 destroy+repair 组合 ----
+            for (; repairAttempt < MAX_RETRY_ATTEMPTS; repairAttempt++) {
+                // ---- 2. Destroy：破坏当前解 ----
+                Chromosome destroyed = destroyOp.apply(current, DESTROY_RATIO_MIN, DESTROY_RATIO_MAX);
+                if (destroyed == null) {
+                    // 换其他 destroy 算子重试
+                    int newDestroyIdx = (destroyIdx + 1 + rnd.nextInt(destroyOperators.size() - 1)) % destroyOperators.size();
+                    destroyIdx = newDestroyIdx;
+                    destroyOp = destroyOperators.get(destroyIdx);
+                    continue;
+                }
+
+                // ---- 3. Repair：修复被破坏的解 ----
+                repaired = repairOp.apply(destroyed, vns, decoder, machines);
+                if (repaired != null) {
+                    break; // 修复成功，退出重试
+                }
+                // 修复失败，换算子重试
+                int newRepairIdx = (repairIdx + 1 + rnd.nextInt(repairOperators.size() - 1)) % repairOperators.size();
+                repairIdx = newRepairIdx;
+                repairOp = repairOperators.get(repairIdx);
+            }
+
+            if (repaired == null) {
+                noImprovementCount++;
+                updateOperatorScores(destroyIdx, repairIdx, SCORE_REJECTED, false);
+                continue;
+            }
+
+            // ---- 4. 禁忌检查 ----
+            boolean tabuHit = tabuSearch.isChromosomeTabu(repaired);
+
+            // ---- 5. 解码 ----
+            decode(decoder, repaired, machines);
+            tabuSearch.addChromosomeToTabu(repaired);
+
+            // ---- 6. 接受准则 ----
+            boolean betterThanBest = isBetter(repaired, best);
+            boolean betterThanCurrent = isBetter(repaired, current);
+
+            boolean accept;
+            double score;
+
+            if (betterThanBest) {
+                // 渴望准则：无条件接受
+                accept = true;
+                score = SCORE_NEW_BEST;
+            } else if (betterThanCurrent) {
+                accept = true;
+                score = SCORE_BETTER;
+            } else if (!tabuHit) {
+                // 模拟退火：接受劣解
+                double delta = current.getFitness() - repaired.getFitness(); // 负值表示更差
+                double acceptProb = Math.exp(delta / temperature);
+                accept = rnd.nextDouble() < acceptProb;
+                score = accept ? SCORE_ACCEPTED : SCORE_REJECTED;
+            } else {
+                accept = false;
+                score = SCORE_REJECTED;
+            }
+
+            updateOperatorScores(destroyIdx, repairIdx, score, true);
+
+            if (accept) {
+                current = lightCopy(repaired);
+                if (betterThanBest) {
+                    best = lightCopy(repaired);
+                    improveCount++;
+                    double delta = best.getFitness() - currentBestFitness;
+                    if (delta > MINOR_IMPROVEMENT_THRESHOLD) {
+                        noImprovementCount = 0;
+                        currentBestFitness = best.getFitness();
+                        if (delta > SIGNIFICANT_IMPROVEMENT_THRESHOLD) {
+                            FileHelper.writeLogFile(String.format(
+                                    "ALNS - 找到更好解(显著), 迭代=%d, fitness=%.12f, destroy=%s, repair=%s",
+                                    iterations, best.getFitness(), destroyOp.name, repairOp.name));
+                        } else {
+                            FileHelper.writeLogFile(String.format(
+                                    "ALNS - 找到更好解(微小), 迭代=%d, fitness=%.12f, delta=%.2e, destroy=%s, repair=%s",
+                                    iterations, best.getFitness(), delta, destroyOp.name, repairOp.name));
+                        }
+                    }
+                }
+            } else {
+                noImprovementCount++;
+            }
+
+            // ---- 7. 温度冷却 ----
+            temperature *= COOLING_RATE;
+            if (temperature < FINAL_TEMPERATURE) {
+                temperature = FINAL_TEMPERATURE;
+            }
+
+            // ---- 8. 段更新权重 ----
+            if ((iter + 1) % SEGMENT_SIZE == 0) {
+                updateWeights();
+            }
+
+            // ---- 9. 提前停止 ----
+            if (noImprovementCount >= MAX_NO_IMPROVE_ITERATIONS) {
+                FileHelper.writeLogFile(String.format(
+                        "ALNS - 提前停止: 连续%d次无改进", MAX_NO_IMPROVE_ITERATIONS));
+                break;
+            }
+            long elapsedMs = System.currentTimeMillis() - startTimeMs;
+            if (elapsedMs > remainingBudgetMs) {
+                FileHelper.writeLogFile(String.format(
+                        "ALNS - 提前停止: 达到时间预算(%.1fmin)", elapsedMs / 60000.0));
+                break;
+            }
+
+            // 每 10 次迭代输出一次状态
+            if ((iter + 1) % 10 == 0) {
+                FileHelper.writeLogFile(String.format(
+                        "ALNS - 迭代%d/%d, 改进=%d, 无改进连续=%d, T=%.4f, fitness=%.12f",
+                        iterations, maxIterations, improveCount, noImprovementCount,
+                        temperature, best.getFitness()));
+            }
+        }
+
+        // 最终权重更新
+        updateWeights();
+
+        FileHelper.writeLogFile(String.format(
+                "ALNS - 结束: 总迭代=%d, 改进次数=%d, 最终fitness=%.12f, 最终温度=%.4f",
+                iterations, improveCount, best.getFitness(), temperature));
+        FileHelper.writeLogFile(KpiLowerBoundCalculator.generateGapReport(best));
+        logOperatorStats();
+        return best;
+    }
+    /**
+     * 轻量拷贝：只复制 generateNeighbor/DelOrder 需要的字段，避免全量 JSON 深拷贝导致 OOM。
+     * result/machines/operatRel 等重型数据共享引用（generateNeighbor 只读，不修改）。
+     */
+    private Chromosome lightCopy(Chromosome source) {
+        Chromosome copy = new Chromosome();
+        copy.setOperationSequencing(new CopyOnWriteArrayList<>(source.getOperationSequencing()));
+        copy.setMachineSelection(new CopyOnWriteArrayList<>(source.getMachineSelection()));
+        copy.setGlobalOpList(new CopyOnWriteArrayList<>(source.getGlobalOpList()));
+        copy.setOrders(new CopyOnWriteArrayList<>(source.getOrders()));
+        copy.setAllOperations(new CopyOnWriteArrayList<>(source.getAllOperations()));
+        copy.setResult(source.getResult());
+        copy.setMachines(source.getMachines());
+        copy.setOperatRel(new CopyOnWriteArrayList<>(source.getOperatRel()));
+        copy.setScenarioID(source.getScenarioID());
+        copy.setBaseTime(source.getBaseTime());
+        copy.setGenerateType(source.getGenerateType());
+        copy.setFitnessLevel(source.getFitnessLevel());
+        copy.setFitness(source.getFitness());
+
+        return copy;
+    }
+    // ====================================================================
+    // Destroy 算子（破坏部分解）
+    // ====================================================================
+
+    /**
+     * Destroy 1: 随机换设备。
+     * 随机选择 destroyRatio 比例的工序，为其更换机器（如果有多个机器选项）。
+     */
+    private Chromosome destroyRandomMachineChange(Chromosome c, double destroyRatioMin, double destroyRatioMax) {
+        List<GAScheduleResult> results = c.getResult();
+        if (results == null || results.isEmpty()) return c;
+
+        // 只选择有多个机器选项的工序
+        List<GAScheduleResult> candidates = results.stream()
+                .filter(r -> {
+                    Entry e = entryByIds.get(r.getOperationId());
+                    return e != null && e.getMachineOptions() != null && e.getMachineOptions().size() > 1;
+                })
+                .collect(Collectors.toList());
+        if (candidates.isEmpty()) return c;
+
+        double ratio = destroyRatioMin + rnd.nextDouble() * (destroyRatioMax - destroyRatioMin);
+        int destroyCount = Math.max(1, (int) (candidates.size() * ratio));
+        Collections.shuffle(candidates, rnd);
+
+        Chromosome copy = copyChromosome(c);
+        CopyOnWriteArrayList<Integer> machineSel = copy.getMachineSelection();
+        if (machineSel == null) return copy;
+
+        // 构建正确的 machineSelection 位置索引：groupId_sequence → globalOpList 索引
+        Map<String, Integer> machinePosIndex = new HashMap<>();
+        List<GlobalOperationInfo> globalOpList = c.getGlobalOpList();
+        for (int i = 0; i < globalOpList.size(); i++) {
+            Entry op = globalOpList.get(i).getOp();
+            machinePosIndex.put(op.getGroupId() + "_" + op.getSequence(), i);
+        }
+
+        for (int i = 0; i < Math.min(destroyCount, candidates.size()); i++) {
+            GAScheduleResult sr = candidates.get(i);
+            Entry entry = entryByIds.get(sr.getOperationId());
+            if (entry == null || entry.getMachineOptions() == null) continue;
+            List<MachineOption> options = entry.getMachineOptions();
+            if (options.size() <= 1) continue;
+            String key = entry.getGroupId() + "_" + entry.getSequence();
+            Integer pos = machinePosIndex.get(key);
+            if (pos == null || pos >= machineSel.size()) continue;
+            // 随机选一个不同的机器（machineSelection 存的是 1-based 序号，不是 machineId）
+            int currentIdx = machineSel.get(pos) - 1; // 转为 0-based
+            int newIdx;
+            if (options.size() == 2) {
+                newIdx = (currentIdx == 0) ? 1 : 0;
+            } else {
+                do {
+                    newIdx = rnd.nextInt(options.size());
+                } while (newIdx == currentIdx);
+            }
+            machineSel.set(pos, newIdx + 1);
+        }
+        return copy;
+    }
+
+    /**
+     * Destroy 2: 瓶颈工序移除。
+     * 找到瓶颈设备上的工序，打乱它们在 operationSequencing 中的位置。
+     */
+    private Chromosome destroyBottleneckOps(Chromosome c, double destroyRatioMin, double destroyRatioMax) {
+        List<GAScheduleResult> results = c.getResult();
+        if (results == null || results.isEmpty()) return c;
+
+        // 简单瓶颈识别：利用率最高的机器
+        Map<Long, Double> utilMap = new HashMap<>();
+        Map<Long, Long> timeMap = new HashMap<>();
+        for (GAScheduleResult r : results) {
+            Long mid = r.getMachineId();
+            utilMap.merge(mid, r.getProcessingTime(), Double::sum);
+            timeMap.merge(mid, (long) (r.getEndTime() - r.getStartTime()), Long::sum);
+        }
+        Long bottleneckMachineId = utilMap.entrySet().stream()
+                .max(Map.Entry.comparingByValue())
+                .map(Map.Entry::getKey)
+                .orElse(null);
+        if (bottleneckMachineId == null) return c;
+
+        List<GAScheduleResult> bottleneckOps = results.stream()
+                .filter(r -> bottleneckMachineId.equals(r.getMachineId()))
+                .collect(Collectors.toList());
+        if (bottleneckOps.size() < 2) return c;
+
+        double ratio = destroyRatioMin + rnd.nextDouble() * (destroyRatioMax - destroyRatioMin);
+        int destroyCount = Math.max(2, (int) (bottleneckOps.size() * ratio));
+
+        Chromosome copy = copyChromosome(c);
+        CopyOnWriteArrayList<Integer> opSeq = copy.getOperationSequencing();
+        if (opSeq == null) return copy;
+
+        // 找到瓶颈工序在 opSeq 中的位置
+        Set<Integer> bottleneckEntryIds = bottleneckOps.stream()
+                .map(GAScheduleResult::getOperationId)
+                .collect(Collectors.toSet());
+        List<Integer> bottleneckPositions = new ArrayList<>();
+        for (int i = 0; i < opSeq.size(); i++) {
+            if (bottleneckEntryIds.contains(opSeq.get(i))) {
+                bottleneckPositions.add(i);
+            }
+        }
+        if (bottleneckPositions.size() < 2) return copy;
+
+        // 随机选择 destroyCount 个位置，打乱它们的值
+        Collections.shuffle(bottleneckPositions, rnd);
+        List<Integer> selectedPositions = bottleneckPositions.subList(0,
+                Math.min(destroyCount, bottleneckPositions.size()));
+        List<Integer> values = new ArrayList<>();
+        for (int pos : selectedPositions) {
+            values.add(opSeq.get(pos));
+        }
+        Collections.shuffle(values, rnd);
+        for (int j = 0; j < selectedPositions.size(); j++) {
+            opSeq.set(selectedPositions.get(j), values.get(j));
+        }
+        return copy;
+    }
+
+    /**
+     * Destroy 3: 延迟工序打乱。
+     * 找到导致延迟的工序，打乱它们的机器选择。
+     */
+    private Chromosome destroyDelayOps(Chromosome c, double destroyRatioMin, double destroyRatioMax) {
+        List<Order> orderList = c.getOrders();
+        if (orderList == null || orderList.isEmpty()) return c;
+
+        // 找到延迟的订单
+        Set<String> delayedOrderIds = orderList.stream()
+                .filter(o -> o.getDelayHours() > 0)
+                .map(Order::getOrderId)
+                .collect(Collectors.toSet());
+        if (delayedOrderIds.isEmpty()) return c;
+
+        List<GAScheduleResult> results = c.getResult();
+        if (results == null || results.isEmpty()) return c;
+
+        List<GAScheduleResult> delayedResults = results.stream()
+                .filter(r -> delayedOrderIds.contains(r.getOrderId()))
+                .collect(Collectors.toList());
+        if (delayedResults.isEmpty()) return c;
+
+        double ratio = destroyRatioMin + rnd.nextDouble() * (destroyRatioMax - destroyRatioMin);
+        int destroyCount = Math.max(1, (int) (delayedResults.size() * ratio));
+        Collections.shuffle(delayedResults, rnd);
+
+        Chromosome copy = copyChromosome(c);
+        CopyOnWriteArrayList<Integer> machineSel = copy.getMachineSelection();
+        if (machineSel == null) return copy;
+
+        // 构建正确的 machineSelection 位置索引：groupId_sequence → globalOpList 索引
+        Map<String, Integer> machinePosIndex = new HashMap<>();
+        List<GlobalOperationInfo> globalOpList = c.getGlobalOpList();
+        for (int i = 0; i < globalOpList.size(); i++) {
+            Entry op = globalOpList.get(i).getOp();
+            machinePosIndex.put(op.getGroupId() + "_" + op.getSequence(), i);
+        }
+
+        for (int i = 0; i < Math.min(destroyCount, delayedResults.size()); i++) {
+            GAScheduleResult sr = delayedResults.get(i);
+            Entry entry = entryByIds.get(sr.getOperationId());
+            if (entry == null || entry.getMachineOptions() == null || entry.getMachineOptions().isEmpty())
+                continue;
+            List<MachineOption> options = entry.getMachineOptions();
+            if (options.size() <= 1) continue;
+            String key = entry.getGroupId() + "_" + entry.getSequence();
+            Integer pos = machinePosIndex.get(key);
+            if (pos == null || pos >= machineSel.size()) continue;
+            // 随机选一个不同的机器（machineSelection 存的是 1-based 序号，不是 machineId）
+            int currentIdx = machineSel.get(pos) - 1;
+            int newIdx;
+            if (options.size() == 2) {
+                newIdx = (currentIdx == 0) ? 1 : 0;
+            } else {
+                do {
+                    newIdx = rnd.nextInt(options.size());
+                } while (newIdx == currentIdx);
+            }
+            machineSel.set(pos, newIdx + 1);
+        }
+        return copy;
+    }
+
+    /**
+     * Destroy 4: 随机区间打乱。
+     * 在 operationSequencing 中随机选一段连续区间，打乱其内部顺序。
+     */
+    private Chromosome destroyRandomSegment(Chromosome c, double destroyRatioMin, double destroyRatioMax) {
+        CopyOnWriteArrayList<Integer> opSeq = c.getOperationSequencing();
+        if (opSeq == null || opSeq.size() < 4) return c;
+
+        Chromosome copy = copyChromosome(c);
+        CopyOnWriteArrayList<Integer> seq = copy.getOperationSequencing();
+        if (seq == null) return copy;
+
+        double ratio = destroyRatioMin + rnd.nextDouble() * (destroyRatioMax - destroyRatioMin);
+        int segmentLen = Math.max(2, (int) (seq.size() * ratio));
+        int start = rnd.nextInt(Math.max(1, seq.size() - segmentLen));
+        int end = Math.min(seq.size(), start + segmentLen);
+
+        // 打乱 [start, end) 区间
+        List<Integer> subList = new ArrayList<>(seq.subList(start, end));
+        Collections.shuffle(subList, rnd);
+        for (int i = start; i < end; i++) {
+            seq.set(i, subList.get(i - start));
+        }
+        return copy;
+    }
+
+    /**
+     * Destroy 5: 负载均衡迁移。
+     * 利用解码后的调度数据，找出利用率最高和最低的机器，
+     * 将超载机器上有多机器选项的工序迁移到低负载机器上。
+     */
+    private Chromosome destroyLoadBalanceTransfer(Chromosome c, double destroyRatioMin, double destroyRatioMax) {
+        List<GAScheduleResult> results = c.getResult();
+        if (results == null || results.isEmpty()) return c;
+
+        // 构建每台机器的利用率数据
+        Map<Long, MachineUtilInfo> machineUtilMap = buildMachineUtilization(results);
+        if (machineUtilMap.size() < 2) return c;
+
+        // 找出超载机器（利用率 > 70%）和低负载机器（利用率 < 40%）
+        List<Long> overloadedMachines = new ArrayList<>();
+        List<Long> underloadedMachines = new ArrayList<>();
+        double totalSpan = 0;
+        for (MachineUtilInfo info : machineUtilMap.values()) {
+            totalSpan = Math.max(totalSpan, info.span);
+        }
+        for (Map.Entry<Long, MachineUtilInfo> entry : machineUtilMap.entrySet()) {
+            double util = totalSpan > 0 ? entry.getValue().span / totalSpan * 100 : 0;
+            if (util > 70) overloadedMachines.add(entry.getKey());
+            if (util < 40) underloadedMachines.add(entry.getKey());
+        }
+        if (overloadedMachines.isEmpty() || underloadedMachines.isEmpty()) return c;
+
+        // 收集超载机器上有多个机器选项的工序
+        Set<Long> overloadedSet = new HashSet<>(overloadedMachines);
+        List<GAScheduleResult> candidates = results.stream()
+                .filter(r -> overloadedSet.contains(r.getMachineId()))
+                .filter(r -> {
+                    Entry e = entryByIds.get(r.getOperationId());
+                    return e != null && e.getMachineOptions() != null && e.getMachineOptions().size() > 1;
+                })
+                .collect(Collectors.toList());
+        if (candidates.isEmpty()) return c;
+
+        double ratio = destroyRatioMin + rnd.nextDouble() * (destroyRatioMax - destroyRatioMin);
+        int destroyCount = Math.max(1, (int) (candidates.size() * ratio));
+        Collections.shuffle(candidates, rnd);
+
+        Chromosome copy = copyChromosome(c);
+        CopyOnWriteArrayList<Integer> machineSel = copy.getMachineSelection();
+        if (machineSel == null) return copy;
+
+        // 构建 machineSelection 位置索引
+        Map<String, Integer> machinePosIndex = new HashMap<>();
+        List<GlobalOperationInfo> globalOpList = c.getGlobalOpList();
+        for (int i = 0; i < globalOpList.size(); i++) {
+            Entry op = globalOpList.get(i).getOp();
+            machinePosIndex.put(op.getGroupId() + "_" + op.getSequence(), i);
+        }
+
+        // 构建低负载机器可用的 machineId 集合
+        Set<Long> underloadedMachineIds = new HashSet<>(underloadedMachines);
+
+        int transferred = 0;
+        for (int i = 0; i < Math.min(destroyCount, candidates.size()); i++) {
+            GAScheduleResult sr = candidates.get(i);
+            Entry entry = entryByIds.get(sr.getOperationId());
+            if (entry == null || entry.getMachineOptions() == null) continue;
+            List<MachineOption> options = entry.getMachineOptions();
+            if (options.size() <= 1) continue;
+
+            // 找到该工序可用的低负载机器选项
+            List<Integer> lowLoadIndices = new ArrayList<>();
+            for (int j = 0; j < options.size(); j++) {
+                if (underloadedMachineIds.contains(options.get(j).getMachineId())) {
+                    lowLoadIndices.add(j);
+                }
+            }
+            if (lowLoadIndices.isEmpty()) continue;
+
+            String key = entry.getGroupId() + "_" + entry.getSequence();
+            Integer pos = machinePosIndex.get(key);
+            if (pos == null || pos >= machineSel.size()) continue;
+
+            // 随机选一个低负载机器
+            int newIdx = lowLoadIndices.get(rnd.nextInt(lowLoadIndices.size()));
+            machineSel.set(pos, newIdx + 1);
+            transferred++;
+        }
+
+        if (transferred > 0) {
+            FileHelper.writeLogFile(String.format(
+                    "ALNS-负载均衡迁移: 超载机器=%d台, 低负载机器=%d台, 转移工序=%d/%d",
+                    overloadedMachines.size(), underloadedMachines.size(), transferred, destroyCount));
+        }
+        return copy;
+    }
+
+    // ====================================================================
+    // Repair 算子（修复被破坏的解）
+    // ====================================================================
+
+    /**
+     * Repair 1: 多轮修复。
+     * 多次调用 VNS 的 generateNeighbor，每次解码并比较，选最优结果。
+     * 这比单次 generateNeighbor 更有可能找到改进。
+     */
+    private Chromosome repairMultiPass(Chromosome c, VariableNeighborhoodSearch vns,
+                                       GeneticDecoder decoder, List<Machine> machines) {
+        Chromosome base = copyChromosome(c);
+        decode(decoder, base, machines);
+
+        Chromosome bestRepair = base;
+        double bestRepairFitness = base.getFitness();
+
+        for (int pass = 0; pass < MULTI_PASS_COUNT; pass++) {
+            // 基于当前最优修复结果生成邻域
+            Chromosome neighbor = vns.generateNeighbor(bestRepair);
+            if (neighbor == null) continue;
+
+            // 解码邻域
+            Chromosome neighborCopy = copyChromosome(neighbor);
+            neighborCopy.setResult(new CopyOnWriteArrayList<>());
+            decode(decoder, neighborCopy, machines);
+
+            if (isBetter(neighborCopy, bestRepair)) {
+                bestRepair = neighborCopy;
+                bestRepairFitness = bestRepair.getFitness();
+            }
+        }
+        return bestRepair;
+    }
+
+    /**
+     * Repair 2: 贪婪换设备。
+     * 通过 VNS 的 generateNeighbor 生成邻居后，再调用局部搜索修复。
+     */
+    private Chromosome repairGreedyMachine(Chromosome c, VariableNeighborhoodSearch vns,
+                                           GeneticDecoder decoder, List<Machine> machines) {
+        // 通过 VNS 的瓶颈感知策略生成一个邻居
+        Chromosome neighbor = vns.generateNeighbor(c);
+        if (neighbor == null) return c;
+        return neighbor;
+    }
+
+    /**
+     * Repair 3: 局部搜索修复。
+     * 通过 VNS 的局部搜索对解进行修复优化。
+     */
+    private Chromosome repairLocalSearch(Chromosome c, VariableNeighborhoodSearch vns,
+                                         GeneticDecoder decoder, List<Machine> machines) {
+        // 深拷贝后解码，再通过 VNS 的 generateNeighbor 生成一个邻居
+        Chromosome copy = copyChromosome(c);
+        decode(decoder, copy, machines);
+        Chromosome neighbor = vns.generateNeighbor(copy);
+        return neighbor != null ? neighbor : copy;
+    }
+
+    // ====================================================================
+    // 自适应权重管理
+    // ====================================================================
+
+    /**
+     * 轮盘赌选择算子。
+     */
+    private int rouletteSelect(double[] weights) {
+        double total = 0.0;
+        for (double w : weights) total += w;
+        if (total <= 0) return rnd.nextInt(weights.length);
+        double rand = rnd.nextDouble() * total;
+        double cumulative = 0.0;
+        for (int i = 0; i < weights.length; i++) {
+            cumulative += weights[i];
+            if (rand <= cumulative) return i;
+        }
+        return weights.length - 1;
+    }
+
+    /**
+     * 更新算子分数（累加到此段的 score 中）。
+     */
+    private void updateOperatorScores(int destroyIdx, int repairIdx, double score, boolean used) {
+        destroyScores[destroyIdx] += score;
+        repairScores[repairIdx] += score;
+        if (used) {
+            destroyUseCount[destroyIdx]++;
+            repairUseCount[repairIdx]++;
+        }
+    }
+
+    /**
+     * 段结束时更新权重。
+     * 权重 = reactionFactor * 旧权重 + (1 - reactionFactor) * (段分数 / 使用次数)
+     */
+    private void updateWeights() {
+        for (int i = 0; i < destroyWeights.length; i++) {
+            double avgScore = destroyUseCount[i] > 0
+                    ? destroyScores[i] / destroyUseCount[i] : 0.0;
+            destroyWeights[i] = Math.max(MIN_WEIGHT,
+                    REACTION_FACTOR * destroyWeights[i] + (1 - REACTION_FACTOR) * avgScore);
+            destroyScores[i] = 0.0;
+            destroyUseCount[i] = 0;
+        }
+        for (int i = 0; i < repairWeights.length; i++) {
+            double avgScore = repairUseCount[i] > 0
+                    ? repairScores[i] / repairUseCount[i] : 0.0;
+            repairWeights[i] = Math.max(MIN_WEIGHT,
+                    REACTION_FACTOR * repairWeights[i] + (1 - REACTION_FACTOR) * avgScore);
+            repairScores[i] = 0.0;
+            repairUseCount[i] = 0;
+        }
+    }
+
+    /**
+     * 输出算子权重统计。
+     */
+    private void logOperatorStats() {
+        StringBuilder sb = new StringBuilder("ALNS - 算子权重: ");
+        sb.append("Destroy[");
+        for (int i = 0; i < destroyOperators.size(); i++) {
+            sb.append(String.format("%s=%.3f", destroyOperators.get(i).name, destroyWeights[i]));
+            if (i < destroyOperators.size() - 1) sb.append(", ");
+        }
+        sb.append("] Repair[");
+        for (int i = 0; i < repairOperators.size(); i++) {
+            sb.append(String.format("%s=%.3f", repairOperators.get(i).name, repairWeights[i]));
+            if (i < repairOperators.size() - 1) sb.append(", ");
+        }
+        sb.append("]");
+        FileHelper.writeLogFile(sb.toString());
+    }
+
+    // ====================================================================
+    // 辅助方法（复用 VNS 风格）
+    // ====================================================================
+
+    /**
+     * 从调度结果构建每台机器的利用率数据。
+     */
+    private Map<Long, MachineUtilInfo> buildMachineUtilization(List<GAScheduleResult> results) {
+        Map<Long, MachineUtilInfo> map = new HashMap<>();
+        for (GAScheduleResult r : results) {
+            Long mid = r.getMachineId();
+            MachineUtilInfo info = map.get(mid);
+            if (info == null) {
+                info = new MachineUtilInfo();
+                info.machineId = mid;
+                info.minStart = r.getStartTime();
+                info.maxEnd = r.getEndTime();
+                info.opCount = 1;
+                info.totalWork = r.getProcessingTime();
+                map.put(mid, info);
+            } else {
+                info.minStart = Math.min(info.minStart, r.getStartTime());
+                info.maxEnd = Math.max(info.maxEnd, r.getEndTime());
+                info.opCount++;
+                info.totalWork += r.getProcessingTime();
+            }
+        }
+        // 计算每台机器的 span
+        for (MachineUtilInfo info : map.values()) {
+            info.span = info.maxEnd - info.minStart;
+        }
+        return map;
+    }
+
+    private void decode(GeneticDecoder decoder, Chromosome chromosome, List<Machine> machines) {
+        chromosome.setResult(new CopyOnWriteArrayList<>());
+        if (cachedMachines == null) {
+            cachedMachines = ProductionDeepCopyUtil.deepCopyList(machines, Machine.class);
+        }
+        chromosome.setMachines(ProductionDeepCopyUtil.deepCopyList(cachedMachines, Machine.class));
+        chromosome.setOrders(ProductionDeepCopyUtil.deepCopyList(
+                new CopyOnWriteArrayList<>(cachedOrders), Order.class));
+        chromosome.setOperatRel(ProductionDeepCopyUtil.deepCopyList(
+                new CopyOnWriteArrayList<>(cachedEntryRel), GroupResult.class));
+        chromosome.setMaterials(ProductionDeepCopyUtil.deepCopyTreeMap(
+                cachedMaterials, String.class, Material.class));
+        chromosome.setAllOperations(ProductionDeepCopyUtil.deepCopyList(
+                new CopyOnWriteArrayList<>(cachedAllOperations), Entry.class));
+
+        List<GAScheduleResult> lockedOrders = GlobalCacheUtil.get("locked_orders_" + chromosome.getScenarioID());
+        if (lockedOrders != null && !lockedOrders.isEmpty()) {
+            chromosome.setResultOld(ProductionDeepCopyUtil.deepCopyList(lockedOrders, GAScheduleResult.class));
+        } else {
+            chromosome.setResultOld(new CopyOnWriteArrayList<>());
+        }
+        decoder.decodeChromosomeWithCache(chromosome, false);
+    }
+
+    private Chromosome copyChromosome(Chromosome c) {
+        Chromosome copy = new Chromosome();
+        if (c.getMachineSelection() != null) {
+            copy.setMachineSelection(new CopyOnWriteArrayList<>(c.getMachineSelection()));
+        }
+        if (c.getOperationSequencing() != null) {
+            copy.setOperationSequencing(new CopyOnWriteArrayList<>(c.getOperationSequencing()));
+        }
+        if (c.getObjectives() != null) {
+            copy.setObjectives(Arrays.copyOf(c.getObjectives(), c.getObjectives().length));
+        }
+        if (c.getGlobalOpList() != null) {
+            copy.setGlobalOpList(new CopyOnWriteArrayList<>(c.getGlobalOpList()));
+        }
+        copy.setMachines(c.getMachines());
+        copy.setOrders(c.getOrders());
+        copy.setOperatRel(c.getOperatRel());
+        copy.setMaterials(c.getMaterials());
+        copy.setAllOperations(c.getAllOperations());
+        copy.setFitness(c.getFitness());
+        copy.setFitnessLevel(c.getFitnessLevel());
+        copy.setResult(c.getResult());
+        copy.setScenarioID(c.getScenarioID());
+        copy.setBaseTime(c.getBaseTime());
+        return copy;
+    }
+
+    private boolean isBetter(Chromosome c1, Chromosome c2) {
+        return fitnessCalculator.isBetter(c1, c2);
+    }
+
+    private Map<Integer, Object> buildEntryKey() {
+        Map<String, Entry> entryMap = new HashMap<>();
+        Map<Integer, Entry> entryByIdMap = new HashMap<>();
+        for (Entry entry : allOperations) {
+
+
+                entryByIdMap.put(entry.getId(), entry);
+
+        }
+        Map<Integer, Object> result = new HashMap<>();
+
+        result.put(1, entryByIdMap);
+        return result;
+    }
+
+    // ====================================================================
+    // 内部类
+    // ====================================================================
+
+    /**
+     * 机器利用率信息（用于负载均衡计算）。
+     */
+    private static class MachineUtilInfo {
+        long machineId;
+        double minStart;
+        double maxEnd;
+        double span;
+        int opCount;
+        double totalWork;
+    }
+
+    /**
+     * Destroy 算子：破坏解的一部分。
+     */
+    private static class DestroyOperator {
+        final String name;
+        final DestroyFunction function;
+
+        DestroyOperator(String name, DestroyFunction function) {
+            this.name = name;
+            this.function = function;
+        }
+
+        Chromosome apply(Chromosome c, double ratioMin, double ratioMax) {
+            try {
+                return function.apply(c, ratioMin, ratioMax);
+            } catch (Exception e) {
+                FileHelper.writeLogFile("ALNS - Destroy算子 " + name + " 异常: " + e.getMessage());
+                return null;
+            }
+        }
+    }
+
+    @FunctionalInterface
+    private interface DestroyFunction {
+        Chromosome apply(Chromosome c, double ratioMin, double ratioMax);
+    }
+
+    /**
+     * Repair 算子：修复被破坏的解。
+     */
+    private static class RepairOperator {
+        final String name;
+        final RepairFunction function;
+
+        RepairOperator(String name, RepairFunction function) {
+            this.name = name;
+            this.function = function;
+        }
+
+        Chromosome apply(Chromosome c, VariableNeighborhoodSearch vns,
+                         GeneticDecoder decoder, List<Machine> machines) {
+            try {
+                return function.apply(c, vns, decoder, machines);
+            } catch (Exception e) {
+                FileHelper.writeLogFile("ALNS - Repair算子 " + name + " 异常: " + e.getMessage());
+                return null;
+            }
+        }
+    }
+
+    @FunctionalInterface
+    private interface RepairFunction {
+        Chromosome apply(Chromosome c, VariableNeighborhoodSearch vns,
+                         GeneticDecoder decoder, List<Machine> machines);
+    }
+}
\ No newline at end of file

src/main/java/com/aps/service/Algorithm/CpSatFjspModel.java

 package com.aps.service.Algorithm;
 
+import com.aps.common.util.FileHelper;
 import com.aps.entity.Algorithm.Chromosome;
 import com.aps.entity.Algorithm.GlobalOperationInfo;
 import com.aps.entity.basic.Entry;
 import com.aps.entity.basic.Machine;
 import com.aps.entity.basic.MachineOption;
+import com.aps.entity.basic.Order;
 import com.google.ortools.Loader;
 import com.google.ortools.sat.CpModel;
 import com.google.ortools.sat.CpSolver;
@@ -16,6 +18,7 @@ import com.google.ortools.sat.LinearExpr;
 import com.google.ortools.sat.Literal;
 
 import java.time.LocalDateTime;
+import java.time.temporal.ChronoUnit;
 import java.util.*;
 import java.util.concurrent.CopyOnWriteArrayList;
 
@@ -48,6 +51,9 @@ public class CpSatFjspModel {
     private final List<Machine> machines;
     private final List<GlobalOperationInfo> globalOpList;
     private final List<Entry> allOperations;
+    private final LocalDateTime baseTime;
+    private final List<Order> orders;
+
     private final int operationCount;
 
     private CpModel model;
@@ -61,11 +67,14 @@ public class CpSatFjspModel {
     public CpSatFjspModel(List<GlobalOperationInfo> globalOpList,
                           List<Entry> allOperations,
                           List<Machine> machines,
+                          List<Order> orders,
                           LocalDateTime baseTime) {
         this.globalOpList = globalOpList;
         this.allOperations = allOperations;
         this.machines = machines;
         this.operationCount = globalOpList.size();
+        this.orders=orders;
+        this.baseTime=baseTime;
     }
 
     private int estimateHorizon() {
@@ -221,28 +230,90 @@ public class CpSatFjspModel {
                     globalOpList.get(i).getOp().getPriority());
         }
 
-        int priorityTerms = 0;
-        for (int i = 0; i < operationCount; i++) {
-            if (globalOpList.get(i).getOp().getPriority() > 0) priorityTerms++;
-        }
+        List<LinearArgument> objTerms = new ArrayList<>();
+        List<Long> objWeights = new ArrayList<>();
 
-        LinearArgument[] objVars = new LinearArgument[1 + priorityTerms];
-        long[] objCoeffs = new long[1 + priorityTerms];
-        objVars[0] = makespanVar;
-        objCoeffs[0] = 100;
+        objTerms.add(makespanVar);
+        objWeights.add(100L);
 
-        int t = 1;
         for (int i = 0; i < operationCount; i++) {
             double priority = globalOpList.get(i).getOp().getPriority();
             if (priority > 0) {
-                objVars[t] = endVars.get(i);
-                objCoeffs[t] = (long)(maxPriority - priority + 1);
-                t++;
+                objTerms.add(endVars.get(i));
+                objWeights.add((long)(maxPriority - priority + 1));
             }
         }
-        model.minimize(LinearExpr.weightedSum(objVars, objCoeffs));
-    }
 
+        // ==================== 新增目标1：Tardiness（延迟时间） ====================
+        // 按订单分组，找到每个订单最后一道工序的end_time
+        Map<Integer, List<Integer>> groupOps = new LinkedHashMap<>();
+        Map<Integer, Long> groupDueDates = new HashMap<>();
+        for (int i = 0; i < operationCount; i++) {
+            GlobalOperationInfo info = globalOpList.get(i);
+            int groupId = info.getGroupId();
+            groupOps.computeIfAbsent(groupId, k -> new ArrayList<>()).add(i);
+            if (!groupDueDates.containsKey(groupId)) {
+                Order order = orders.stream()
+                        .filter(o -> info.getOp().getOrderId().equals(o.getOrderId()))
+                        .findFirst().orElse(null);
+                if (order != null && order.getDueDate() != null) {
+                    long dueMin = ChronoUnit.MINUTES.between(baseTime, order.getDueDate());
+                    groupDueDates.put(groupId, dueMin * 60);
+                }
+            }
+        }
+
+        long totalTardinessWeight = 80L;
+        for (Map.Entry<Integer, List<Integer>> entry : groupOps.entrySet()) {
+            int groupId = entry.getKey();
+            List<Integer> opIndices = entry.getValue();
+
+            // 找到该订单最后完成的工序
+            IntVar lastEnd = model.newIntVar(0, horizonSeconds, "lastEnd_g" + groupId);
+            List<LinearArgument> endCandidates = new ArrayList<>();
+            for (int idx : opIndices) {
+                endCandidates.add(endVars.get(idx));
+            }
+            model.addMaxEquality(lastEnd, endCandidates.toArray(new LinearArgument[0]));
+
+            Long dueDateSec = groupDueDates.get(groupId);
+            if (dueDateSec != null) {
+                IntVar tardiness = model.newIntVar(0, horizonSeconds, "tardiness_g" + groupId);
+                LinearArgument diff = LinearExpr.sum(new LinearArgument[]{
+                        lastEnd, LinearExpr.constant(-dueDateSec)});
+
+                model.addMaxEquality(tardiness, new LinearArgument[]{diff, LinearExpr.constant(0)});
+                objTerms.add(tardiness);
+                objWeights.add(totalTardinessWeight);
+            }
+        }
+
+        // ==================== 新增目标2：Machine Load Balance（负载均衡） ====================
+        // 简化版本：取每台机器上所有工序的最大end_time，加入目标函数
+        // 这样 CP-SAT 会倾向于让各机器的完工时间更均匀
+        Map<Long, List<LinearArgument>> machineEndTimes = new HashMap<>();
+        for (int i = 0; i < operationCount; i++) {
+            GlobalOperationInfo info = globalOpList.get(i);
+            List<MachineOption> options = info.getOp().getMachineOptions();
+            if (options == null || options.isEmpty()) continue;
+            for (MachineOption mo : options) {
+                machineEndTimes.computeIfAbsent(mo.getMachineId(), k -> new ArrayList<>()).add(endVars.get(i));
+            }
+        }
+
+        long loadBalanceWeight = 30L;
+        for (Map.Entry<Long, List<LinearArgument>> entry : machineEndTimes.entrySet()) {
+            IntVar machineMakespan = model.newIntVar(0, horizonSeconds, "mk_m" + entry.getKey());
+            model.addMaxEquality(machineMakespan, entry.getValue().toArray(new LinearArgument[0]));
+            objTerms.add(machineMakespan);
+            objWeights.add(loadBalanceWeight);
+        }
+
+        model.minimize(LinearExpr.weightedSum(
+                objTerms.toArray(new LinearArgument[0]),
+                objWeights.stream().mapToLong(Long::longValue).toArray()));
+    }
+   private boolean enableLogging=true;
     /**
      * 单次求解，返回一个 Chromosome
      */
@@ -258,8 +329,13 @@ public class CpSatFjspModel {
 
         CpSolverStatus status = solver.solve(model);
         if (status == CpSolverStatus.OPTIMAL || status == CpSolverStatus.FEASIBLE) {
+            long makespanSec = (long) solver.value(makespanVar);
+            FileHelper.log("[CpSatFjsp] 单次求解 状态=" + status
+                    + " CP-SAT目标值=" + String.format("%,d", (long) solver.objectiveValue())
+                    + " makespan=" + String.format("%,d秒 (%.1f小时)", makespanSec, makespanSec / 3600.0),enableLogging);
             return extractChromosome(solver);
         }
+        FileHelper.log("[CpSatFjsp] 单次求解 状态=" + status + "（无解）",enableLogging);
         return null;
     }
 
@@ -328,15 +404,22 @@ public class CpSatFjspModel {
 
             CpSolverStatus status = solver.solve(model);
             if (status == CpSolverStatus.OPTIMAL || status == CpSolverStatus.FEASIBLE) {
+                long msSec = (long) solver.value(makespanVar);
+                FileHelper.log("[CpSatFjsp] 第" + (round + 1) + "轮 状态=" + status
+                        + " CP-SAT目标=" + String.format("%,d", (long) solver.objectiveValue())
+                        + " makespan=" + String.format("%,d秒(%.1f小时)", msSec, msSec / 3600.0),enableLogging);
                 Chromosome chromo = extractChromosome(solver);
                 if (chromo != null && !containsDuplicate(results, chromo)) {
                     chromo.setGsOrls(4);
                     chromo.setGenerateType("CP-SAT");
                     results.add(chromo);
                 }
+            } else {
+                FileHelper.log("[CpSatFjsp] 第" + (round + 1) + "轮 状态=" + status + "（无解）",enableLogging);
             }
         }
 
+        FileHelper.log("[CpSatFjsp] 多样性解生成完成，共" + results.size() + "个解",enableLogging);
         return results;
     }
 

src/main/java/com/aps/service/Algorithm/CpSatInitializer.java

@@ -93,7 +93,7 @@ public class CpSatInitializer {
      */
     private List<Chromosome> smallScaleSolve(List<GlobalOperationInfo> globalOpList,
                                              int targetCount, int timeBudgetSec) {
-        CpSatFjspModel model = new CpSatFjspModel(globalOpList, allOperations, machines, baseTime);
+        CpSatFjspModel model = new CpSatFjspModel(globalOpList, allOperations, machines,orders, baseTime);
         return model.generateDiverseSolutions(
                 Math.min(targetCount, 8), timeBudgetSec, true);
     }
@@ -103,7 +103,7 @@ public class CpSatInitializer {
      */
     private List<Chromosome> mediumScaleSolve(List<GlobalOperationInfo> globalOpList,
                                               int targetCount, int timeBudgetSec) {
-        CpSatFjspModel model = new CpSatFjspModel(globalOpList, allOperations, machines, baseTime);
+        CpSatFjspModel model = new CpSatFjspModel(globalOpList, allOperations, machines,orders, baseTime);
         int effectiveTarget = Math.min(targetCount, 5);
         int perSolveTime = Math.max(timeBudgetSec, 15);
         return model.generateDiverseSolutions(effectiveTarget, perSolveTime, true);
@@ -158,7 +158,7 @@ public class CpSatInitializer {
         }
 
         CpSatFjspModel model = new CpSatFjspModel(
-                bottleneckOps, allOperations, bottleneckMachines, baseTime);
+                bottleneckOps, allOperations, bottleneckMachines,orders, baseTime);
         List<Chromosome> cpSatResults = model.generateDiverseSolutions(
                 effectiveTarget, timeBudgetSec, true);
 

src/main/java/com/aps/service/Algorithm/CpSatLnsNeighborhood.java

+package com.aps.service.Algorithm;
+
+import com.aps.common.util.FileHelper;
+import com.aps.common.util.GlobalCacheUtil;
+import com.aps.common.util.ProductionDeepCopyUtil;
+import com.aps.entity.Algorithm.Chromosome;
+import com.aps.entity.Algorithm.GAScheduleResult;
+import com.aps.entity.Algorithm.GlobalOperationInfo;
+import com.aps.entity.Algorithm.IDAndChildID.GroupResult;
+import com.aps.entity.Algorithm.ObjectiveWeights;
+import com.aps.entity.basic.*;
+import com.google.ortools.Loader;
+import com.google.ortools.sat.CpModel;
+import com.google.ortools.sat.CpSolver;
+import com.google.ortools.sat.CpSolverStatus;
+import com.google.ortools.sat.IntVar;
+import com.google.ortools.sat.IntervalVar;
+import com.google.ortools.sat.LinearArgument;
+import com.google.ortools.sat.LinearExpr;
+import com.google.ortools.sat.Literal;
+
+import java.util.*;
+import java.util.concurrent.CopyOnWriteArrayList;
+import java.util.stream.Collectors;
+import java.util.stream.IntStream;
+
+/**
+ * LNS (Large Neighborhood Search) + OR-Tools CP-SAT
+ *
+ * 从 NSGA-II 帕累托前沿取一个染色体：
+ *   1. 随机"释放" 10%~30% 的工序（让 CP-SAT 重排）
+ *   2. 其余工序"冻结"（保持机器和时间不变）
+ *   3. CP-SAT 在 5~15 秒内给出一个更好的调度
+ *   4. decode 后若目标改进则接受，否则回退
+ *
+ * 循环多轮。
+ */
+public class CpSatLnsNeighborhood {
+
+    private static boolean NATIVE_LIBRARY_LOADED = false;
+
+    static {
+        try {
+            Loader.loadNativeLibraries();
+            Class.forName("com.google.ortools.sat.CpModel");
+            NATIVE_LIBRARY_LOADED = true;
+            FileHelper.writeLogFile("[CpSatLns] OR-Tools 已加载");
+        } catch (Throwable t) {
+            FileHelper.writeLogFile("[CpSatLns] OR-Tools 本机库不可用，LNS 被禁用：" + t.getMessage());
+        }
+    }
+
+    private final List<Entry> allOperations;
+    private final List<Machine> machines;
+    private final FitnessCalculator fitnessCalculator;
+
+    private  GeneticOperations geneticOperations;
+
+    private List<Machine> cachedMachines;
+    private List<Order> cachedOrders;
+    private List<GroupResult> cachedEntryRel;
+    private TreeMap<String, Material> cachedMaterials;
+    private List<Entry> cachedAllOperations;
+
+    private final Random random = new Random(20260618L);
+
+    /**
+     * @param allOperations   全工序列表
+     * @param machines        全机器列表
+     * @param fitnessCalculator 适应度计算器（与 HybridAlgorithm 一致）
+     */
+    public CpSatLnsNeighborhood(List<Entry> allOperations, List<Machine> machines,List<Order> orders,
+                                TreeMap<String, Material> materials,List<GroupResult> entryRel,
+                                FitnessCalculator fitnessCalculator) {
+        this.allOperations = allOperations;
+        this.machines = machines;
+        this.fitnessCalculator = fitnessCalculator;
+        geneticOperations=new GeneticOperations();
+
+        // 预缓存解码需要的深拷贝列表，避免重复拷贝
+        cachedAllOperations = ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList<>(allOperations), Entry.class);
+        cachedOrders = ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList<>(orders), Order.class);
+        cachedEntryRel = ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList<>(entryRel), GroupResult.class);
+        cachedMaterials = ProductionDeepCopyUtil.deepCopyTreeMap(materials, String.class, Material.class);
+
+    }
+
+    /**
+     * 对帕累托前沿做 LNS 重优化。
+     * 返回的 list 包含：改进后的前沿 + 原始前沿（让下游的帕累托归并去重）
+     */
+    public List<Chromosome> runLnsOnParetoFront(List<Chromosome> paretoFront, GeneticDecoder sharedDecoder) {
+        if (!NATIVE_LIBRARY_LOADED || paretoFront == null || paretoFront.isEmpty()) return paretoFront;
+
+        int totalOps = allOperations.size();
+        int lnsOuterRounds = Math.min(paretoFront.size(), 8);
+        int lnsInnerRounds = Math.max(3, Math.min(15, 8000 / Math.max(100, totalOps)));
+        int releasedOps = Math.max(50, (int)(totalOps * 0.15));
+        int cpsatTimeSeconds = Math.max(5, Math.min(15, 12000 / Math.max(100, totalOps)));
+
+        if(totalOps<50)
+        {
+            return  null;
+        }
+
+        FileHelper.writeLogFile("[CpSatLns] 开始 LNS 重优化：工序=" + totalOps
+                + "，每轮释放=" + releasedOps + "，CP-SAT=" + cpsatTimeSeconds + "s，"
+                + "前沿解数=" + lnsOuterRounds + "，每解迭代=" + lnsInnerRounds);
+
+        List<Chromosome> newFront = new ArrayList<>();
+        int totalAttempts = 0, totalImproves = 0;
+
+        for (int i = 0; i < lnsOuterRounds && i < paretoFront.size(); i++) {
+            Chromosome current = lightCopy(paretoFront.get(i));
+            Chromosome beat = ProductionDeepCopyUtil.deepCopy(paretoFront.get(i), Chromosome.class);
+
+            // sharedDecoder.serialDecode(current);
+            double currentFitness = current.getFitness();
+
+            double[] curObj = current.getObjectives();
+            FileHelper.log("[CpSatLns] 帕累托前沿第" + (i + 1) + "个解 → 适应度="
+                    + String.format("%.4f", currentFitness)
+                    + (curObj != null ? " makespan=" + String.format("%.1f", curObj[0]) : ""),true);
+
+            for (int round = 0; round < lnsInnerRounds; round++) {
+                totalAttempts++;
+                Chromosome neighbor = optimizeNeighborhood(current, releasedOps, cpsatTimeSeconds);
+                if (neighbor == null) break;
+
+                decode(sharedDecoder,neighbor,machines);
+
+                if (fitnessCalculator.isBetter(neighbor, current)) {
+                    double[] nObj = neighbor.getObjectives();
+                    FileHelper.log("[CpSatLns]   第" + (round + 1) + "轮 ✅接受"
+                            + " 新适应度=" + String.format("%.4f", neighbor.getFitness())
+                            + " 旧适应度=" + String.format("%.4f", currentFitness)
+                            + (nObj != null ? " 新makespan=" + String.format("%.1f", nObj[0]) : ""),true);
+                    current = neighbor;
+                    beat = ProductionDeepCopyUtil.deepCopy(neighbor, Chromosome.class);
+
+                    currentFitness = neighbor.getFitness();
+                    totalImproves++;
+                }
+            }
+            newFront.add(beat);
+        }
+
+        newFront.addAll(paretoFront);
+
+        FileHelper.log("[CpSatLns] 结束：尝试=" + totalAttempts
+                + "，改进=" + totalImproves + "，合并后解数=" + newFront.size(),true);
+        return newFront;
+    }
+
+    // ----------------------------------------------------------------
+    // 单轮邻域重优化
+    // ----------------------------------------------------------------
+    public Chromosome optimizeNeighborhood(Chromosome base, int releaseCount, int timeLimitSec) {
+
+        List<GlobalOperationInfo> globalOpList = base.getGlobalOpList();
+        if (globalOpList == null || globalOpList.isEmpty()) return null;
+
+        CopyOnWriteArrayList<Integer> ms = base.getMachineSelection();
+        if (ms == null) return null;
+
+        int totalCount = globalOpList.size();
+        releaseCount = Math.min(releaseCount, totalCount);
+
+        long horizon = 30L * 24 * 60 * 60; // 30 天（秒）
+
+        // ================== 1. 从 Result 中读取每道工序的机器+时间 ==================
+        Map<Integer, GAScheduleResult> scheduleMap = new HashMap<>();
+        if (base.getResult() != null) {
+            for (GAScheduleResult r : base.getResult()) {
+                int key = (r.getGroupId() * 1000000) + r.getSeq();
+                scheduleMap.put(key, r);
+            }
+        }
+
+        long[] opStartSec = new long[totalCount];
+        long[] opDurationSec = new long[totalCount];
+        long[] opMachineId = new long[totalCount];
+
+        for (int i = 0; i < totalCount; i++) {
+            GlobalOperationInfo info = globalOpList.get(i);
+            Entry op = info.getOp();
+            int key = (info.getGroupId() * 1000000) + info.getSequence();
+            GAScheduleResult res = scheduleMap.get(key);
+
+            if (res != null) {
+                opStartSec[i] = res.getStartTime();
+                opMachineId[i] = res.getMachineId();
+                opDurationSec[i] = Math.max(60L, (long)(res.getProcessingTime() * 60));
+            } else {
+                opDurationSec[i] = Math.max(60L, (long)(op.getMinProcessingTime() * 60));
+                opStartSec[i] = 0;
+                List<MachineOption> options = op.getMachineOptions();
+                opMachineId[i] = (options != null && !options.isEmpty()) ? options.get(0).getMachineId() : -1;
+            }
+        }
+
+        // ================== 2. 随机选"释放"的工序 ==================
+        BitSet released = new BitSet(totalCount);
+        if (releaseCount >= totalCount) {
+            released.set(0, totalCount);
+        } else {
+            List<Integer> indices = IntStream.range(0, totalCount)
+                    .boxed().collect(Collectors.toList());
+            Collections.shuffle(indices, random);
+            for (int k = 0; k < releaseCount; k++) released.set(indices.get(k));
+        }
+
+        // ================== 3. 构建 CP-SAT 模型 ==================
+        try {
+            CpModel model = new CpModel();
+
+            Map<Long, Integer> machineIdx = new HashMap<>();
+            for (int m = 0; m < machines.size(); m++) machineIdx.put(machines.get(m).getId(), m);
+
+            List<List<IntervalVar>> machineIntervals = new ArrayList<>();
+            for (int m = 0; m < machines.size(); m++) machineIntervals.add(new ArrayList<>());
+
+            IntVar[] startVars = new IntVar[totalCount];
+            // presence[i][j] = 第 i 个释放工序选第 j 台机器的布尔
+            List<Literal[]> presenceMatrix = new ArrayList<>();
+
+            // 处理释放工序
+            for (int i = 0; i < totalCount; i++) {
+                List<MachineOption> options = globalOpList.get(i).getOp().getMachineOptions();
+                presenceMatrix.add(null);
+                if (!released.get(i)) continue;
+                if (options == null || options.isEmpty()) continue;
+
+                long baseStart = Math.max(0, opStartSec[i]);
+                long lower = Math.max(0, baseStart - 3600);
+                long upper = Math.min(horizon, baseStart + 10800);
+                if (upper <= lower) upper = lower + 3600;
+
+                startVars[i] = model.newIntVar(lower, upper, "start_" + i);
+
+                List<Literal> presList = new ArrayList<>();
+                for (int j = 0; j < options.size(); j++) {
+                    MachineOption mo = options.get(j);
+                    long procSec = Math.max(60L, (long)(mo.getProcessingTime() * 60));
+                    Literal pres = model.newBoolVar("pres_" + i + "_" + j);
+                    presList.add(pres);
+
+                    IntVar endVar = model.newIntVar(lower + procSec, upper + procSec,
+                            "end_" + i + "_" + j);
+                    model.addEquality(LinearExpr.sum(new LinearArgument[]{
+                            startVars[i], LinearExpr.constant(procSec)}), endVar);
+
+                    IntervalVar iv = model.newOptionalIntervalVar(
+                            startVars[i], model.newConstant(procSec), endVar, pres,
+                            "iv_" + i + "_m" + mo.getMachineId());
+                    int mIdx = machineIdx.getOrDefault(mo.getMachineId(), -1);
+                    if (mIdx >= 0) machineIntervals.get(mIdx).add(iv);
+                }
+
+                Literal[] arr = presList.toArray(new Literal[0]);
+                if (arr.length > 0) {
+                    model.addExactlyOne(arr);
+                    presenceMatrix.set(i, arr);
+                }
+            }
+
+            // 冻结工序：固定 interval
+            for (int i = 0; i < totalCount; i++) {
+                if (released.get(i)) continue;
+                long start = Math.max(0, opStartSec[i]);
+                long dur = Math.max(60L, opDurationSec[i]);
+                long mId = opMachineId[i];
+                int mIdx = machineIdx.getOrDefault(mId, -1);
+                if (mIdx >= 0) {
+                    IntervalVar iv = model.newFixedInterval(start, dur, "frozen_" + i);
+                    machineIntervals.get(mIdx).add(iv);
+                }
+            }
+
+            // ================== 4. 同订单工序顺序 ==================
+            Map<Integer, List<Integer>> orderOpsMap = new LinkedHashMap<>();
+            for (int i = 0; i < totalCount; i++) {
+                orderOpsMap.computeIfAbsent(
+                        globalOpList.get(i).getGroupId(), k -> new ArrayList<>()).add(i);
+            }
+            for (List<Integer> ops : orderOpsMap.values()) {
+                ops.sort(Comparator.comparingInt(idx -> globalOpList.get(idx).getSequence()));
+                for (int k = 0; k < ops.size() - 1; k++) {
+                    int prev = ops.get(k);
+                    int next = ops.get(k + 1);
+                    long prevDuration = Math.max(60L, opDurationSec[prev]);
+
+                    LinearArgument prevEnd;
+                    LinearArgument nextStart;
+                    if (released.get(prev)) {
+                        prevEnd = LinearExpr.sum(new LinearArgument[]{
+                                startVars[prev], LinearExpr.constant(prevDuration)});
+                    } else {
+                        prevEnd = LinearExpr.constant(opStartSec[prev] + prevDuration);
+                    }
+                    if (released.get(next)) {
+                        nextStart = startVars[next];
+                    } else {
+                        nextStart = LinearExpr.constant(opStartSec[next]);
+                    }
+                    model.addLessOrEqual(prevEnd, nextStart);
+                }
+            }
+
+            // ================== 5. 每台机器 no_overlap ==================
+            for (List<IntervalVar> list : machineIntervals) {
+                if (list.size() > 1) {
+                    model.addNoOverlap(list.toArray(new IntervalVar[0]));
+                }
+            }
+
+            // ================== 6. 目标：makespan + 优先级加权 + 延迟 + 负载均衡 ==================
+            IntVar makespan = model.newIntVar(0, horizon, "makespan");
+            List<LinearArgument> objTerms = new ArrayList<>();
+            List<Long> objWeights = new ArrayList<>();
+
+            objTerms.add(makespan);
+            objWeights.add(100L);
+
+            double maxPriority = 1.0;
+            for (int i = 0; i < totalCount; i++) {
+                maxPriority = Math.max(maxPriority, globalOpList.get(i).getOp().getPriority());
+            }
+
+            for (int i = 0; i < totalCount; i++) {
+                long dur = Math.max(60L, opDurationSec[i]);
+                LinearArgument endVar;
+                if (released.get(i)) {
+                    endVar = LinearExpr.sum(new LinearArgument[]{
+                            startVars[i], LinearExpr.constant(dur)});
+                } else {
+                    endVar = LinearExpr.constant(opStartSec[i] + dur);
+                }
+                model.addLessOrEqual(endVar, makespan);
+
+                if (released.get(i)) {
+                    double priority = globalOpList.get(i).getOp().getPriority();
+                    long weight = (long)(maxPriority - priority + 1);
+                    objTerms.add(endVar);
+                    objWeights.add(weight);
+                }
+            }
+
+            // 延迟时间（只对释放的工序所属订单）
+            long tardinessWeight = 80L;
+            Map<Integer, List<LinearArgument>> groupEnds = new HashMap<>();
+            for (int i = 0; i < totalCount; i++) {
+                if (!released.get(i)) continue;
+                GlobalOperationInfo info = globalOpList.get(i);
+                groupEnds.computeIfAbsent(info.getGroupId(), k -> new ArrayList<>())
+                        .add(LinearExpr.sum(new LinearArgument[]{
+                                startVars[i], LinearExpr.constant(Math.max(60L, opDurationSec[i]))}));
+            }
+            for (Map.Entry<Integer, List<LinearArgument>> entry : groupEnds.entrySet()) {
+                IntVar lastEnd = model.newIntVar(0, horizon, "lastEnd_g" + entry.getKey());
+                model.addMaxEquality(lastEnd, entry.getValue().toArray(new LinearArgument[0]));
+                // 简化：延迟作为惩罚项加入（实际可加 due_date 判断）
+                objTerms.add(lastEnd);
+                objWeights.add(tardinessWeight);
+            }
+
+            // 机器负载均衡
+            long loadWeight = 30L;
+            Map<Long, List<LinearArgument>> machineEnds = new HashMap<>();
+            for (int i = 0; i < totalCount; i++) {
+                if (!released.get(i)) continue;
+                GlobalOperationInfo info = globalOpList.get(i);
+                List<com.aps.entity.basic.MachineOption> options = info.getOp().getMachineOptions();
+                if (options == null || options.isEmpty()) continue;
+                LinearArgument endVar = LinearExpr.sum(new LinearArgument[]{
+                        startVars[i], LinearExpr.constant(Math.max(60L, opDurationSec[i]))});
+                for (com.aps.entity.basic.MachineOption mo : options) {
+                    machineEnds.computeIfAbsent(mo.getMachineId(), k -> new ArrayList<>()).add(endVar);
+                }
+            }
+            for (Map.Entry<Long, List<LinearArgument>> entry : machineEnds.entrySet()) {
+                IntVar machineMk = model.newIntVar(0, horizon, "mk_m" + entry.getKey());
+                model.addMaxEquality(machineMk, entry.getValue().toArray(new LinearArgument[0]));
+                objTerms.add(machineMk);
+                objWeights.add(loadWeight);
+            }
+
+            model.minimize(LinearExpr.weightedSum(
+                    objTerms.toArray(new LinearArgument[0]),
+                    objWeights.stream().mapToLong(Long::longValue).toArray()));
+
+            // ================== 7. 求解 ==================
+            CpSolver solver = new CpSolver();
+            solver.getParameters().setMaxTimeInSeconds(timeLimitSec);
+            solver.getParameters().setNumSearchWorkers(4);
+            solver.getParameters().setLogSearchProgress(false);
+
+            CpSolverStatus status = solver.solve(model);
+            if (status != CpSolverStatus.OPTIMAL && status != CpSolverStatus.FEASIBLE) {
+                FileHelper.writeLogFile("[CpSatLns] 邻域求解 状态=" + status + "（无解），释放=" + releaseCount + "道工序");
+                return null;
+            }
+            long cpObjective = (long) solver.objectiveValue();
+            long cpMakespan = (long) solver.value(makespan);
+            FileHelper.writeLogFile("[CpSatLns] 邻域求解 状态=" + status
+                    + " 释放工序=" + releaseCount
+                    + " CP-SAT目标=" + String.format("%,d", cpObjective)
+                    + " makespan=" + String.format("%,d秒(%.1f小时)", cpMakespan, cpMakespan / 3600.0));
+
+            // ================== 8. 写回新染色体的 machineSelection ==================
+            Chromosome neighbor = base.deepCopy();
+            CopyOnWriteArrayList<Integer> newMs = new CopyOnWriteArrayList<>(
+                    neighbor.getMachineSelection() != null
+                            ? neighbor.getMachineSelection()
+                            : new ArrayList<>()
+            );
+            while (newMs.size() < totalCount) newMs.add(1);
+
+            for (int i = 0; i < totalCount; i++) {
+                if (!released.get(i)) continue;
+                Literal[] arr = presenceMatrix.get(i);
+                if (arr == null) continue;
+                for (int j = 0; j < arr.length; j++) {
+                    if (solver.value(arr[j]) == 1L) {
+                        newMs.set(i, j + 1);  // 1-based
+                        break;
+                    }
+                }
+            }
+
+            neighbor.setMachineSelection(newMs);
+            neighbor.setGenerateType("LNS-CPSAT");
+            neighbor.setGsOrls(5);
+            return neighbor;
+
+        } catch (Exception e) {
+            FileHelper.writeLogFile("[CpSatLns] 邻域重优化异常：" + e.getMessage());
+            return null;
+        }
+    }
+    /**
+     * 轻量拷贝：只复制 generateNeighbor/DelOrder 需要的字段，避免全量 JSON 深拷贝导致 OOM。
+     * result/machines/operatRel 等重型数据共享引用（generateNeighbor 只读，不修改）。
+     */
+    private Chromosome lightCopy(Chromosome source) {
+        Chromosome copy = new Chromosome();
+        copy.setOperationSequencing(new CopyOnWriteArrayList<>(source.getOperationSequencing()));
+        copy.setMachineSelection(new CopyOnWriteArrayList<>(source.getMachineSelection()));
+        copy.setGlobalOpList(new CopyOnWriteArrayList<>(source.getGlobalOpList()));
+        copy.setOrders(new CopyOnWriteArrayList<>(source.getOrders()));
+        copy.setAllOperations(new CopyOnWriteArrayList<>(source.getAllOperations()));
+        copy.setResult(source.getResult());
+        copy.setMachines(source.getMachines());
+        copy.setOperatRel(new CopyOnWriteArrayList<>(source.getOperatRel()));
+        copy.setScenarioID(source.getScenarioID());
+        copy.setBaseTime(source.getBaseTime());
+        copy.setGenerateType(source.getGenerateType());
+        copy.setFitnessLevel(source.getFitnessLevel());
+        copy.setFitness(source.getFitness());
+        copy.setObjectives(source.getObjectives());
+        geneticOperations.DelOrder(copy);
+        return copy;
+    }
+    /**
+     * 解码染色体
+     */
+    private void decode(GeneticDecoder decoder, Chromosome chromosome , List<Machine> machines) {
+        // MS 校验：解码前检查 machineSelection 与 machineOptions 是否匹配
+        List<GlobalOperationInfo> gops = chromosome.getGlobalOpList();
+        List<Integer> msCheck = chromosome.getMachineSelection();
+        if (gops != null && msCheck != null) {
+            int msErrors = 0;
+            StringBuilder sb = new StringBuilder();
+            for (int i = 0; i < Math.min(gops.size(), msCheck.size()); i++) {
+                Entry op = gops.get(i).getOp();
+                int msVal = msCheck.get(i);
+                if (op != null && op.getMachineOptions() != null
+                        && (msVal < 1 || msVal > op.getMachineOptions().size())) {
+                    msErrors++;
+                    if (msErrors <= 3) {
+                        sb.append(String.format(" [idx=%d 订单%d工序%d ms=%d range=1-%d]",
+                                i, op.getGroupId(), op.getSequence(), msVal, op.getMachineOptions().size()));
+                    }
+                }
+            }
+            if (msErrors > 0) {
+                FileHelper.log(String.format("decode-MS校验失败: 共%d处越界 %s", msErrors, sb.toString()));
+            }
+        }
+
+        chromosome.setResult(new CopyOnWriteArrayList<>());
+
+
+
+        // 使用缓存的列表，避免重复深拷贝
+        chromosome.setMachines(ProductionDeepCopyUtil.deepCopyList(machines, Machine.class));
+        chromosome.setOrders(ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList<>(cachedOrders), Order.class));
+        chromosome.setOperatRel(ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList<>(cachedEntryRel), GroupResult.class));
+        chromosome.setMaterials(ProductionDeepCopyUtil.deepCopyTreeMap(cachedMaterials, String.class, Material.class));
+        chromosome.setAllOperations(ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList<>(allOperations), Entry.class));
+
+        // 加载锁定工单到ResultOld
+        List<GAScheduleResult> lockedOrders = GlobalCacheUtil.get("locked_orders_" + chromosome.getScenarioID());
+        if (lockedOrders != null && !lockedOrders.isEmpty()) {
+            chromosome.setResultOld(ProductionDeepCopyUtil.deepCopyList(lockedOrders, GAScheduleResult.class));
+        } else {
+            chromosome.setResultOld(new CopyOnWriteArrayList<>());
+        }
+        decoder.decodeChromosomeWithCache(chromosome,false);
+    }
+
+}

src/main/java/com/aps/service/Algorithm/GeneticDecoder.java

@@ -695,11 +695,6 @@ public class GeneticDecoder {
             int scheduledCount = orderProcessCounter.get(groupId);
 
 
-            if(groupId==7)
-            {
-                int k=0;
-            }
-
             List<Entry> orderOps=new ArrayList<>();
 
             boolean orderIsJit=orderDueDate.get(groupId)>0;
@@ -743,7 +738,7 @@ public class GeneticDecoder {
                 } else {
                     orderAnchor = bom.computeSemiFinishedAnchor(this, groupId, entrysBygroupId,
                             opMachineKeyMap, chromosome,
-                            scheduleIndexById, machineTasksCache, machineIdMap, entryIndexById,_globalParam.isIsCheckMp());
+                            scheduleIndexById, machineTasksCache, machineIdMap, entryIndexById,_globalParam.isIsCheckMp(),null);
                     if (orderAnchor < 0) {
                         orderIsJit = false;
                         orderSchedulingInfo.put(groupId,
@@ -3524,6 +3519,7 @@ if(geneDetails!=null&&geneDetails.size()>0)
 
     private void calculateScheduleResult(Chromosome chromosome) {
         double[] Objectives = new double[_globalParam.getObjectiveWeights().size()];
+        double[] weights = new double[_globalParam.getObjectiveWeights().size()];
 
         int i = 0;
         for (ObjectiveConfig config : _globalParam.getObjectiveConfigs()) {
@@ -3535,7 +3531,9 @@ if(geneDetails!=null&&geneDetails.size()>0)
                             .max()
                             .orElse(0);
                     Objectives[i] = makespan;
+                    weights[i] = config.getWeight();
                     chromosome.setMakespan(makespan);
+
                 }
                 if (GlobalParam.OBJECTIVE_TARDINESS.equals(config.getName())) {
                     // 2. 交付期满足情况（最小化延迟）
@@ -3604,6 +3602,16 @@ if(geneDetails!=null&&geneDetails.size()>0)
         }
 
         chromosome.setObjectives(Objectives);
+
+        // 计算各 KPI 的理论下界（用于计算 Gap = (current - lowerBound) / lowerBound）
+        try {
+            KpiLowerBoundCalculator.computeAndSetLowerBounds(chromosome, _globalParam);
+        } catch (Exception e) {
+            // 下界计算异常不应影响主流程
+            com.aps.common.util.FileHelper.writeLogFile(
+                    "KPI 下界计算异常: " + e.getClass().getSimpleName() + " - " + e.getMessage());
+        }
+
         FitnessCalculator fitnessCalculator = new FitnessCalculator();
 
         chromosome.setFitnessLevel(fitnessCalculator.calculateFitness(chromosome, _globalParam));

src/main/java/com/aps/service/Algorithm/HillClimbing.java

@@ -76,7 +76,7 @@ public class HillClimbing {
         Chromosome current = ProductionDeepCopyUtil.deepCopy(chromosome, Chromosome.class);
 
         Chromosome best = ProductionDeepCopyUtil.deepCopy(chromosome, Chromosome.class);
-        decoder.DelOrder(current);
+
 
         // 构建位置索引映射：groupId_sequence -> position
         Map<String, Integer> positionIndex = buildPositionIndex(current);
@@ -119,7 +119,7 @@ public class HillClimbing {
                             positionIndex = buildPositionIndex(current);
                             entryIndex = buildEntryIndex(current, entrys);
                             MachinePositionIndex = buildEntryMachinePositionIndex(current);
-                            decoder.DelOrder(current);
+                           // decoder.DelOrder(current);
                             break;
                         }
                     }

src/main/java/com/aps/service/Algorithm/HybridAlgorithm.java

@@ -49,6 +49,8 @@ public class HybridAlgorithm {
     private String sceneId;
 
     private VariableNeighborhoodSearch _vns;
+
+    private AdaptiveLargeNeighborhoodSearch _ALNS;
     // 初始化算法实例
     private HillClimbing _hillClimbing;
     private SimulatedAnnealing _simulatedAnnealing;
@@ -116,6 +118,7 @@ public class HybridAlgorithm {
         // 初始化变邻域搜索
         _vns = new VariableNeighborhoodSearch( allOperations,orders,materials,_entryRel,  _fitnessCalculator );
         _vns.initMachineSelectFrequency();
+        _ALNS = new AdaptiveLargeNeighborhoodSearch( allOperations,orders,materials,_entryRel,  _fitnessCalculator );
         _hillClimbing = new HillClimbing(allOperations,orders,materials,_entryRel,  _fitnessCalculator);
         _simulatedAnnealing = new SimulatedAnnealing( allOperations,orders,materials,_entryRel,  _fitnessCalculator);
         _tabuSearch = new TabuSearch(allOperations,orders,materials,_entryRel,  _fitnessCalculator);
@@ -148,6 +151,26 @@ public class HybridAlgorithm {
         if (population == null || population.isEmpty()) {
             throw new RuntimeException("初始种群为空，请检查 populationSize、种群初始化和解码结果");
         }
+        if (_GlobalParam.isOptimizer()) {
+            FileHelper.writeLogFile("LNS-CPSAT 邻域重优化-----------开始-------种群=" + population.size());
+            try {
+
+                CpSatLnsNeighborhood lns = new CpSatLnsNeighborhood(
+                        allOperations, machines,orders,materials,_entryRel,_fitnessCalculator);
+                List<Chromosome> lnsResult = lns.runLnsOnParetoFront(population, sharedDecoder);
+                if (lnsResult != null && !lnsResult.isEmpty()) {
+                  //  Chromosomedecode(sharedDecoder, param, allOperations, globalOpList, lnsResult);
+                    population = chromosomeDistinctByObjectives(lnsResult);
+                    if (population == null || population.isEmpty()) {
+                        population = lnsResult;
+                    }
+                }
+            } catch (Throwable t) {
+                FileHelper.writeLogFile("LNS-CPSAT 失败，跳过：" + t.getMessage());
+            }
+            FileHelper.writeLogFile("LNS-CPSAT 邻域重优化-----------结束-------种群="
+                    + (population == null ? 0 : population.size()));
+        }
       //  if(1==1)
       //  return getBestChromosome(population.get(0), param.getBaseTime(), starttime);
         // 步骤2：对初始种群进行爬山法局部优化
@@ -178,9 +201,9 @@ public class HybridAlgorithm {
 
             return getBestChromosome(saHcOptimized, param.getBaseTime(), starttime);
         }
-        if(opcount>=800 ) {
+        if(opcount>800&&opcount<2000 ) {
             Chromosome best=population.get(0);
-
+            best = _ALNS.search(best,_tabuSearch,_vns, sharedDecoder, machines);
             best = _simulatedAnnealing.search(best, _tabuSearch, _vns, sharedDecoder, machines);
 
             best = _vns.search(best,_tabuSearch, sharedDecoder, machines);
@@ -189,6 +212,29 @@ public class HybridAlgorithm {
 
             return getBestChromosome(best, param.getBaseTime(), starttime);
 
+        }else {
+            Chromosome best=population.get(0);
+            int topN = Math.min(3, population.size());
+            for (int iter = 0; iter < topN; iter++) {
+                FileHelper.writeLogFile("迭代进化------"+iter+"-----开始-------");
+
+                 System.gc();
+
+                Chromosome chromosome=population.get(iter);
+                chromosome = _ALNS.search(chromosome,_tabuSearch,_vns, sharedDecoder, machines);
+                chromosome = _simulatedAnnealing.search(chromosome, _tabuSearch, _vns, sharedDecoder, machines);
+
+                chromosome = _vns.search(chromosome,_tabuSearch, sharedDecoder, machines);
+
+                if(_fitnessCalculator.isBetter(chromosome,best)) {
+                    FileHelper.writeLogFile("迭代进化------发现更优解-----------");
+                    writeKpi(chromosome);
+                    best= chromosome;
+                }
+                FileHelper.writeLogFile("迭代进化------"+iter+"-----结束-------");
+                // 周期性 GC 释放内存
+                System.gc();
+            }
         }
 
 
@@ -514,7 +560,47 @@ public class HybridAlgorithm {
 //        }
     }
 
+    private  void  writeKpi(Chromosome chromosome) {
+        String fitness = "";
+        double[] fitness1 = chromosome.getFitnessLevel();
+        if (fitness1 != null) {
+            for (int i = 0; i < fitness1.length; i++) {
+                fitness += fitness1[i] + ",";
+            }
+        } else {
+            fitness = "null (未计算)";
+        }
+
+        log(String.format("变邻域搜索 - kpi：%s", fitness),true);
+        if(chromosome.getMakespan()!=0) {
+            log(String.format("变邻域搜索 - kpi-Makespan: %f",  chromosome.getMakespan()));
+        }
+        if(chromosome.getDelayTime()!=0) {
+            log(String.format("变邻域搜索 - kpi-DelayTime: %f",  chromosome.getDelayTime()));
+        }
+        if(chromosome.getTotalChangeoverTime()!=0) {
 
+            log(String.format("变邻域搜索 - kpi-ChangeoverTime: %f", chromosome.getTotalChangeoverTime()));
+        }
+        if(chromosome.getMachineLoadStd()!=0) {
+            log(String.format("变邻域搜索 - kpi-MachineLoad: %f", chromosome.getMachineLoadStd()));
+        }
+        if(chromosome.getTotalFlowTime()!=0) {
+            log(String.format("变邻域搜索 - kpi-FlowTime: %f",chromosome.getTotalFlowTime()));
+        }
+
+        // ==================== 打印各 KPI 的 Gap ====================
+        log(KpiLowerBoundCalculator.generateGapReport(chromosome));
+
+    }
+    private void log(String message) {
+        log( message,  true);
+    }
+    private void log(String message, boolean enableLogging) {
+        if (enableLogging ) {
+            FileHelper.writeLogFile(message);
+        }
+    }
 
 }
 

src/main/java/com/aps/service/Algorithm/Initialization.java

@@ -1137,9 +1137,15 @@ public class Initialization {
             List<Chromosome> heuristicPopulation =
                     generateHeuristicInitialPopulation(subParam,remaining);
             for (Chromosome chromo : heuristicPopulation) {
-                chromo.setOrders(new CopyOnWriteArrayList<>(orders));
+
+if(chromo!=null)
+{
+    chromo.setOrders(new CopyOnWriteArrayList<>(orders));
+    population.add(chromo);
+}
+
             }
-            population.addAll(heuristicPopulation);
+
         }
 
         long cpSatCount = population.stream()

src/main/java/com/aps/service/Algorithm/KpiLowerBoundCalculator.java

+package com.aps.service.Algorithm;
+
+import com.aps.entity.Algorithm.Chromosome;
+import com.aps.entity.Algorithm.GAScheduleResult;
+import com.aps.entity.Algorithm.IDAndChildID.GroupResult;
+import com.aps.entity.Algorithm.IDAndChildID.NodeInfo;
+import com.aps.entity.basic.Entry;
+import com.aps.entity.basic.GlobalParam;
+import com.aps.entity.basic.ObjectiveConfig;
+
+import java.util.*;
+import java.util.stream.Collectors;
+
+/**
+ * 计算各 KPI 的理论下界（Lower Bound）。
+ *
+ * <p>Gap = (current - lowerBound) / lowerBound（lowerBound > 0 时）
+ * <p>当 lowerBound = 0 时（如 Tardiness），Gap = current（表示偏离 0 的绝对量）
+ *
+ * <p>各维度的理论下界：
+ * <ul>
+ *   <li>Makespan（最大完工时间）：DAG 关键路径长度（拓扑序 + 动态规划）</li>
+ *   <li>FlowTime（总流程时间）：所有工序加工时间之和（∑ processingTime）</li>
+ *   <li>SetupTime（总换型时间）：若换型时间只由产品类型决定则为 0；若必须换型则为换型时间之和</li>
+ *   <li>MachineLoad（机器负载均衡标准差）：完美均衡下为 0</li>
+ *   <li>Tardiness（总延迟时间）：0（可全部按时交付）</li>
+ * </ul>
+ *
+ * 作者：佟礼
+ */
+public class KpiLowerBoundCalculator {
+
+    /**
+     * 对已解码的 chromosome 计算各维度的理论下界，并存入 chromosome.LowerBoundObjectives。
+     *
+     * @param chromosome 已完成 decode 的染色体
+     * @param globalParam 全局参数（用于确定 objectives 数组的顺序与配置）
+     */
+    public static void computeAndSetLowerBounds(Chromosome chromosome, GlobalParam globalParam) {
+        double[] objectives = chromosome.getObjectives();
+        if (objectives == null || objectives.length == 0) return;
+
+        double[] lowerBounds = new double[objectives.length];
+
+        // 按 GlobalParam.objectiveConfigs 的顺序（与 objectives 数组一一对应）
+        // 注意：globalParam.objectiveConfigs 已按 level 排序
+        List<ObjectiveConfig> configs = globalParam.getObjectiveConfigs();
+
+        for (int i = 0; i < objectives.length && i < configs.size(); i++) {
+            ObjectiveConfig config = configs.get(i);
+            if (!config.isEnabled()) {
+                lowerBounds[i] = 0.0;
+                continue;
+            }
+            String name = config.getName();
+            if (GlobalParam.OBJECTIVE_MAKESPAN.equals(name)) {
+                lowerBounds[i] = computeMakespanLowerBound(chromosome);
+            } else if (GlobalParam.OBJECTIVE_TARDINESS.equals(name)) {
+                // 延迟理论上界为 0（全部按时）
+                lowerBounds[i] = 0.0;
+            } else if (GlobalParam.OBJECTIVE_SETUP_TIME.equals(name)) {
+                // 换型时间下界：若换型只由产品类型决定则为 0
+                lowerBounds[i] = 0.0;
+            } else if (GlobalParam.OBJECTIVE_FLOW_TIME.equals(name)) {
+                lowerBounds[i] = computeFlowTimeLowerBound(chromosome);
+            } else if (GlobalParam.OBJECTIVE_MACHINE_LOAD.equals(name)) {
+                // 负载均衡标准差理论上界为 0
+                lowerBounds[i] = 0.0;
+            } else {
+                lowerBounds[i] = 0.0;
+            }
+        }
+
+        chromosome.setLowerBoundObjectives(lowerBounds);
+    }
+
+    /**
+     * 计算 Makespan（最大完工时间）的理论下界 = DAG 关键路径长度。
+     *
+     * <p>思路：对 job shop 问题，完工时间下界 = 所有工序的最长无冲突调度路径长度，
+     * 即在不考虑机器冲突的条件下，从根节点到终点的最长加权和路径（边权重 = 加工时间）。
+     * 通过拓扑排序 + 动态规划实现（时间复杂度 O(V+E)）。
+     */
+    private static double computeMakespanLowerBound(Chromosome chromosome) {
+        List<GAScheduleResult> result = chromosome.getResult();
+        List<GroupResult> operatRel = chromosome.getOperatRel();
+        List<Entry> allOperations = chromosome.getAllOperations();
+
+        if (result == null || result.isEmpty()) return 0.0;
+
+        // ---- 1. 构建 nodeId -> processingTime 映射 ----
+        Map<Integer, Double> entryProcessingTime = new HashMap<>();
+        if (allOperations != null) {
+            for (Entry e : allOperations) {
+                entryProcessingTime.put(e.getId(), (double) e.getMinProcessingTime());
+            }
+        }
+        // 如果 allOperations 缺失，从 result 推算（使用 flowTime 作为加工时间近似）
+        if (entryProcessingTime.isEmpty()) {
+            for (GAScheduleResult r : result) {
+                entryProcessingTime.put(r.getOperationId(), Math.max(1.0, r.getProcessingTime()));
+            }
+        }
+
+        // ---- 2. 构建 DAG：nodeId -> [childNodeIds] ----
+        Map<Integer, List<Integer>> dag = new HashMap<>();
+        Set<Integer> allNodeIds = new HashSet<>();
+        // 初始化所有节点
+        for (GAScheduleResult r : result) {
+            int eid = r.getOperationId();
+            dag.putIfAbsent(eid, new ArrayList<>());
+            allNodeIds.add(eid);
+        }
+        // 添加边：parent -> child（基于 GroupResult.newParentIds）
+        if (operatRel != null) {
+            for (GroupResult gr : operatRel) {
+                List<NodeInfo> nodes = gr.getNodeInfoList();
+                if (nodes == null) continue;
+                for (NodeInfo node : nodes) {
+                    Integer nodeId = node.getGlobalSerial();
+                    dag.putIfAbsent(nodeId, new ArrayList<>());
+                    allNodeIds.add(nodeId);
+                    List<Integer> childIds = node.getNewChildIds();
+                    if (childIds != null) {
+
+
+
+                        for (Integer childId : childIds) {
+
+                                dag.get(nodeId).add(childId);
+                                allNodeIds.add(childId);
+
+                        }
+                    }
+                }
+            }
+        }
+
+        // ---- 3. 计算入度 ----
+        Map<Integer, Integer> inDegree = new HashMap<>();
+        for (Integer nodeId : allNodeIds) {
+            inDegree.put(nodeId, 0);
+        }
+        for (Integer parent : dag.keySet()) {
+            for (Integer child : dag.get(parent)) {
+                inDegree.merge(child, 1, Integer::sum);
+            }
+        }
+
+        // ---- 4. 拓扑排序 + 动态规划找最长路径 ----
+        // earliest[nodeId] = 从任意根节点到 nodeId 的最长路径
+        Map<Integer, Double> earliest = new HashMap<>();
+        Queue<Integer> queue = new LinkedList<>();
+
+        // 初始化：入度为 0 的节点 earliest = processingTime
+        for (Integer nodeId : allNodeIds) {
+            if (inDegree.get(nodeId) == 0) {
+                double pt = entryProcessingTime.getOrDefault(nodeId, 0.0);
+                earliest.put(nodeId, pt);
+                queue.offer(nodeId);
+            }
+        }
+
+        while (!queue.isEmpty()) {
+            Integer parent = queue.poll();
+            double parentEarliest = earliest.getOrDefault(parent, 0.0);
+            double parentPt = entryProcessingTime.getOrDefault(parent, 0.0);
+            double parentFinish = parentEarliest; // finish = start + pt，这里 start = parentEarliest - pt
+
+            for (Integer child : dag.getOrDefault(parent, Collections.emptyList())) {
+                double childPt = entryProcessingTime.getOrDefault(child, 0.0);
+                double childEarliestCandidate = parentEarliest + childPt;
+                earliest.put(child, Math.max(earliest.getOrDefault(child, 0.0), childEarliestCandidate));
+                inDegree.merge(child, -1, Integer::sum);
+                if (inDegree.get(child) == 0) {
+                    queue.offer(child);
+                }
+            }
+        }
+
+        // ---- 5. 关键路径 = 所有节点 earliest 的最大值 ----
+        double criticalPath = 0.0;
+        for (double val : earliest.values()) {
+            if (val > criticalPath) criticalPath = val;
+        }
+        return criticalPath;
+    }
+
+    /**
+     * 计算 FlowTime（总流程时间）的理论下界 = 所有工序的加工时间之和。
+     */
+    private static double computeFlowTimeLowerBound(Chromosome chromosome) {
+        List<GAScheduleResult> result = chromosome.getResult();
+        if (result == null || result.isEmpty()) return 0.0;
+        return result.stream()
+                .mapToDouble(r -> Math.max(1.0, r.getProcessingTime()))
+                .sum();
+    }
+
+    /**
+     * 计算总 Tardiness 的 Gap。
+     * lowerBound = 0，所以 Gap = current（即延迟小时数的绝对值）。
+     * 结果为 (current - 0) / 1 = current，与延迟同量纲。
+     */
+    public static double computeTardinessGap(double tardiness) {
+        if (tardiness <= 0) return 0.0;
+        return tardiness; // lowerBound=0，用 1 做分母
+    }
+
+    /**
+     * 计算 Makespan Gap。
+     */
+    public static double computeMakespanGap(double makespan, double lowerBound) {
+        if (lowerBound <= 0) return 0.0;
+        return (makespan - lowerBound) / lowerBound;
+    }
+
+    /**
+     * 计算 FlowTime Gap。
+     */
+    public static double computeFlowTimeGap(double flowTime, double lowerBound) {
+        if (lowerBound <= 0) return 0.0;
+        return (flowTime - lowerBound) / lowerBound;
+    }
+
+    /**
+     * 计算 MachineLoad Gap（标准差的下界为 0）。
+     */
+    public static double computeMachineLoadGap(double machineLoadStd) {
+        if (machineLoadStd <= 0) return 0.0;
+        return machineLoadStd; // lowerBound=0
+    }
+
+    /**
+     * 格式化 Gap 为百分比字符串。
+     */
+    public static String formatGap(double gap) {
+        if (gap == 0.0) return "0.00%";
+        return String.format("%.2f%%", gap * 100.0);
+    }
+
+    /**
+     * 生成 KPI Gap 报告字符串（供调用方直接打印）。
+     * 格式：`KPI Gap 报告 (越小越好): [Makespan: cur / LB → Gap=x%] [FlowTime: ...] ...`
+     */
+    public static String generateGapReport(Chromosome chromosome) {
+        double[] objectives = chromosome.getObjectives();
+        double[] lowerBounds = chromosome.getLowerBoundObjectives();
+        if (objectives == null || objectives.length == 0) {
+            return "KPI Gap 报告: 无 objectives 数据";
+        }
+        String[] names = {"Makespan", "FlowTime", "SetupTime", "MachineLoad", "Tardiness"};
+        StringBuilder sb = new StringBuilder("KPI Gap 报告 (越小越好): ");
+        for (int i = 0; i < objectives.length; i++) {
+            double current = objectives[i];
+            double lb = (lowerBounds != null && i < lowerBounds.length) ? lowerBounds[i] : 0.0;
+            double gap;
+            if (lb > 0) {
+                gap = (current - lb) / lb;
+            } else if (current > 0) {
+                gap = current; // lowerBound=0 时 gap = current 本身（绝对偏离量）
+            } else {
+                gap = 0.0;
+            }
+            String name = i < names.length ? names[i] : ("Obj[" + i + "]");
+            sb.append(String.format("[%s: %.4f / LB=%.4f → Gap=%s] ",
+                    name, current, lb, formatGap(gap)));
+        }
+        return sb.toString();
+    }
+}

src/main/java/com/aps/service/Algorithm/RoutingDataService.java

@@ -265,6 +265,7 @@ public class RoutingDataService {
                     List<ProdEquipment> Equipments = ProdEquipments.stream()
                             .filter(t -> t.getExecId().equals(op.getExecId()))
                             .collect(Collectors.toList());
+                    double minProcessingTime=999999999;
                     if (Equipments != null && Equipments.size() > 0) {
                         List<MachineOption> mos = new ArrayList<>();
                         for (ProdEquipment e : Equipments) {
@@ -277,7 +278,7 @@ public class RoutingDataService {
                                 totalprocessTime=e.getSpeed()/e.getSingleOut().doubleValue()*entry.getQuantity();
 
                             }
-
+                            minProcessingTime=Math.min(minProcessingTime,totalprocessTime);
                             if(machineIds.containsKey(e.getEquipId()))
                             {
                                if( machineIds.get(e.getEquipId())<totalprocessTime)
@@ -307,6 +308,7 @@ public class RoutingDataService {
 
                             mos.add(mo);
                         }
+entry.setMinProcessingTime(minProcessingTime);
                         entry.setMachineOptions(mos);
                     }
                 }

src/main/java/com/aps/service/Algorithm/SimulatedAnnealing.java

+// Source code is decompiled from a .class file using FernFlower decompiler (from Intellij IDEA).
 package com.aps.service.Algorithm;
 
 import com.aps.common.util.FileHelper;
 import com.aps.common.util.GlobalCacheUtil;
 import com.aps.common.util.ProductionDeepCopyUtil;
-import com.aps.entity.Algorithm.*;
+import com.aps.entity.Algorithm.Chromosome;
+import com.aps.entity.Algorithm.GAScheduleResult;
 import com.aps.entity.Algorithm.IDAndChildID.GroupResult;
-import com.aps.entity.basic.*;
-
-import java.util.*;
-import java.util.concurrent.*;
+import com.aps.entity.basic.Entry;
+import com.aps.entity.basic.Machine;
+import com.aps.entity.basic.Material;
+import com.aps.entity.basic.Order;
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.Random;
+import java.util.Set;
+import java.util.TreeMap;
+import java.util.concurrent.ArrayBlockingQueue;
+import java.util.concurrent.CopyOnWriteArrayList;
+import java.util.concurrent.ExecutorService;
+import java.util.concurrent.ThreadPoolExecutor;
+import java.util.concurrent.TimeUnit;
 import java.util.stream.Collectors;
 
-/**
- * 模拟退火算法
- */
 public class SimulatedAnnealing {
     private final Random rnd = new Random();
-
-    // ==================== 改进判断参数 ====================
-    private static final double SIGNIFICANT_IMPROVEMENT_THRESHOLD = 0.0001;  // 显著改进阈值：只有改进超过这个值才重置无改进计数
+    private static final double SIGNIFICANT_IMPROVEMENT_THRESHOLD = 1.0E-4;
+    private List<Entry> allOperations;
+    private List<Order> orders;
+    private TreeMap<String, Material> materials;
+    private List<GroupResult> _entryRel;
+    private FitnessCalculator fitnessCalculator;
+    private Map<String, Entry> entrys;
+    private Map<Integer, Entry> entrybyids;
+    private List<Machine> cachedMachines;
+    private List<Order> cachedOrders;
+    private List<GroupResult> cachedEntryRel;
+    private TreeMap<String, Material> cachedMaterials;
+    private List<Entry> cachedAllOperations;
+    private final ExecutorService decodeExecutor;
 
     private void log(String message) {
-        log(message, false);
+        this.log(message, false);
     }
 
     private void log(String message, boolean enableLogging) {
         if (enableLogging) {
             FileHelper.writeLogFile(message);
         }
-    }
-
-    private List<Entry> allOperations;
 
-    private  List<Order> orders;
-    private  TreeMap<String, Material> materials;
-
-    private  List<GroupResult> _entryRel;
-
-    private FitnessCalculator fitnessCalculator;
-
-    private Map<String, Entry> entrys;
-    private  Map<Integer, Entry> entrybyids;
-
-    public SimulatedAnnealing( List<Entry> allOperations, List<Order> orders,
-                               TreeMap<String, Material> materials,List<GroupResult> entryRel,  FitnessCalculator _fitnessCalculator) {
+    }
 
+    public SimulatedAnnealing(List<Entry> allOperations, List<Order> orders, TreeMap<String, Material> materials, List<GroupResult> entryRel, FitnessCalculator _fitnessCalculator) {
+        this.decodeExecutor = new ThreadPoolExecutor(Runtime.getRuntime().availableProcessors() - 1, Runtime.getRuntime().availableProcessors() - 1, 0L, TimeUnit.MILLISECONDS, new ArrayBlockingQueue(200), new ThreadPoolExecutor.CallerRunsPolicy());
         this.allOperations = allOperations;
         this.orders = orders;
         this.materials = materials;
-
-        _entryRel=entryRel;
-
-        Map<Integer, Object> mp = buildEntryKey();
+        this._entryRel = entryRel;
+        Map<Integer, Object> mp = this.buildEntryKey();
         this.fitnessCalculator = _fitnessCalculator;
-        entrys=(Map<String, Entry>)mp.get(1);
-        entrybyids=(Map<Integer, Entry>)mp.get(2);
-
-    }
-    private final ExecutorService decodeExecutor = new ThreadPoolExecutor(
-            Runtime.getRuntime().availableProcessors() - 1, // 核心线程数=CPU-1，无切换开销
-            Runtime.getRuntime().availableProcessors() - 1, // 最大线程数=核心数
-            0L, TimeUnit.MILLISECONDS,
-            new ArrayBlockingQueue<>(200), // 有界队列，避免内存溢出
-            new ThreadPoolExecutor.CallerRunsPolicy() // 任务满了主线程执行，不丢失任务
-    );
-    public List<Chromosome> batchSearch(List<Chromosome> chromosomes,VariableNeighborhoodSearch vns, GeneticDecoder decoder, List<Machine> machines) {
-        List<Chromosome> saHcOptimized=new ArrayList<>();
-//        CompletableFuture.allOf(chromosomes.stream()
-//                        .map(chromosome -> CompletableFuture.runAsync(() -> {
-//                            Chromosome optimized = searchWithHillClimbing(chromosome, decoder, param);
-//                            saHcOptimized.add(optimized);
-//                        }, decodeExecutor))
-//                        .toArray(CompletableFuture[]::new))
-//                .join();
-        for (Chromosome chromosome:chromosomes) {
-            Chromosome optimized = searchWithHillClimbing(chromosome,vns, decoder, machines);
-            saHcOptimized.add(optimized);
-
-        }
-        return saHcOptimized;
+        this.entrys = (Map)mp.get(1);
+        this.entrybyids = (Map)mp.get(2);
+        this.cachedAllOperations = ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList(allOperations), Entry.class);
+        this.cachedOrders = ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList(orders), Order.class);
+        this.cachedEntryRel = ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList(entryRel), GroupResult.class);
+        this.cachedMaterials = ProductionDeepCopyUtil.deepCopyTreeMap(materials, String.class, Material.class);
     }
 
-    public Chromosome batchSearchGetMax(List<Chromosome> chromosomes,VariableNeighborhoodSearch vns, GeneticDecoder decoder, List<Machine> machines) {
-        List<Chromosome> saHcOptimized=batchSearch(chromosomes,vns,decoder,machines);
+    public List<Chromosome> batchSearch(List<Chromosome> chromosomes, VariableNeighborhoodSearch vns, GeneticDecoder decoder, List<Machine> machines) {
+        List<Chromosome> saHcOptimized = new ArrayList();
 
-        int bestidx= Getbest(saHcOptimized,null);
-        if(bestidx>-1) {
-            return saHcOptimized.get(bestidx);
+        for(Chromosome chromosome : chromosomes) {
+            Chromosome optimized = this.searchWithHillClimbing(chromosome, vns, decoder, machines);
+            saHcOptimized.add(optimized);
         }
-        return  null;
 
+        return saHcOptimized;
     }
 
-    /**
-     * 模拟退火搜索，当温度降低到一定程度后切换到爬山法
-     * 流程：模拟退火全局探索（按概率接受劣解）→ 降温 → 温度低时爬山法局部求精 → 输出最优
-     */
-    public Chromosome searchWithHillClimbing(Chromosome chromosome, VariableNeighborhoodSearch vns,GeneticDecoder decoder, List<Machine> machines) {
-        log("模拟退火+爬山法 - 开始执行",true);
-
-        Chromosome current = ProductionDeepCopyUtil.deepCopy(chromosome, Chromosome.class);
-
-        Chromosome best = ProductionDeepCopyUtil.deepCopy(chromosome, Chromosome.class);
-        writeKpi(best);
+    public Chromosome batchSearchGetMax(List<Chromosome> chromosomes, VariableNeighborhoodSearch vns, GeneticDecoder decoder, List<Machine> machines) {
+        List<Chromosome> saHcOptimized = this.batchSearch(chromosomes, vns, decoder, machines);
+        int bestidx = this.Getbest(saHcOptimized, (Chromosome)null);
+        return bestidx > -1 ? (Chromosome)saHcOptimized.get(bestidx) : null;
+    }
 
-        // 记录初始KPI用于计算改进率
-        double[] initialFitnessLevel = best.getFitnessLevel().clone();
+    public Chromosome searchWithHillClimbing(Chromosome chromosome, VariableNeighborhoodSearch vns, GeneticDecoder decoder, List<Machine> machines) {
+        this.log("模拟退火+爬山法 - 开始执行", true);
+        Chromosome current = vns.copyChromosome(chromosome);
+        this.decode(decoder, current, machines);
+        Chromosome best = this.lightCopy(current);
+        this.writeKpi(best);
+        double[] initialFitnessLevel = (double[])best.getFitnessLevel().clone();
         double initialFitness = best.getFitness();
-
-        //  log("模拟退火+爬山法 - 初始化解码完成");
-
-        // 初始化温度（优化：更快收敛）
-        double temperature = 100.0;
-        double coolingRate = 0.90;  // 优化：降温更快
-        double temperatureThreshold = 5.0;  // 优化：温度阈值更高
-        int maxIterations = 100;  // 优化：从300减少到100次
+        double temperature = (double)100.0F;
+        double coolingRate = 0.9;
+        double temperatureThreshold = (double)5.0F;
+        int maxIterations = 100;
         int noImproveCount = 0;
-        int maxNoImprove = 15; // 优化：从30减少到15次
-
-        // 新增：改进率监控参数
-        int stagnantWindow = 15; // 观察窗口大小
-        int[] recentImprovements = new int[stagnantWindow]; // 记录最近窗口内的改进情况
-        double improvementRateThreshold = 0.05; // 改进率阈值(5%)
-
-        log(String.format("模拟退火+爬山法 - 参数配置：温度=%.1f, 降温率=%.2f, 阈值=%.1f, 最大迭代=%d, 最大无改进=%d",
-                temperature, coolingRate, temperatureThreshold, maxIterations, maxNoImprove));
-
+        int maxNoImprove = 15;
+        int stagnantWindow = 15;
+        int[] recentImprovements = new int[stagnantWindow];
+        double improvementRateThreshold = 0.05;
+        this.log(String.format("模拟退火+爬山法 - 参数配置：温度=%.1f, 降温率=%.2f, 阈值=%.1f, 最大迭代=%d, 最大无改进=%d", temperature, coolingRate, temperatureThreshold, maxIterations, maxNoImprove));
         int acceptCount = 0;
         int improveCount = 0;
         int significantImproveCount = 0;
         int totalIterations = 0;
 
-        for (int i = 0; i < maxIterations; i++) {
+        for(int i = 0; i < maxIterations; ++i) {
             totalIterations = i + 1;
             boolean improved = false;
             decoder.DelOrder(current);
-            // 1. 使用智能策略生成邻域解（找瓶颈工序/设备）
             Chromosome neighbor = vns.generateNeighbor(current);
-
-            // 2. 解码
-            decode(decoder, neighbor,machines);
-
-            // 3. 计算能量差
-            double energyDifference = calculateEnergyDifference(neighbor, current);
-
-            // 4. 按概率接受新解（模拟退火核心：有概率接受劣解）
+            this.decode(decoder, neighbor, machines);
+            double energyDifference = this.calculateEnergyDifference(neighbor, current);
             boolean accepted = false;
-            if (energyDifference > 0 || rnd.nextDouble() < Math.exp(energyDifference / temperature)) {
+            if (energyDifference > (double)0.0F || this.rnd.nextDouble() < Math.exp(energyDifference / temperature)) {
                 current = neighbor;
                 accepted = true;
-                acceptCount++;
-
-                // 更新全局最优
-                if (isBetter(current, best)) {
-                    best = ProductionDeepCopyUtil.deepCopy(current, Chromosome.class);
-                    writeKpi(best);
+                ++acceptCount;
+                if (this.isBetter(neighbor, best)) {
+                    best = this.lightCopy(neighbor);
+                    this.writeKpi(best);
                     improved = true;
-                    improveCount++;
-                    boolean isSignificant = isSignificantImprovement(current, best);
+                    ++improveCount;
+                    boolean isSignificant = this.isSignificantImprovement(neighbor, best);
                     if (isSignificant) {
-                        noImproveCount = 0; // 只有显著改进才重置无改进计数
-                        significantImproveCount++;
-                        logImprovementDetails(best, initialFitnessLevel, initialFitness, totalIterations);
-                        log(String.format("模拟退火+爬山法 - 迭代%d：找到更优解(显著)，fitness=%.4f", totalIterations, best.getFitness()),true);
+                        noImproveCount = 0;
+                        ++significantImproveCount;
+                        this.logImprovementDetails(best, initialFitnessLevel, initialFitness, totalIterations);
+                        this.log(String.format("模拟退火+爬山法 - 迭代%d：找到更优解(显著)，fitness=%.4f", totalIterations, best.getFitness()), true);
                     } else {
-                        // 微小改进也接受，但不重置计数
-                        log(String.format("模拟退火+爬山法 - 迭代%d：找到更优解(微小)，fitness=%.4f", totalIterations, best.getFitness()),true);
+                        this.log(String.format("模拟退火+爬山法 - 迭代%d：找到更优解(微小)，fitness=%.4f", totalIterations, best.getFitness()), true);
                     }
                 }
-
             }
 
             if (!improved) {
-                noImproveCount++;
+                ++noImproveCount;
             }
 
-            // 记录本次改进情况
             if (totalIterations <= stagnantWindow) {
                 recentImprovements[totalIterations - 1] = improved ? 1 : 0;
             }
 
-            // 5. 降温
             temperature *= coolingRate;
+            if (totalIterations % 10 == 0) {
+                System.gc();
+            }
 
-            // 每30次迭代输出一次状态
-            if ((totalIterations) % 30 == 0) {
-                log(String.format("模拟退火+爬山法 - 迭代%d/%d：温度=%.4f, 接受数=%d, 改进数=%d, 无改进连续=%d, 总改进率=%.2f%%",
-                        totalIterations, maxIterations, temperature, acceptCount, improveCount, noImproveCount,
-                        totalIterations > 0 ? (double)improveCount / totalIterations * 100 : 0));
+            if (totalIterations % 30 == 0) {
+                this.log(String.format("模拟退火+爬山法 - 迭代%d/%d：温度=%.4f, 接受数=%d, 改进数=%d, 无改进连续=%d, 总改进率=%.2f%%", totalIterations, maxIterations, temperature, acceptCount, improveCount, noImproveCount, totalIterations > 0 ? (double)improveCount / (double)totalIterations * (double)100.0F : (double)0.0F));
             }
 
-            // 6. 提前停止条件
             boolean shouldStop = false;
             String stopReason = "";
-
             if (temperature < temperatureThreshold) {
                 shouldStop = true;
                 stopReason = "温度低于阈值";
@@ -201,194 +164,141 @@ public class SimulatedAnnealing {
                 shouldStop = true;
                 stopReason = String.format("连续无改进达到上限(%d次)", maxNoImprove);
             } else if (totalIterations >= stagnantWindow) {
-                // 检查改进率是否过低
-                double recentImproveRate = calculateRecentImprovementRate(recentImprovements, stagnantWindow);
+                double recentImproveRate = this.calculateRecentImprovementRate(recentImprovements, stagnantWindow);
                 if (recentImproveRate < improvementRateThreshold) {
                     shouldStop = true;
-                    stopReason = String.format("最近%d次迭代改进率过低(%.2f%%)", stagnantWindow, recentImproveRate * 100);
+                    stopReason = String.format("最近%d次迭代改进率过低(%.2f%%)", stagnantWindow, recentImproveRate * (double)100.0F);
                 }
             }
 
             if (shouldStop) {
-                log(String.format("模拟退火+爬山法 - 提前停止：%s，迭代%d次，最终温度=%.4f",
-                        stopReason, totalIterations, temperature));
-                logFinalSummary(best, initialFitnessLevel, initialFitness, improveCount, significantImproveCount, totalIterations);
-
-                log("模拟退火+爬山法 - 切换到爬山法求精");
-                HillClimbing hillClimbing = new HillClimbing( allOperations,orders,materials,_entryRel,fitnessCalculator);
+                this.log(String.format("模拟退火+爬山法 - 提前停止：%s，迭代%d次，最终温度=%.4f", stopReason, totalIterations, temperature));
+                this.logFinalSummary(best, initialFitnessLevel, initialFitness, improveCount, significantImproveCount, totalIterations);
+                this.log("模拟退火+爬山法 - 切换到爬山法求精");
+                HillClimbing hillClimbing = new HillClimbing(this.allOperations, this.orders, this.materials, this._entryRel, this.fitnessCalculator);
                 Chromosome refined = hillClimbing.search(best, decoder, machines);
-                log("模拟退火+爬山法 - 爬山法求精完成");
-
+                this.log("模拟退火+爬山法 - 爬山法求精完成");
                 return refined;
             }
         }
 
-        log(String.format("模拟退火+爬山法 - 完成所有%d次迭代，最终fitness=%.4f",
-                maxIterations, best.getFitness()),true);
-        logFinalSummary(best, initialFitnessLevel, initialFitness, improveCount, significantImproveCount, totalIterations);
-
-        // 7. 输出全局最优排产
+        this.log(String.format("模拟退火+爬山法 - 完成所有%d次迭代，最终fitness=%.4f", maxIterations, best.getFitness()), true);
+        this.logFinalSummary(best, initialFitnessLevel, initialFitness, improveCount, significantImproveCount, totalIterations);
         return best;
     }
 
-    /**
-     * 模拟退火搜索
-     * 流程：模拟退火全局探索（按概率接受劣解）→ 降温 → 输出最优
-     */
-    public Chromosome search(Chromosome chromosome,TabuSearch tabusearch,VariableNeighborhoodSearch vns, GeneticDecoder decoder, List<Machine> machines) {
-        log("模拟退火 - 开始执行",true);
-
+    public Chromosome search(Chromosome chromosome, TabuSearch tabusearch, VariableNeighborhoodSearch vns, GeneticDecoder decoder, List<Machine> machines) {
+        this.log("模拟退火 - 开始执行", true);
         Chromosome current = vns.copyChromosome(chromosome);
-
-        decode(decoder, current,machines);
-
-
-        Chromosome best = ProductionDeepCopyUtil.deepCopy(current, Chromosome.class);
-        writeKpi(best);
-        // 初始化解码
-
-        // 记录初始KPI用于计算改进率
-        double[] initialFitnessLevel = best.getFitnessLevel().clone();
+        this.decode(decoder, current, machines);
+        Chromosome best = this.lightCopy(current);
+        this.writeKpi(best);
+        double[] initialFitnessLevel = (double[])best.getFitnessLevel().clone();
         double initialFitness = best.getFitness();
-
-        //  log("模拟退火+爬山法 - 初始化解码完成");
-
-        // 初始化温度（优化：更快收敛）
-        double temperature = 100.0;
-        double coolingRate = 0.90;  // 优化：降温更快
-        double temperatureThreshold = 5.0;  // 优化：温度阈值更高
-        int maxIterations = 80;  // 优化：从300减少到80次
+        double temperature = (double)100.0F;
+        double coolingRate = 0.9;
+        double temperatureThreshold = (double)5.0F;
+        int maxIterations = 80;
         int noImproveCount = 0;
-        int maxNoImprove = 10; // 优化：从20减少到10次
-
-        // 新增：改进率监控参数
-        int stagnantWindow = 10; // 观察窗口大小
-        int[] recentImprovements = new int[stagnantWindow]; // 记录最近窗口内的改进情况
-        double improvementRateThreshold = 0.001; // 改进率阈值
-
-        log(String.format("模拟退火 - 参数配置：温度=%.1f, 降温率=%.2f, 阈值=%.1f, 最大迭代=%d, 最大无改进=%d",
-                temperature, coolingRate, temperatureThreshold, maxIterations, maxNoImprove));
-
+        int maxNoImprove = 10;
+        int stagnantWindow = 10;
+        int[] recentImprovements = new int[stagnantWindow];
+        double improvementRateThreshold = 0.001;
+        this.log(String.format("模拟退火 - 参数配置：温度=%.1f, 降温率=%.2f, 阈值=%.1f, 最大迭代=%d, 最大无改进=%d", temperature, coolingRate, temperatureThreshold, maxIterations, maxNoImprove));
         int acceptCount = 0;
         int improveCount = 0;
         int significantImproveCount = 0;
         int totalIterations = 0;
 
-        for (int i = 0; i < maxIterations; i++) {
+        for(int i = 0; i < maxIterations; ++i) {
             totalIterations = i + 1;
             boolean improved = false;
-            log(String.format("模拟退火 - 迭代%d：", totalIterations));
+            this.log(String.format("模拟退火 - 迭代%d：", totalIterations));
             decoder.DelOrder(current);
-
-            // 1. 使用智能策略生成邻域解（找瓶颈工序/设备）
             Chromosome neighbor = vns.generateNeighbor(current);
-            // 2. 解码
-            decode(decoder, neighbor,machines);
-            // 跳过禁忌解（除非是最优解）
-            if (tabusearch.isTabu(neighbor.getGeneStr()) && !isBetter(current,best)) {
+            this.decode(decoder, neighbor, machines);
+            if (tabusearch.isTabu(neighbor.getGeneStr()) && !this.isBetter(current, best)) {
                 temperature *= coolingRate;
-                noImproveCount++;
-                // 记录本次无改进
+                ++noImproveCount;
                 if (totalIterations <= stagnantWindow) {
                     recentImprovements[totalIterations - 1] = 0;
                 }
-                continue;
-            }
-
-            // 3. 计算能量差
-            double energyDifference = calculateEnergyDifference(neighbor, current);
-
-            // 4. 按概率接受新解（模拟退火核心：有概率接受劣解）
-            boolean accepted = false;
-            if (energyDifference > 0 || rnd.nextDouble() < Math.exp(energyDifference / temperature)) {
-                current = neighbor;
-
-                acceptCount++;
-                if (isBetter(current, best)) {
-                    best = ProductionDeepCopyUtil.deepCopy(current, Chromosome.class);
-                    tabusearch.addToTabuList(best.getGeneStr());
-                    writeKpi(best);
-                    improved = true;
-                    improveCount++;
-                    boolean isSignificant = isSignificantImprovement(current, best);
-                    if (isSignificant) {
-                        noImproveCount = 0;
-                        significantImproveCount++;
-                        logImprovementDetails(best, initialFitnessLevel, initialFitness, totalIterations);
-                        log(String.format("模拟退火 - 迭代%d：找到更优解(显著)，fitness=%.4f", totalIterations, best.getFitness()));
-                    } else {
-                        log(String.format("模拟退火 - 迭代%d：找到更优解(微小)，fitness=%.4f", totalIterations, best.getFitness()));
+            } else {
+                double energyDifference = this.calculateEnergyDifference(neighbor, current);
+                boolean accepted = false;
+                if (energyDifference > (double)0.0F || this.rnd.nextDouble() < Math.exp(energyDifference / temperature)) {
+                    current = neighbor;
+                    ++acceptCount;
+                    if (this.isBetter(neighbor, best)) {
+                        best = this.lightCopy(neighbor);
+                        tabusearch.addToTabuList(best.getGeneStr());
+                        this.writeKpi(best);
+                        improved = true;
+                        ++improveCount;
+                        boolean isSignificant = this.isSignificantImprovement(neighbor, best);
+                        if (isSignificant) {
+                            noImproveCount = 0;
+                            ++significantImproveCount;
+                            this.logImprovementDetails(best, initialFitnessLevel, initialFitness, totalIterations);
+                            this.log(String.format("模拟退火 - 迭代%d：找到更优解(显著)，fitness=%.4f", totalIterations, best.getFitness()));
+                        } else {
+                            this.log(String.format("模拟退火 - 迭代%d：找到更优解(微小)，fitness=%.4f", totalIterations, best.getFitness()));
+                        }
                     }
                 }
-              //  decoder.DelOrder(current);
-            }
 
-            if (!improved) {
-                noImproveCount++;
-            }
-
-            // 记录本次改进情况
-            if (totalIterations <= stagnantWindow) {
-                recentImprovements[totalIterations - 1] = improved ? 1 : 0;
-            }
-
-            // 5. 降温
-            temperature *= coolingRate;
+                if (!improved) {
+                    ++noImproveCount;
+                }
 
-            // 每次迭代都输出状态
-            log(String.format("模拟退火 - 迭代%d/%d：温度=%.4f, 接受数=%d, 改进数=%d, 无改进连续=%d, 总改进率=%.2f%%",
-                    totalIterations, maxIterations, temperature, acceptCount, improveCount, noImproveCount,
-                    totalIterations > 0 ? (double)improveCount / totalIterations * 100 : 0));
+                if (totalIterations <= stagnantWindow) {
+                    recentImprovements[totalIterations - 1] = improved ? 1 : 0;
+                }
 
-            // 6. 检查提前停止条件
-            boolean shouldStop = false;
-            String stopReason = "";
+                temperature *= coolingRate;
+                if (totalIterations % 10 == 0) {
+                    System.gc();
+                }
 
-            if (temperature < temperatureThreshold) {
-                shouldStop = true;
-                stopReason = "温度低于阈值";
-            } else if (noImproveCount >= maxNoImprove) {
-                shouldStop = true;
-                stopReason = String.format("连续无改进达到上限(%d次)", maxNoImprove);
-            } else if (totalIterations >= stagnantWindow) {
-                // 检查改进率是否过低
-                double recentImproveRate = calculateRecentImprovementRate(recentImprovements, stagnantWindow);
-                if (recentImproveRate < improvementRateThreshold) {
+                this.log(String.format("模拟退火 - 迭代%d/%d：温度=%.4f, 接受数=%d, 改进数=%d, 无改进连续=%d, 总改进率=%.2f%%", totalIterations, maxIterations, temperature, acceptCount, improveCount, noImproveCount, totalIterations > 0 ? (double)improveCount / (double)totalIterations * (double)100.0F : (double)0.0F));
+                boolean shouldStop = false;
+                String stopReason = "";
+                if (temperature < temperatureThreshold) {
                     shouldStop = true;
-                    stopReason = String.format("最近%d次迭代改进率过低(%.2f%%)", stagnantWindow, recentImproveRate * 100);
+                    stopReason = "温度低于阈值";
+                } else if (noImproveCount >= maxNoImprove) {
+                    shouldStop = true;
+                    stopReason = String.format("连续无改进达到上限(%d次)", maxNoImprove);
+                } else if (totalIterations >= stagnantWindow) {
+                    double recentImproveRate = this.calculateRecentImprovementRate(recentImprovements, stagnantWindow);
+                    if (recentImproveRate < improvementRateThreshold) {
+                        shouldStop = true;
+                        stopReason = String.format("最近%d次迭代改进率过低(%.2f%%)", stagnantWindow, recentImproveRate * (double)100.0F);
+                    }
                 }
-            }
 
-            if (shouldStop) {
-                log(String.format("模拟退火 - 提前停止：%s，迭代%d次，最终温度=%.4f",
-                        stopReason, totalIterations, temperature));
-                logFinalSummary(best, initialFitnessLevel, initialFitness, improveCount, significantImproveCount, totalIterations);
-                return best;
+                if (shouldStop) {
+                    this.log(String.format("模拟退火 - 提前停止：%s，迭代%d次，最终温度=%.4f", stopReason, totalIterations, temperature));
+                    this.logFinalSummary(best, initialFitnessLevel, initialFitness, improveCount, significantImproveCount, totalIterations);
+                    return best;
+                }
             }
         }
 
-        log(String.format("模拟退火 - 完成所有%d次迭代，最终fitness=%.4f",
-                maxIterations, best.getFitness()),true);
-        logFinalSummary(best, initialFitnessLevel, initialFitness, improveCount, significantImproveCount, totalIterations);
-
-        // 7. 输出全局最优排产
+        this.log(String.format("模拟退火 - 完成所有%d次迭代，最终fitness=%.4f", maxIterations, best.getFitness()), true);
+        this.logFinalSummary(best, initialFitnessLevel, initialFitness, improveCount, significantImproveCount, totalIterations);
         return best;
     }
 
-    /**
-     * 记录改进详情
-     */
     private void logImprovementDetails(Chromosome best, double[] initialFitnessLevel, double initialFitness, int iteration) {
         StringBuilder sb = new StringBuilder("模拟退火 - 改进详情: 迭代" + iteration + ", ");
         double[] currentFitness = best.getFitnessLevel();
-
-        // 处理null或空数组的情况
-        if (currentFitness != null && currentFitness.length > 0 &&
-                initialFitnessLevel != null && initialFitnessLevel.length > 0) {
+        if (currentFitness != null && currentFitness.length > 0 && initialFitnessLevel != null && initialFitnessLevel.length > 0) {
             int minLength = Math.min(currentFitness.length, initialFitnessLevel.length);
-            for (int i = 0; i < minLength; i++) {
+
+            for(int i = 0; i < minLength; ++i) {
                 double improvement = currentFitness[i] - initialFitnessLevel[i];
-                sb.append(String.format("KPI%d: %.4f→%.4f(+%.4f) ", i+1, initialFitnessLevel[i], currentFitness[i], improvement));
+                sb.append(String.format("KPI%d: %.4f→%.4f(+%.4f) ", i + 1, initialFitnessLevel[i], currentFitness[i], improvement));
             }
         } else {
             sb.append("(KPI数据不可用) ");
@@ -396,219 +306,191 @@ public class SimulatedAnnealing {
 
         double totalImprovement = best.getFitness() - initialFitness;
         sb.append(String.format("总Fitness: %.4f→%.4f(+%.4f)", initialFitness, best.getFitness(), totalImprovement));
-        log(sb.toString());
+        this.log(sb.toString());
     }
 
-    /**
-     * 计算最近改进率
-     */
     private double calculateRecentImprovementRate(int[] recentImprovements, int windowSize) {
         int improveCount = 0;
-        for (int i = 0; i < windowSize; i++) {
+
+        for(int i = 0; i < windowSize; ++i) {
             improveCount += recentImprovements[i];
         }
-        return (double) improveCount / windowSize;
+
+        return (double)improveCount / (double)windowSize;
     }
 
-    /**
-     * 记录最终总结
-     */
     private void logFinalSummary(Chromosome best, double[] initialFitnessLevel, double initialFitness, int improveCount, int significantImproveCount, int totalIterations) {
         StringBuilder sb = new StringBuilder("模拟退火 - 最终总结: ");
         double[] currentFitness = best.getFitnessLevel();
-
-        sb.append(String.format("总迭代%d次, 成功改进%d次(显著%d次), 改进率%.2f%%. ",
-                totalIterations, improveCount, significantImproveCount,
-                totalIterations > 0 ? (double)improveCount / totalIterations * 100 : 0));
-
-        // 处理null或空数组的情况
-        if (currentFitness != null && currentFitness.length > 0 &&
-                initialFitnessLevel != null && initialFitnessLevel.length > 0) {
+        sb.append(String.format("总迭代%d次, 成功改进%d次(显著%d次), 改进率%.2f%%. ", totalIterations, improveCount, significantImproveCount, totalIterations > 0 ? (double)improveCount / (double)totalIterations * (double)100.0F : (double)0.0F));
+        if (currentFitness != null && currentFitness.length > 0 && initialFitnessLevel != null && initialFitnessLevel.length > 0) {
             int minLength = Math.min(currentFitness.length, initialFitnessLevel.length);
-            for (int i = 0; i < minLength; i++) {
+
+            for(int i = 0; i < minLength; ++i) {
                 double improvement = currentFitness[i] - initialFitnessLevel[i];
-                sb.append(String.format("KPI%d: %.4f→%.4f(%.2f%%) ", i+1, initialFitnessLevel[i], currentFitness[i],
-                        initialFitnessLevel[i] > 0 ? improvement / initialFitnessLevel[i] * 100 : 0));
+                sb.append(String.format("KPI%d: %.4f→%.4f(%.2f%%) ", i + 1, initialFitnessLevel[i], currentFitness[i], initialFitnessLevel[i] > (double)0.0F ? improvement / initialFitnessLevel[i] * (double)100.0F : (double)0.0F));
             }
         } else {
             sb.append("(KPI数据不可用) ");
         }
 
         double totalImprovement = best.getFitness() - initialFitness;
-        sb.append(String.format("总Fitness: %.4f→%.4f(%.2f%%)", initialFitness, best.getFitness(),
-                initialFitness > 0 ? totalImprovement / initialFitness * 100 : 0));
-        log(sb.toString());
+        sb.append(String.format("总Fitness: %.4f→%.4f(%.2f%%)", initialFitness, best.getFitness(), initialFitness > (double)0.0F ? totalImprovement / initialFitness * (double)100.0F : (double)0.0F));
+        this.log(sb.toString());
     }
 
-
-    private  void  writeKpi(Chromosome chromosome) {
+    private void writeKpi(Chromosome chromosome) {
         String fitness = "";
         double[] fitness1 = chromosome.getFitnessLevel();
         if (fitness1 != null) {
-            for (int i = 0; i < fitness1.length; i++) {
-                fitness += fitness1[i] + ",";
+            for(int i = 0; i < fitness1.length; ++i) {
+                fitness = fitness + fitness1[i] + ",";
             }
         } else {
             fitness = "null (未计算)";
         }
 
-        log(String.format("模拟退火 - kpi：%s", fitness),true);
-        if(chromosome.getMakespan()!=0) {
-            FileHelper.writeLogFile(String.format("模拟退火 - kpi-Makespan: %f",  chromosome.getMakespan()));
+        this.log(String.format("模拟退火 - kpi：%s", fitness), true);
+        if (chromosome.getMakespan() != (double)0.0F) {
+            FileHelper.writeLogFile(String.format("模拟退火 - kpi-Makespan: %f", chromosome.getMakespan()));
         }
-        if(chromosome.getDelayTime()!=0) {
-            FileHelper.writeLogFile(String.format("模拟退火 - kpi-DelayTime: %f",  chromosome.getDelayTime()));
+
+        if (chromosome.getDelayTime() != (double)0.0F) {
+            FileHelper.writeLogFile(String.format("模拟退火 - kpi-DelayTime: %f", chromosome.getDelayTime()));
         }
-        if(chromosome.getTotalChangeoverTime()!=0) {
 
+        if (chromosome.getTotalChangeoverTime() != (double)0.0F) {
             FileHelper.writeLogFile(String.format("模拟退火 - kpi-ChangeoverTime: %f", chromosome.getTotalChangeoverTime()));
         }
-        if(chromosome.getMachineLoadStd()!=0) {
+
+        if (chromosome.getMachineLoadStd() != (double)0.0F) {
             FileHelper.writeLogFile(String.format("模拟退火 - kpi-MachineLoad: %f", chromosome.getMachineLoadStd()));
         }
-        if(chromosome.getTotalFlowTime()!=0) {
+
+        if (chromosome.getTotalFlowTime() != (double)0.0F) {
             FileHelper.writeLogFile(String.format("模拟退火 - kpi-FlowTime: %f", chromosome.getTotalFlowTime()));
         }
+
     }
 
-    /**
-     * 计算能量差（基于fitnessLevel数组的比较）
-     */
     private double calculateEnergyDifference(Chromosome neighbor, Chromosome current) {
         double[] neighborFitness = neighbor.getFitnessLevel();
         double[] currentFitness = current.getFitnessLevel();
+        double diff = (double)0.0F;
 
-        // 计算加权能量差
-        double diff = 0;
-        for (int i = 0; i < neighborFitness.length; i++) {
-            diff += (neighborFitness[i] - currentFitness[i]);
+        for(int i = 0; i < neighborFitness.length; ++i) {
+            diff += neighborFitness[i] - currentFitness[i];
         }
+
         return diff;
     }
 
-    /**
-     * 按优先级分组工序
-     */
     private Map<Double, List<Entry>> groupOperationsByPriority() {
-        Map<Double, List<Entry>> groups = new HashMap<>();
-        for (Entry op : allOperations) {
+        Map<Double, List<Entry>> groups = new HashMap();
+
+        for(Entry op : this.allOperations) {
             double priority = op.getPriority();
-            groups.computeIfAbsent(priority, k -> new ArrayList<>()).add(op);
+            ((List)groups.computeIfAbsent(priority, (k) -> new ArrayList())).add(op);
         }
 
-        // 过滤掉：设备只有一个且只有一个GroupId的优先级组
-        Map<Double, List<Entry>> filteredGroups = new HashMap<>();
-        for (Map.Entry<Double, List<Entry>> entry : groups.entrySet()) {
-            List<Entry> ops = entry.getValue();
-
-            // 检查是否所有工序都只有一个设备选项
-            boolean allSingleMachine = ops.stream()
-                    .allMatch(op -> op.getMachineOptions().size() <= 1);
+        Map<Double, List<Entry>> filteredGroups = new HashMap();
 
-            // 检查是否只有一个GroupId
-            Set<Integer> groupIds = ops.stream()
-                    .map(Entry::getGroupId)
-                    .collect(Collectors.toSet());
-
-            // 如果两个条件都满足，过滤掉这个优先级组
-            if (!(allSingleMachine && groupIds.size() <= 1)) {
-                filteredGroups.put(entry.getKey(), ops);
+        for(Map.Entry<Double, List<Entry>> entry : groups.entrySet()) {
+            List<Entry> ops = (List)entry.getValue();
+            boolean allSingleMachine = ops.stream().allMatch((opx) -> opx.getMachineOptions().size() <= 1);
+            Set<Integer> groupIds = (Set)ops.stream().map(Entry::getGroupId).collect(Collectors.toSet());
+            if (!allSingleMachine || groupIds.size() > 1) {
+                filteredGroups.put((Double)entry.getKey(), ops);
             }
         }
 
         return filteredGroups;
     }
 
-    /**
-     * 构建Entry索引：op.getGroupId() + "_" + op.getSequence() -> Entry
-     */
     private Map<Integer, Object> buildEntryKey() {
-        Map<Integer, Object> index0 = new HashMap<>();
-        Map<String, Entry> index = new HashMap<>();
-        Map<Integer, Entry> index2 = new HashMap<>();
-        List<Entry> allOps = this.allOperations;
+        Map<Integer, Object> index0 = new HashMap();
+        Map<String, Entry> index = new HashMap();
+        Map<Integer, Entry> index2 = new HashMap();
 
-        for (Entry op : allOps) {
+        for(Entry op : this.allOperations) {
             String key = op.getGroupId() + "_" + op.getSequence();
-
             index.put(key, op);
             index2.put(op.getId(), op);
         }
-        index0.put(1,index);
-        index0.put(2,index2);
+
+        index0.put(1, index);
+        index0.put(2, index2);
         return index0;
     }
 
-    /**
-     * 解码染色体
-     */
-    /**
-     * 解码染色体
-     */
-    private void decode(GeneticDecoder decoder, Chromosome chromosome , List<Machine> machines) {
-        chromosome.setResult(new CopyOnWriteArrayList<>());
-
-        // 假设Machine类有拷贝方法，或使用MapStruct等工具进行映射
-
-        chromosome.setMachines(ProductionDeepCopyUtil.deepCopyList(machines,Machine.class)); // 简单拷贝，实际可能需要深拷贝
-        chromosome.setOrders(ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList<>(orders), Order.class) ); // 简单拷贝，实际可能需要深拷贝
-        chromosome.setOperatRel(ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList<>(_entryRel), GroupResult.class) ); // 简单拷贝，实际可能需要深拷贝
-        chromosome.setMaterials(ProductionDeepCopyUtil.deepCopyTreeMap(materials,String.class, Material.class)); // 简单拷贝，实际可能需要深拷贝
-
-        chromosome.setAllOperations(ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList<>(allOperations), Entry.class) ); // 简单拷贝，实际可能需要深拷贝
-        //chromosome.setObjectiveWeights(_objectiveWeights);
-        //  chromosome.setBaseTime(param.getBaseTime());
-        //  chromosome.setInitMachines(ProductionDeepCopyUtil.deepCopyList(machines,Machine.class)); // 简单拷贝，实际可能需要深拷贝
-        //   _sceneService.saveChromosomeToFile(chromosome, "12345679");
-
-        // 加载锁定工单到ResultOld
-        List<GAScheduleResult> lockedOrders = GlobalCacheUtil.get("locked_orders_" + chromosome.getScenarioID());
+    private void decode(GeneticDecoder decoder, Chromosome chromosome, List<Machine> machines) {
+        chromosome.setResult(new CopyOnWriteArrayList());
+        if (this.cachedMachines == null) {
+            this.cachedMachines = ProductionDeepCopyUtil.deepCopyList(machines, Machine.class);
+        }
+
+        chromosome.setMachines(ProductionDeepCopyUtil.deepCopyList(this.cachedMachines, Machine.class));
+        chromosome.setOrders(ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList(this.cachedOrders), Order.class));
+        chromosome.setOperatRel(ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList(this.cachedEntryRel), GroupResult.class));
+        chromosome.setMaterials(ProductionDeepCopyUtil.deepCopyTreeMap(this.cachedMaterials, String.class, Material.class));
+        chromosome.setAllOperations(ProductionDeepCopyUtil.deepCopyList(new CopyOnWriteArrayList(this.cachedAllOperations), Entry.class));
+        List<GAScheduleResult> lockedOrders = (List)GlobalCacheUtil.get("locked_orders_" + chromosome.getScenarioID());
         if (lockedOrders != null && !lockedOrders.isEmpty()) {
             chromosome.setResultOld(ProductionDeepCopyUtil.deepCopyList(lockedOrders, GAScheduleResult.class));
-            log("将 " + lockedOrders.size() + " 个锁定工单加载到初始种群中");
+            this.log("将 " + lockedOrders.size() + " 个锁定工单加载到初始种群中");
         } else {
-            chromosome.setResultOld(new CopyOnWriteArrayList<>());
+            chromosome.setResultOld(new CopyOnWriteArrayList());
         }
-        decoder.decodeChromosomeWithCache(chromosome,false);
 
+        decoder.decodeChromosomeWithCache(chromosome, false);
     }
 
+    private Chromosome lightCopy(Chromosome source) {
+        Chromosome copy = new Chromosome();
+        copy.setOperationSequencing(new CopyOnWriteArrayList(source.getOperationSequencing()));
+        copy.setMachineSelection(new CopyOnWriteArrayList(source.getMachineSelection()));
+        copy.setGlobalOpList(new CopyOnWriteArrayList(source.getGlobalOpList()));
+        copy.setOrders(new CopyOnWriteArrayList(source.getOrders()));
+        copy.setAllOperations(new CopyOnWriteArrayList(source.getAllOperations()));
+        copy.setResult(source.getResult());
+        copy.setMachines(source.getMachines());
+        copy.setOperatRel(new CopyOnWriteArrayList(source.getOperatRel()));
+        copy.setScenarioID(source.getScenarioID());
+        copy.setBaseTime(source.getBaseTime());
+        copy.setGenerateType(source.getGenerateType());
+        copy.setFitnessLevel(source.getFitnessLevel());
+        copy.setFitness(source.getFitness());
+        return copy;
+    }
 
-    /**
-     * 比较两个染色体的优劣（基于fitnessLevel多层次比较）
-     */
     private boolean isBetter(Chromosome c1, Chromosome c2) {
-        return fitnessCalculator.isBetter(c1,c2);
+        return this.fitnessCalculator.isBetter(c1, c2);
     }
 
-    /**
-     * 判断是否为显著改进（只有超过阈值的改进才重置无改进计数）
-     */
     private boolean isSignificantImprovement(Chromosome newChromo, Chromosome oldChromo) {
-        if (!isBetter(newChromo, oldChromo)) {
+        if (!this.isBetter(newChromo, oldChromo)) {
             return false;
+        } else {
+            double newFitness = newChromo.getFitness();
+            double oldFitness = oldChromo.getFitness();
+            return newFitness - oldFitness > 1.0E-4;
         }
-        double newFitness = newChromo.getFitness();
-        double oldFitness = oldChromo.getFitness();
-        return (newFitness - oldFitness) > SIGNIFICANT_IMPROVEMENT_THRESHOLD;
     }
-    private int  Getbest(List<Chromosome> candidates,Chromosome best) {
-        // 找出最佳候选方案
 
+    private int Getbest(List<Chromosome> candidates, Chromosome best) {
         int bestidx = -1;
-        if(best==null)
-        {
-            best=candidates.get(0);
-            bestidx=0;
+        if (best == null) {
+            best = (Chromosome)candidates.get(0);
+            bestidx = 0;
         }
-        for (int i = 0; i < candidates.size(); i++) {
-            Chromosome candidate = candidates.get(i);
-            if (isBetter(candidate, best)) {
+
+        for(int i = 0; i < candidates.size(); ++i) {
+            Chromosome candidate = (Chromosome)candidates.get(i);
+            if (this.isBetter(candidate, best)) {
                 bestidx = i;
             }
         }
 
         return bestidx;
     }
-
-}
+}
\ No newline at end of file

src/main/java/com/aps/service/Algorithm/VariableNeighborhoodSearch.java

@@ -71,32 +71,8 @@ public class VariableNeighborhoodSearch {
     private static final double DIVERSITY_WEIGHT = 0.4; // 设备选择多样性权重（越高越倾向选择次数少的设备）
     private static final double RANDOM_NOISE_FOR_MACHINE = 0.1; // 机器选择的随机扰动因子
 
-    // 日志级别
-    private static final int LOG_LEVEL_DEBUG = 0;
-    private static final int LOG_LEVEL_INFO = 1;
-    private static final int LOG_LEVEL_WARN = 2;
-    private int currentLogLevel = LOG_LEVEL_INFO;
-
-    // 局部搜索优化
     private static final int MAX_LOCAL_SEARCH_NEIGHBORS = 2; // 从5减少到2，大幅减少解码次数
 
-    private void log(String message) {
-        log(message, LOG_LEVEL_INFO, false);
-    }
-
-    private void log(String message, boolean enableLogging) {
-        log(message, LOG_LEVEL_INFO, enableLogging);
-    }
-
-    private void log(String message, int level) {
-        log(message, level, false);
-    }
-
-    private void log(String message, int level, boolean enableLogging) {
-        if (enableLogging && level >= currentLogLevel) {
-            FileHelper.writeLogFile(message);
-        }
-    }
 
     private List<Entry> allOperations;
 
@@ -140,6 +116,23 @@ public class VariableNeighborhoodSearch {
     private TreeMap<String, Material> cachedMaterials;
     private List<Entry> cachedAllOperations;
 
+    // CP-SAT 邻域需要的缓存
+    private GeneticDecoder cpSatDecoder;
+    private List<Machine> cpSatMachines;
+    private ObjectiveWeights cpSatWeights;
+    private boolean cpSatEnabled = false;
+
+    /**
+     * 启用 CP-SAT 邻域
+     */
+    public void enableCpSatNeighborhood(GeneticDecoder decoder, List<Machine> machines, ObjectiveWeights weights) {
+        this.cpSatDecoder = decoder;
+        this.cpSatMachines = machines;
+        this.cpSatWeights = weights;
+        this.cpSatEnabled = true;
+        FileHelper.writeLogFile("[VNS-CpSat] CP-SAT 邻域已启用");
+    }
+
     private  GeneticOperations geneticOperations;
 
     // 邻域结构成功率统计（用于 search() 方法）
@@ -261,26 +254,41 @@ public class VariableNeighborhoodSearch {
         // HybridShake - 混合抖动：综合调整
         neighborhoods.add(new NeighborhoodStructure("HybridShake", this::hybridShakeWrapper));
 
+        // CpSatLocalOptimize - CP-SAT 局部重优化（仅当启用时）
+        neighborhoods.add(new NeighborhoodStructure("CpSatLocalOptimize", this::cpSatLocalOptimizeWrapper));
+
         return neighborhoods;
     }
 
     /**
-     * 定义邻域结构（按成功率排序）
+     * 定义邻域结构（按成功率排序，CP-SAT 始终保留一个名额）
      */
     private List<NeighborhoodStructure> defineNeighborhoods() {
         // 按成功率排序
         List<NeighborhoodWithStats> sorted = new ArrayList<>(neighborhoodsWithStats);
         sorted.sort((a, b) -> Double.compare(b.getSuccessRate(), a.getSuccessRate()));
 
-        // 提取 NeighborhoodStructure
         List<NeighborhoodStructure> result = new ArrayList<>();
-        // for (NeighborhoodWithStats ns : sorted) {
-        //     result.add(ns.structure);
-        // }
 
+        // 始终保留 CP-SAT 邻域（避免因成功率低被挤出前 N 名后永远无法被选中）
+        NeighborhoodWithStats cpSatNs = null;
         int maxNeighborhoods = Math.min(MAX_NEIGHBORHOODS, sorted.size());
         for (int i = 0; i < maxNeighborhoods; i++) {
-            result.add(sorted.get(i).structure);
+            NeighborhoodWithStats ns = sorted.get(i);
+            if ("CpSatLocalOptimize".equals(ns.structure.name)) {
+                cpSatNs = ns;
+            }
+            result.add(ns.structure);
+        }
+
+        // 如果 CP-SAT 未被前 N 名选中且已启用，额外追加
+        if (cpSatNs == null && cpSatEnabled) {
+            for (NeighborhoodWithStats ns : neighborhoodsWithStats) {
+                if ("CpSatLocalOptimize".equals(ns.structure.name)) {
+                    result.add(ns.structure);
+                    break;
+                }
+            }
         }
 
         return result;
@@ -304,7 +312,7 @@ public class VariableNeighborhoodSearch {
     /**
      * 对种群中的每个个体进行变邻域搜索
      */
-    public List<Chromosome> search(List<Chromosome> population,TabuSearch tabuSearch, GeneticDecoder decoder, List<Machine> machines) {
+    public List<Chromosome> search(List<Chromosome> population, TabuSearch tabuSearch, GeneticDecoder decoder, List<Machine> machines) {
         List<Chromosome> improvedPopulation = new ArrayList<>();
 
         for (Chromosome chromosome : population) {
@@ -339,8 +347,9 @@ public class VariableNeighborhoodSearch {
         log("变邻域搜索(共用禁忌表) - 开始执行",true);
 
         // 深拷贝当前染色体
-        Chromosome current = ProductionDeepCopyUtil.deepCopy(chromosome, Chromosome.class);
-        Chromosome best = ProductionDeepCopyUtil.deepCopy(chromosome, Chromosome.class);
+        Chromosome current = lightCopy(chromosome);
+        Chromosome best = copyChromosome(chromosome);
+
         writeKpi(best);
 
         // 记录初始KPI用于跟踪改进
@@ -421,9 +430,9 @@ public class VariableNeighborhoodSearch {
                 }
 
                 if (accept) {
-                    current = ProductionDeepCopyUtil.deepCopy(localBest, Chromosome.class);
+                    current = lightCopy(localBest);
                     if (betterThanBest) {
-                        best = ProductionDeepCopyUtil.deepCopy(localBest, Chromosome.class);
+                        best = lightCopy(localBest);
                         writeKpi(best);
                         totalImprovements++;
                         roundHadImprovement = true;
@@ -535,6 +544,8 @@ public class VariableNeighborhoodSearch {
 
         logVNSFinalSummary(best, initialFitnessLevel, initialFitness, totalRounds, totalImprovements, totalSignificantImprovements);
         log(String.format("变邻域搜索(融合禁忌) - 结束, 总轮次=%d", totalRounds), true);
+
+        decode(decoder, best, machines);
         return best;
     }
 
@@ -616,21 +627,25 @@ public class VariableNeighborhoodSearch {
 
         log(String.format("变邻域搜索 - kpi：%s", fitness),true);
         if(chromosome.getMakespan()!=0) {
-            FileHelper.writeLogFile(String.format("变邻域搜索 - kpi-Makespan: %f",  chromosome.getMakespan()));
+            log(String.format("变邻域搜索 - kpi-Makespan: %f",  chromosome.getMakespan()));
         }
         if(chromosome.getDelayTime()!=0) {
-            FileHelper.writeLogFile(String.format("变邻域搜索 - kpi-DelayTime: %f",  chromosome.getDelayTime()));
+            log(String.format("变邻域搜索 - kpi-DelayTime: %f",  chromosome.getDelayTime()));
         }
         if(chromosome.getTotalChangeoverTime()!=0) {
 
-            FileHelper.writeLogFile(String.format("变邻域搜索 - kpi-ChangeoverTime: %f", chromosome.getTotalChangeoverTime()));
+            log(String.format("变邻域搜索 - kpi-ChangeoverTime: %f", chromosome.getTotalChangeoverTime()));
         }
         if(chromosome.getMachineLoadStd()!=0) {
-            FileHelper.writeLogFile(String.format("变邻域搜索 - kpi-MachineLoad: %f", chromosome.getMachineLoadStd()));
+            log(String.format("变邻域搜索 - kpi-MachineLoad: %f", chromosome.getMachineLoadStd()));
         }
         if(chromosome.getTotalFlowTime()!=0) {
-            FileHelper.writeLogFile(String.format("变邻域搜索 - kpi-FlowTime: %f",chromosome.getTotalFlowTime()));
+            log(String.format("变邻域搜索 - kpi-FlowTime: %f",chromosome.getTotalFlowTime()));
         }
+
+        // ==================== 打印各 KPI 的 Gap ====================
+        log(KpiLowerBoundCalculator.generateGapReport(chromosome));
+
     }
 
     /**
@@ -2125,12 +2140,23 @@ public class VariableNeighborhoodSearch {
         Chromosome neighbor=new  Chromosome();
         neighbor.setGenerateType(chromosome.getGenerateType());
         neighbor.setID(UUID.randomUUID().toString());
-        neighbor.setOperationSequencing(chromosome.getOperationSequencing());
-        neighbor.setMachineSelection(chromosome.getMachineSelection());
+        neighbor.setOperationSequencing(new CopyOnWriteArrayList<>(chromosome.getOperationSequencing()));
+        neighbor.setMachineSelection(new CopyOnWriteArrayList<>(chromosome.getMachineSelection()));
+        neighbor.setGlobalOpList(new CopyOnWriteArrayList<>(chromosome.getGlobalOpList()));
+
         neighbor.setScenarioID(chromosome.getScenarioID());
         neighbor.setBaseTime(chromosome.getBaseTime());
         neighbor.setFitnessLevel(chromosome.getFitnessLevel());
-        neighbor.setGlobalOpList(chromosome.getGlobalOpList());
+        // 拷贝 orders/allOperations/operatRel，确保 DelOrder 能正常清理 SF 工序
+        if (chromosome.getOrders() != null) {
+            neighbor.setOrders(new CopyOnWriteArrayList<>(chromosome.getOrders()));
+        }
+        if (chromosome.getAllOperations() != null) {
+            neighbor.setAllOperations(new CopyOnWriteArrayList<>(chromosome.getAllOperations()));
+        }
+        if (chromosome.getOperatRel() != null) {
+            neighbor.setOperatRel(new CopyOnWriteArrayList<>(chromosome.getOperatRel()));
+        }
         return neighbor;
     }
 
@@ -2380,25 +2406,36 @@ public class VariableNeighborhoodSearch {
         CopyOnWriteArrayList<Integer> os = neighbor.getOperationSequencing();
         CopyOnWriteArrayList<Integer> ms = neighbor.getMachineSelection();
 
+        log(String.format("generateSameMachineSwapNeighbor: os.size=%d, ms.size=%d, idx1=%d, globalOpList.size=%d",
+                os.size(), ms.size(), idx1,
+                neighbor.getGlobalOpList() != null ? neighbor.getGlobalOpList().size() : 0));
+
         if (os.size() < 2 || ms.size() < 2) {
+            log("generateSameMachineSwapNeighbor: os或ms太小，返回");
             return neighbor;
         }
 
         // ========== 修复1: op1 必须非 null ==========
         Entry op1 = positionIndex.get(idx1);
         if (op1 == null) {
+            log(String.format("generateSameMachineSwapNeighbor: positionIndex中找不到idx1=%d", idx1));
             return neighbor;
         }
+        log(String.format("generateSameMachineSwapNeighbor: op1=订单%d工序%d, machineOptions.size=%d",
+                op1.getGroupId(), op1.getSequence(), op1.getMachineOptions().size()));
 
         // ========== 修复2: machinePositionIndex 查找必须非 null 且有效 ==========
         String op1Key = op1.getGroupId() + "_" + op1.getSequence();
         Integer maPos1 = machinePositionIndex.get(op1Key);
         if (maPos1 == null || maPos1 < 0 || maPos1 >= ms.size()) {
+            log(String.format("generateSameMachineSwapNeighbor: maPos1=%s无效 (ms.size=%d)", maPos1, ms.size()));
             return neighbor;
         }
 
         int machineSeq1 = ms.get(maPos1);
         if (machineSeq1 < 1 || machineSeq1 > op1.getMachineOptions().size()) {
+            log(String.format("generateSameMachineSwapNeighbor: machineSeq1=%d超出范围[1-%d]",
+                    machineSeq1, op1.getMachineOptions().size()));
             return neighbor;
         }
 
@@ -2444,6 +2481,11 @@ public class VariableNeighborhoodSearch {
         Collections.swap(os, idx1, idx2);
         neighbor.setOperationSequencing(os);
 
+        log(String.format("generateSameMachineSwapNeighbor: swap完成 idx1=%d(订单%d工序%d) <-> idx2=%d(订单%d工序%d), 同机器候选数=%d",
+                idx1, op1.getGroupId(), op1.getSequence(),
+                idx2, positionIndex.get(idx2).getGroupId(), positionIndex.get(idx2).getSequence(),
+                sameMachineOsPositions.size()));
+
         return neighbor;
     }
 
@@ -2512,8 +2554,8 @@ public class VariableNeighborhoodSearch {
 
         Chromosome best =copyChromosome(chromosome);
         decode(decoder, best, machines);
-        Chromosome current = ProductionDeepCopyUtil.deepCopy(best, Chromosome.class);
-        geneticOperations.DelOrder(current);
+        Chromosome current = lightCopy(best);
+
         writeKpi(best);
         // 预定义邻域结构，避免每次循环重复创建
         List<NeighborhoodStructure> neighborhoods = defineNeighborhoods();
@@ -2558,6 +2600,29 @@ public class VariableNeighborhoodSearch {
      * 解码染色体
      */
     private void decode(GeneticDecoder decoder, Chromosome chromosome , List<Machine> machines) {
+        // MS 校验：解码前检查 machineSelection 与 machineOptions 是否匹配
+        List<GlobalOperationInfo> gops = chromosome.getGlobalOpList();
+        List<Integer> msCheck = chromosome.getMachineSelection();
+        if (gops != null && msCheck != null) {
+            int msErrors = 0;
+            StringBuilder sb = new StringBuilder();
+            for (int i = 0; i < Math.min(gops.size(), msCheck.size()); i++) {
+                Entry op = gops.get(i).getOp();
+                int msVal = msCheck.get(i);
+                if (op != null && op.getMachineOptions() != null
+                        && (msVal < 1 || msVal > op.getMachineOptions().size())) {
+                    msErrors++;
+                    if (msErrors <= 3) {
+                        sb.append(String.format(" [idx=%d 订单%d工序%d ms=%d range=1-%d]",
+                                i, op.getGroupId(), op.getSequence(), msVal, op.getMachineOptions().size()));
+                    }
+                }
+            }
+            if (msErrors > 0) {
+                log(String.format("decode-MS校验失败: 共%d处越界 %s", msErrors, sb.toString()));
+            }
+        }
+
         chromosome.setResult(new CopyOnWriteArrayList<>());
 
         // 缓存 Machine 列表（第一次调用时缓存）
@@ -2724,7 +2789,28 @@ public class VariableNeighborhoodSearch {
         }
         return index;
     }
-
+    /**
+     * 轻量拷贝：只复制 generateNeighbor/DelOrder 需要的字段，避免全量 JSON 深拷贝导致 OOM。
+     * result/machines/operatRel 等重型数据共享引用（generateNeighbor 只读，不修改）。
+     */
+    private Chromosome lightCopy(Chromosome source) {
+        Chromosome copy = new Chromosome();
+        copy.setOperationSequencing(new CopyOnWriteArrayList<>(source.getOperationSequencing()));
+        copy.setMachineSelection(new CopyOnWriteArrayList<>(source.getMachineSelection()));
+        copy.setGlobalOpList(new CopyOnWriteArrayList<>(source.getGlobalOpList()));
+        copy.setOrders(new CopyOnWriteArrayList<>(source.getOrders()));
+        copy.setAllOperations(new CopyOnWriteArrayList<>(source.getAllOperations()));
+        copy.setResult(source.getResult());
+        copy.setMachines(source.getMachines());
+        copy.setOperatRel(new CopyOnWriteArrayList<>(source.getOperatRel()));
+        copy.setScenarioID(source.getScenarioID());
+        copy.setBaseTime(source.getBaseTime());
+        copy.setGenerateType(source.getGenerateType());
+        copy.setFitnessLevel(source.getFitnessLevel());
+        copy.setFitness(source.getFitness());
+        geneticOperations.DelOrder(copy);
+        return copy;
+    }
     /**
      * 构建位置->Entry索引
      */
@@ -3675,10 +3761,10 @@ public class VariableNeighborhoodSearch {
             // 随机选择一个不同于当前的机器
             int newSelection;
             if (options.size() == 2) {
-                newSelection = (currentSelection == 1) ? 0 : 1;
+                newSelection = (currentSelection == 1) ? 2 : 1;
             } else {
                 do {
-                    newSelection = rnd.nextInt(options.size());
+                    newSelection = rnd.nextInt(options.size()) + 1;
                 } while (newSelection == currentSelection);
             }
 
@@ -3736,4 +3822,43 @@ public class VariableNeighborhoodSearch {
         log("HybridShake: 混合抖动完成");
         return neighbor;
     }
+
+    /**
+     * CP-SAT 局部重优化邻域
+     * 在当前调度的基础上，释放一部分工序，用 CP-SAT 重优化机器选择
+     */
+    private Chromosome cpSatLocalOptimizeWrapper(Chromosome chromosome) {
+        if (!cpSatEnabled || cpSatDecoder == null || cpSatMachines == null) {
+            return null;
+        }
+
+        try {
+            // CP-SAT 需要解码后的时间信息，先做一次解码
+            cpSatDecoder.serialDecode(chromosome);
+
+            // 只做一次快速重优化（释放约 10% 的工序）
+            CpSatLnsNeighborhood lns = new CpSatLnsNeighborhood(
+                    cachedAllOperations, cpSatMachines,orders,materials,_entryRel, fitnessCalculator);
+            int releaseCount = Math.max(30, (int)(cachedAllOperations.size() * 0.10));
+            int timeLimitSec = Math.min(8, 12000 / Math.max(100, cachedAllOperations.size()));
+
+            Chromosome neighbor = lns.optimizeNeighborhood(chromosome, releaseCount, timeLimitSec);
+            if (neighbor != null) {
+                log("CpSatLocalOptimize: 生成邻居成功");
+            }
+            return neighbor;
+
+        } catch (Exception e) {
+            log("CpSatLocalOptimize: 异常 - " + e.getMessage());
+            return null;
+        }
+    }
+    public static void log(String message) {
+        log(message, true);
+    }
+
+    public static void log(String message, boolean enableLogging) {
+       FileHelper.log(message, enableLogging);
+    }
+
 }
\ No newline at end of file

src/test/java/com/aps/demo/PlanResultServiceTest.java

@@ -43,8 +43,8 @@ public class PlanResultServiceTest {
 
      //     planResultService.execute2("64E64F6B68094AF38CEDC418630C3CC2");//2000
            //  planResultService.execute2("E1448B3C9C8743DEAB39708F2CFE348A");//倒排bomces
-     //  planResultService.execute2("197083D0D26A449EB179AC103C753FD3");
-        planResultService.execute2("F8F147BD627C47B1A190399DD7A697F6");
+      planResultService.execute2("85DA28EC5F4643449E65A51253D5F127");
+      //  planResultService.execute2("F8F147BD627C47B1A190399DD7A697F6");
 
       //  planResultService.execute2("9FEDFD92BB6A4675BF9B1CC64505D1AB");