diff --git a/viewsh-module-ops/viewsh-module-environment-biz/src/main/java/com/viewsh/module/ops/environment/integration/listener/BadgeDeviceStatusEventListener.java b/viewsh-module-ops/viewsh-module-environment-biz/src/main/java/com/viewsh/module/ops/environment/integration/listener/BadgeDeviceStatusEventListener.java
index a9cc44c5..cd0b7857 100644
--- a/viewsh-module-ops/viewsh-module-environment-biz/src/main/java/com/viewsh/module/ops/environment/integration/listener/BadgeDeviceStatusEventListener.java
+++ b/viewsh-module-ops/viewsh-module-environment-biz/src/main/java/com/viewsh/module/ops/environment/integration/listener/BadgeDeviceStatusEventListener.java
@@ -15,9 +15,6 @@ import jakarta.annotation.Resource;
 import lombok.extern.slf4j.Slf4j;
 import org.springframework.context.event.EventListener;
 import org.springframework.stereotype.Component;
-import org.springframework.transaction.PlatformTransactionManager;
-import org.springframework.transaction.TransactionDefinition;
-import org.springframework.transaction.support.TransactionTemplate;
 
 /**
  * 工牌设备状态事件监听器
@@ -87,9 +84,6 @@ public class BadgeDeviceStatusEventListener {
     @Resource
     private OrderLifecycleManager orderLifecycleManager;
 
-    @Resource
-    private PlatformTransactionManager transactionManager;
-
     /**
      * 监听工单状态变更事件，同步更新设备工单关联
      * <p>
@@ -180,40 +174,27 @@ public class BadgeDeviceStatusEventListener {
 
     /**
      * 处理工单推送状态（首次设置工单关联）
+     * <p>
+     * 若 Redis 里检测到旧 orderId（正常业务不应出现），仅打 ERROR 告警并清理 Redis 关联。
+     * 此前版本会在此处"自动取消旧工单"，但那是对"数据已错乱"场景的暴力兜底：
+     * <ul>
+     *   <li>取消使用 REQUIRES_NEW 独立事务且吞异常，失败时新单照常落地，旧单残留，形成越清越多</li>
+     *   <li>真正的防线应在 DispatchEngine.autoDispatchNext 入口做设备空闲校验</li>
+     * </ul>
+     * 现改为被动告警，暴露问题等待定位，避免误杀保洁员正在执行的任务。
      */
     private void handleDispatched(Long deviceId, Long orderId, OpsOrderDO order) {
-        // 检查并清理旧工单（防止工单切换时状态残留）
         BadgeDeviceStatusDTO deviceStatus = badgeDeviceStatusService.getBadgeStatus(deviceId);
         if (deviceStatus != null && deviceStatus.getCurrentOpsOrderId() != null) {
             Long oldOrderId = deviceStatus.getCurrentOpsOrderId();
             if (!oldOrderId.equals(orderId)) {
-                log.warn("[BadgeDeviceStatusEventListener] 派发新工单时检测到旧工单残留: " +
-                        "deviceId={}, oldOrderId={}, newOrderId={}", deviceId, oldOrderId, orderId);
-
-                // 检查旧工单是否仍在进行中，如果是则先取消
                 OpsOrderDO oldOrder = opsOrderMapper.selectById(oldOrderId);
-                if (oldOrder != null) {
-                    WorkOrderStatusEnum oldStatus = WorkOrderStatusEnum.fromStatus(oldOrder.getStatus());
-                    if (oldStatus == WorkOrderStatusEnum.DISPATCHED
-                            || oldStatus == WorkOrderStatusEnum.CONFIRMED
-                            || oldStatus == WorkOrderStatusEnum.ARRIVED) {
-                        // 旧工单仍在进行，先取消
-                        // 使用 REQUIRES_NEW 独立事务，避免内层异常标记外层事务 rollback-only
-                        log.warn("[BadgeDeviceStatusEventListener] 取消残留的旧工单: oldOrderId={}", oldOrderId);
-                        try {
-                            TransactionTemplate txTemplate = new TransactionTemplate(transactionManager);
-                            txTemplate.setPropagationBehavior(TransactionDefinition.PROPAGATION_REQUIRES_NEW);
-                            txTemplate.executeWithoutResult(status -> {
-                                orderLifecycleManager.cancelOrder(oldOrderId, deviceId,
-                                        OperatorTypeEnum.SYSTEM, "新工单派发，自动取消旧工单");
-                            });
-                        } catch (Exception e) {
-                            log.error("[BadgeDeviceStatusEventListener] 取消旧工单失败: oldOrderId={}", oldOrderId, e);
-                        }
-                    }
-                }
+                String oldStatus = oldOrder != null ? oldOrder.getStatus() : "NOT_FOUND";
+                log.error("[BadgeDeviceStatusEventListener] 派发新工单时检测到旧工单残留（数据可能已错乱，需人工核查）: " +
+                                "deviceId={}, oldOrderId={}, oldStatus={}, newOrderId={}",
+                        deviceId, oldOrderId, oldStatus, orderId);
 
-                // 确保设备状态清理（无论旧工单是否取消成功）
+                // 清理 Redis 中对旧工单的关联（纯 Redis 操作，不触达状态机）
                 badgeDeviceStatusService.clearCurrentOrder(deviceId);
             }
         }
diff --git a/viewsh-module-ops/viewsh-module-ops-biz/src/main/java/com/viewsh/module/ops/core/dispatch/DispatchEngineImpl.java b/viewsh-module-ops/viewsh-module-ops-biz/src/main/java/com/viewsh/module/ops/core/dispatch/DispatchEngineImpl.java
index 48ab3741..1cf884d8 100644
--- a/viewsh-module-ops/viewsh-module-ops-biz/src/main/java/com/viewsh/module/ops/core/dispatch/DispatchEngineImpl.java
+++ b/viewsh-module-ops/viewsh-module-ops-biz/src/main/java/com/viewsh/module/ops/core/dispatch/DispatchEngineImpl.java
@@ -178,6 +178,22 @@ public class DispatchEngineImpl implements DispatchEngine {
     public DispatchResult autoDispatchNext(Long completedOrderId, Long assigneeId) {
         log.info("任务完成后自动派发下一单: completedOrderId={}, assigneeId={}", completedOrderId, assigneeId);
 
+        if (assigneeId == null) {
+            log.warn("autoDispatchNext 缺少执行人，跳过派发: completedOrderId={}", completedOrderId);
+            return DispatchResult.success("缺少执行人，跳过派发", null);
+        }
+
+        // 空闲校验：若执行人仍挂着其他活跃工单（DISPATCHED/CONFIRMED/ARRIVED/PAUSED），
+        // 说明设备尚未真正空闲，不应再派发新任务——否则会触发"同一设备并行多单"的状态错乱，
+        // 典型场景是管理员手动取消一个僵尸 DISPATCHED 单时，handleCancelled 会调到这里。
+        List<OpsOrderDO> activeOrders = orderMapper.selectActiveByAssignee(assigneeId, completedOrderId);
+        if (!activeOrders.isEmpty()) {
+            OpsOrderDO head = activeOrders.get(0);
+            log.info("执行人仍有活跃工单，跳过自动派发: assigneeId={}, completedOrderId={}, activeCount={}, sampleOrderId={}, sampleStatus={}",
+                    assigneeId, completedOrderId, activeOrders.size(), head.getId(), head.getStatus());
+            return DispatchResult.success("执行人非空闲，跳过派发", assigneeId);
+        }
+
         Long fallbackAreaId = null;
         OpsOrderDO completedOrder = orderMapper.selectById(completedOrderId);
         if (completedOrder != null) {
diff --git a/viewsh-module-ops/viewsh-module-ops-biz/src/main/java/com/viewsh/module/ops/dal/mysql/workorder/OpsOrderMapper.java b/viewsh-module-ops/viewsh-module-ops-biz/src/main/java/com/viewsh/module/ops/dal/mysql/workorder/OpsOrderMapper.java
index f1454b99..6df51062 100644
--- a/viewsh-module-ops/viewsh-module-ops-biz/src/main/java/com/viewsh/module/ops/dal/mysql/workorder/OpsOrderMapper.java
+++ b/viewsh-module-ops/viewsh-module-ops-biz/src/main/java/com/viewsh/module/ops/dal/mysql/workorder/OpsOrderMapper.java
@@ -92,6 +92,28 @@ public interface OpsOrderMapper extends BaseMapperX<OpsOrderDO> {
                 .last("LIMIT 1"));
     }
 
+    /**
+     * 查询执行人名下尚未结束的工单（DISPATCHED/CONFIRMED/ARRIVED/PAUSED）
+     * <p>
+     * 用于 autoDispatchNext 等调度入口的空闲校验：若该执行人仍挂着活跃工单，
+     * 则不应再派发新任务，避免"越清越多"的级联派发。
+     *
+     * @param assigneeId     执行人ID（工牌设备ID）
+     * @param excludeOrderId 需要排除的工单ID（通常是刚完成/取消触发本次调度的工单），可传 null
+     * @return 活跃工单列表，按创建时间升序
+     */
+    default List<OpsOrderDO> selectActiveByAssignee(Long assigneeId, Long excludeOrderId) {
+        return selectList(new LambdaQueryWrapperX<OpsOrderDO>()
+                .eq(OpsOrderDO::getAssigneeId, assigneeId)
+                .in(OpsOrderDO::getStatus,
+                        WorkOrderStatusEnum.DISPATCHED.getStatus(),
+                        WorkOrderStatusEnum.CONFIRMED.getStatus(),
+                        WorkOrderStatusEnum.ARRIVED.getStatus(),
+                        WorkOrderStatusEnum.PAUSED.getStatus())
+                .ne(excludeOrderId != null, OpsOrderDO::getId, excludeOrderId)
+                .orderByAsc(OpsOrderDO::getCreateTime));
+    }
+
     // ==================== 统计聚合查询 ====================
 
     /**