系统概述

本平台面向城市环卫管理场景，整合移动端应用、物联网感知与空间数据分析能力，构建覆盖垃圾收运全流程的数字化监管体系。核心解决传统模式下车辆调度粗放、设施状态感知滞后、作业质量难追溯等痛点。

核心数据模型设计

采用空间数据库扩展方案，关键实体关系如下：

erDiagram
    TRUCK ||--o{ MISSION : "执行"
    TRUCK {
        string PlateNumber PK
        enum EnergyType
        decimal LoadCapacity
    }
    MISSION {
        uuid MissionID PK
        geometry Trajectory
        timestamp DispatchTime
        timestamp CompleteTime
    }
    WASTEBOX ||--o{ MISSION : "关联"
    WASTEBOX ||--o{ SENSOR_LOG : "产生"
    WASTEBOX {
        int BoxNo PK
        geography Position
        decimal Volume
        enum BoxCategory
    }
    SENSOR_LOG {
        bigint LogID PK
        int BoxNo FK
        tinyint OccupancyRate
        decimal Celsius
        timestamp CaptureAt
    }

数据存储层实现

利用PostGIS扩展支持地理空间运算，表结构设计兼顾查询性能与扩展性：

-- 作业任务轨迹表（含空间索引）
CREATE TABLE Mission (
    MissionID UUID DEFAULT gen_random_uuid() PRIMARY KEY,
    TruckID VARCHAR(10) NOT NULL,
    RoutePath GEOGRAPHY(LINESTRING, 4326),
    PlannedDistance DECIMAL(8,2),
    ActualDistance DECIMAL(8,2),
    CreatedAt TIMESTAMPTZ DEFAULT NOW(),
    SPATIAL INDEX idx_route_path (RoutePath)
);

-- 智能垃圾桶实时状态表（分区存储按周轮转）
CREATE TABLE SensorLog (
    LogID BIGSERIAL,
    BoxNo INT NOT NULL,
    FillPct TINYINT CHECK (FillPct BETWEEN 0 AND 100),
    BatteryVolt DECIMAL(4,2),
    ReportedAt TIMESTAMPTZ DEFAULT NOW(),
    PRIMARY KEY (LogID, ReportedAt)
) PARTITION BY RANGE (ReportedAt);

-- 创建近期分区
CREATE TABLE SensorLog_2024w35 PARTITION OF SensorLog
    FOR VALUES FROM ('2024-08-26') TO ('2024-09-02');

路径优化核心算法

针对垃圾收运场景改进传统遗传算法，引入载重约束与时间窗惩罚：

public class CollectionScheduler {
    private static final int ELITE_COUNT = 5;
    private static final double CROSSOVER_RATE = 0.85;
    
    public RoutePlan solveVRP(List<WasteNode> pickups, Depot depot) {
        Population current = initDiversePopulation(pickups, depot);
        
        for (int iter = 0; iter < maxIterations; iter++) {
            // 非支配排序处理多目标（距离+碳排+时效）
            List<RoutePlan> paretoFront = fastNonDominatedSort(current);
            
            Population offspring = new Population();
            // 保留精英个体
            offspring.addAll(paretoFront.subList(0, ELITE_COUNT));
            
            while (offspring.size() < populationSize) {
                RoutePlan p1 = tournamentSelect(current, 3);
                RoutePlan p2 = tournamentSelect(current, 3);
                
                if (ThreadLocalRandom.current().nextDouble() < CROSSOVER_RATE) {
                    RoutePlan[] children = orderCrossover(p1, p2);
                    offspring.add(mutateWith2Opt(children[0]));
                    offspring.add(mutateWith2Opt(children[1]));
                }
            }
            current = offspring;
        }
        return extractBestBalance(current);
    }
    
    private double calculateCarbonCost(RoutePlan plan) {
        double total = 0.0;
        for (Leg segment : plan.getSegments()) {
            double factor = segment.getTruck().getEmissionFactor();
            total += segment.getMeter() / 1000.0 * factor;
        }
        return total;
    }
}

Android端地图集成

采用高德地图SDK实现作业可视化，关键封装如下：

public class DispatchMapLayer {
    private MapView mapView;
    private HeatmapOverlay heatmap;
    
    public void renderCollectionStatus(List<BinStatus> realtimeData) {
        List<WeightedLatLng> heatPoints = realtimeData.stream()
            .filter(b -> b.getFillPct() > 60)
            .map(b -> new WeightedLatLng(
                new LatLng(b.getLat(), b.getLng()),
                b.getFillPct() / 100.0
            ))
            .collect(Collectors.toList());
        
        HeatmapTileProvider provider = new HeatmapTileProvider.Builder()
            .weightedData(heatPoints)
            .radius(40)
            .gradient(urgencyGradient)
            .build();
            
        heatmap.setTileProvider(provider);
        mapView.invalidate();
    }
    
    public void animateTruckMovement(String truckId, List<LatLng>> trace) {
        SmoothMoveMarker mover = new SmoothMoveMarker(mapView.getMap());
        mover.setDescriptor(BitmapDescriptorFactory
            .fromResource(R.drawable.ic_truck_moving));
        mover.setPoints(trace);
        mover.setTotalDuration(30);
        mover.startSmoothMove();
    }
}

物联网硬件对接

垃圾桶终端采用LoRaWAN组网，数据上行协议设计：

字节位	含义	说明
0-1	帧头	固定0xAA55
2	设备类型	0x01=超声波,0x02=红外
3-4	设备编号	BCD编码
5	填充高度	0-255对应0-100%
6	温度	偏移-40℃,范围-40~85
7-8	电池电压	单位mV/10

异常检测与预警

基于滑动窗口统计识别异常模式：

@Service
public class AnomalyDetector {
    
    public Optional<Alert> analyze(BinTelemetry telemetry) {
        DescriptiveStatistics stats = ringBuffer
            .get(telemetry.getBoxNo());
        
        if (stats.getN() < MIN_SAMPLES) {
            return Optional.empty();
        }
        
        double zScore = (telemetry.getTemp() - stats.getMean()) 
            / stats.getStandardDeviation();
            
        if (zScore > 3.5) {
            return Optional.of(Alert.builder()
                .level(CRITICAL)
                .type(TEMPERATURE_SPIKE)
                .suggestedAction("调度最近车辆现场核查")
                .build());
        }
        
        // 填充速率异常（疑似非法倾倒）
        double fillVelocity = estimateFillVelocity(telemetry);
        if (fillVelocity > THRESHOLD_RAPID_FILL) {
            return Optional.of(Alert.builder()
                .level(WARNING)
                .type(RAPID_FILL_DETECTED)
                .build());
        }
        return Optional.empty();
    }
}

技术选型总结

• 移动端: Android原生 + Jetpack Compose UI层
• 地图引擎: 高德地图SDK（路径规划+轨迹纠偏）
• 后端服务: Spring Boot + WebFlux响应式处理
• 空间数据库: PostgreSQL 15 + PostGIS 3.3
• 消息队列: RabbitMQ处理设备上行数据
• 缓存层: Redis Geo数据结构支持附近车辆检索