DolphinDB生产指标监控：KPI实时计算

七夜zippoe2026-07-06 11:01

目录

    • 摘要
    • 一、生产指标概述
      • [1.1 KPI指标体系](#1.1 KPI指标体系)
      • [1.2 核心KPI指标](#1.2 核心KPI指标)
      • [1.3 指标监控目标](#1.3 指标监控目标)
    • 二、KPI定义与计算
      • [2.1 KPI定义表](#2.1 KPI定义表)
      • [2.2 产量计算](#2.2 产量计算)
      • [2.3 良品率计算](#2.3 良品率计算)
      • [2.4 OEE计算](#2.4 OEE计算)
    • 三、实时统计
      • [3.1 时间窗口统计](#3.1 时间窗口统计)
      • [3.2 滑动窗口统计](#3.2 滑动窗口统计)
      • [3.3 实时聚合引擎](#3.3 实时聚合引擎)
    • 四、趋势分析
      • [4.1 趋势数据](#4.1 趋势数据)
      • [4.2 同比环比](#4.2 同比环比)
      • [4.3 预测分析](#4.3 预测分析)
    • 五、指标告警
      • [5.1 告警规则](#5.1 告警规则)
      • [5.2 告警推送](#5.2 告警推送)
    • 六、可视化展示
      • [6.1 KPI看板数据](#6.1 KPI看板数据)
      • [6.2 KPI排名](#6.2 KPI排名)
    • 七、实战案例
      • [7.1 完整KPI监控系统](#7.1 完整KPI监控系统)
    • 八、总结
    • 参考资料

摘要

本文深入讲解DolphinDB生产指标监控技术。从KPI定义到指标计算，从实时统计到趋势分析，从指标告警到可视化展示，全面介绍生产指标监控的核心方法。通过丰富的代码示例，帮助读者掌握KPI实时计算的核心技能。

一、生产指标概述

1.1 KPI指标体系

#mermaid-svg-iK9yB9hrM0x8HJUd{font-family:"trebuchet ms",verdana,arial,sans-serif;font-size:16px;fill:#333;}@keyframes edge-animation-frame{from{stroke-dashoffset:0;}}@keyframes dash{to{stroke-dashoffset:0;}}#mermaid-svg-iK9yB9hrM0x8HJUd .edge-animation-slow{stroke-dasharray:9,5!important;stroke-dashoffset:900;animation:dash 50s linear infinite;stroke-linecap:round;}#mermaid-svg-iK9yB9hrM0x8HJUd .edge-animation-fast{stroke-dasharray:9,5!important;stroke-dashoffset:900;animation:dash 20s linear infinite;stroke-linecap:round;}#mermaid-svg-iK9yB9hrM0x8HJUd .error-icon{fill:#552222;}#mermaid-svg-iK9yB9hrM0x8HJUd .error-text{fill:#552222;stroke:#552222;}#mermaid-svg-iK9yB9hrM0x8HJUd .edge-thickness-normal{stroke-width:1px;}#mermaid-svg-iK9yB9hrM0x8HJUd .edge-thickness-thick{stroke-width:3.5px;}#mermaid-svg-iK9yB9hrM0x8HJUd .edge-pattern-solid{stroke-dasharray:0;}#mermaid-svg-iK9yB9hrM0x8HJUd .edge-thickness-invisible{stroke-width:0;fill:none;}#mermaid-svg-iK9yB9hrM0x8HJUd .edge-pattern-dashed{stroke-dasharray:3;}#mermaid-svg-iK9yB9hrM0x8HJUd .edge-pattern-dotted{stroke-dasharray:2;}#mermaid-svg-iK9yB9hrM0x8HJUd .marker{fill:#333333;stroke:#333333;}#mermaid-svg-iK9yB9hrM0x8HJUd .marker.cross{stroke:#333333;}#mermaid-svg-iK9yB9hrM0x8HJUd svg{font-family:"trebuchet ms",verdana,arial,sans-serif;font-size:16px;}#mermaid-svg-iK9yB9hrM0x8HJUd p{margin:0;}#mermaid-svg-iK9yB9hrM0x8HJUd .label{font-family:"trebuchet ms",verdana,arial,sans-serif;color:#333;}#mermaid-svg-iK9yB9hrM0x8HJUd .cluster-label text{fill:#333;}#mermaid-svg-iK9yB9hrM0x8HJUd .cluster-label span{color:#333;}#mermaid-svg-iK9yB9hrM0x8HJUd .cluster-label span p{background-color:transparent;}#mermaid-svg-iK9yB9hrM0x8HJUd .label text,#mermaid-svg-iK9yB9hrM0x8HJUd span{fill:#333;color:#333;}#mermaid-svg-iK9yB9hrM0x8HJUd .node rect,#mermaid-svg-iK9yB9hrM0x8HJUd .node circle,#mermaid-svg-iK9yB9hrM0x8HJUd .node ellipse,#mermaid-svg-iK9yB9hrM0x8HJUd .node polygon,#mermaid-svg-iK9yB9hrM0x8HJUd .node path{fill:#ECECFF;stroke:#9370DB;stroke-width:1px;}#mermaid-svg-iK9yB9hrM0x8HJUd .rough-node .label text,#mermaid-svg-iK9yB9hrM0x8HJUd .node .label text,#mermaid-svg-iK9yB9hrM0x8HJUd .image-shape .label,#mermaid-svg-iK9yB9hrM0x8HJUd .icon-shape .label{text-anchor:middle;}#mermaid-svg-iK9yB9hrM0x8HJUd .node .katex path{fill:#000;stroke:#000;stroke-width:1px;}#mermaid-svg-iK9yB9hrM0x8HJUd .rough-node .label,#mermaid-svg-iK9yB9hrM0x8HJUd .node .label,#mermaid-svg-iK9yB9hrM0x8HJUd .image-shape .label,#mermaid-svg-iK9yB9hrM0x8HJUd .icon-shape .label{text-align:center;}#mermaid-svg-iK9yB9hrM0x8HJUd .node.clickable{cursor:pointer;}#mermaid-svg-iK9yB9hrM0x8HJUd .root .anchor path{fill:#333333!important;stroke-width:0;stroke:#333333;}#mermaid-svg-iK9yB9hrM0x8HJUd .arrowheadPath{fill:#333333;}#mermaid-svg-iK9yB9hrM0x8HJUd .edgePath .path{stroke:#333333;stroke-width:2.0px;}#mermaid-svg-iK9yB9hrM0x8HJUd .flowchart-link{stroke:#333333;fill:none;}#mermaid-svg-iK9yB9hrM0x8HJUd .edgeLabel{background-color:rgba(232,232,232, 0.8);text-align:center;}#mermaid-svg-iK9yB9hrM0x8HJUd .edgeLabel p{background-color:rgba(232,232,232, 0.8);}#mermaid-svg-iK9yB9hrM0x8HJUd .edgeLabel rect{opacity:0.5;background-color:rgba(232,232,232, 0.8);fill:rgba(232,232,232, 0.8);}#mermaid-svg-iK9yB9hrM0x8HJUd .labelBkg{background-color:rgba(232, 232, 232, 0.5);}#mermaid-svg-iK9yB9hrM0x8HJUd .cluster rect{fill:#ffffde;stroke:#aaaa33;stroke-width:1px;}#mermaid-svg-iK9yB9hrM0x8HJUd .cluster text{fill:#333;}#mermaid-svg-iK9yB9hrM0x8HJUd .cluster span{color:#333;}#mermaid-svg-iK9yB9hrM0x8HJUd div.mermaidTooltip{position:absolute;text-align:center;max-width:200px;padding:2px;font-family:"trebuchet ms",verdana,arial,sans-serif;font-size:12px;background:hsl(80, 100%, 96.2745098039%);border:1px solid #aaaa33;border-radius:2px;pointer-events:none;z-index:100;}#mermaid-svg-iK9yB9hrM0x8HJUd .flowchartTitleText{text-anchor:middle;font-size:18px;fill:#333;}#mermaid-svg-iK9yB9hrM0x8HJUd rect.text{fill:none;stroke-width:0;}#mermaid-svg-iK9yB9hrM0x8HJUd .icon-shape,#mermaid-svg-iK9yB9hrM0x8HJUd .image-shape{background-color:rgba(232,232,232, 0.8);text-align:center;}#mermaid-svg-iK9yB9hrM0x8HJUd .icon-shape p,#mermaid-svg-iK9yB9hrM0x8HJUd .image-shape p{background-color:rgba(232,232,232, 0.8);padding:2px;}#mermaid-svg-iK9yB9hrM0x8HJUd .icon-shape .label rect,#mermaid-svg-iK9yB9hrM0x8HJUd .image-shape .label rect{opacity:0.5;background-color:rgba(232,232,232, 0.8);fill:rgba(232,232,232, 0.8);}#mermaid-svg-iK9yB9hrM0x8HJUd .label-icon{display:inline-block;height:1em;overflow:visible;vertical-align:-0.125em;}#mermaid-svg-iK9yB9hrM0x8HJUd .node .label-icon path{fill:currentColor;stroke:revert;stroke-width:revert;}#mermaid-svg-iK9yB9hrM0x8HJUd :root{--mermaid-font-family:"trebuchet ms",verdana,arial,sans-serif;} 生产KPI体系
产量指标
生产KPI
质量指标
效率指标
能耗指标

1.2 核心KPI指标

指标 计算公式 说明
产量 合格品数量 生产总量
良品率 合格品/总产量 × 100% 质量指标
OEE 可用率×性能率×良品率 综合效率
能耗 总能耗/产量 单位能耗

1.3 指标监控目标

目标 说明
实时性 指标实时计算
准确性 数据准确可靠
可视化 指标直观展示
告警性 异常及时告警

二、KPI定义与计算

2.1 KPI定义表

python 复制代码 
// KPI定义表
kpiDefinitions = table(
    ["production", "quality_rate", "oee", "energy_consumption"] as kpi_id,
    ["产量", "良品率", "OEE", "单位能耗"] as kpi_name,
    ["count(product_id)", "sum(qualified)/count(*)*100", "availability*performance*quality", "sum(energy)/count(product_id)"] as formula,
    ["件", "%", "%", "kWh/件"] as unit,
    [1000, 95, 85, 10] as target
)

2.2 产量计算

python 复制代码 
// 产量数据表
share streamTable(100000:0, 
    `product_id`device_id`timestamp`quantity`qualified,
    [SYMBOL, SYMBOL, TIMESTAMP, INT, BOOL]) as production_stream

// 实时产量统计
share table(1:0, 
    `time_window`device_id`production_count`qualified_count,
    [TIMESTAMP, SYMBOL, LONG, LONG]) as production_agg

// 产量聚合引擎
productionEngine = createTimeSeriesEngine("production_engine", 60000,
    <[count(*) as production_count,
      sum(qualified) as qualified_count]>,
    production_agg, `timestamp, `device_id)

subscribeTable(, "production_stream", "production_agg", -1, productionEngine, true)

2.3 良品率计算

python 复制代码 
// 良品率计算
def calculateQualityRate(deviceId, startTime, endTime) {
    data = select count(*) as total,
                  sum(qualified) as qualified
           from production_stream
           where device_id = deviceId
           and timestamp between startTime and endTime
    
    if (data.total[0] == 0) {
        return 0.0
    }
    
    return data.qualified[0] * 100.0 / data.total[0]
}

// 批量良品率
def batchQualityRate(startTime, endTime) {
    return select device_id,
                  sum(qualified) * 100.0 / count(*) as quality_rate
           from production_stream
           where timestamp between startTime and endTime
           group by device_id
}

2.4 OEE计算

python 复制代码 
// OEE数据表
share table(1:0, 
    `device_id`timestamp`availability`performance`quality,
    [SYMBOL, TIMESTAMP, DOUBLE, DOUBLE, DOUBLE]) as oee_data

// OEE计算
def calculateOEE(deviceId, startTime, endTime) {
    // 可用率 = 实际运行时间 / 计划运行时间
    availability = getAvailability(deviceId, startTime, endTime)
    
    // 性能率 = 实际产量 / 理论产量
    performance = getPerformance(deviceId, startTime, endTime)
    
    // 质量率 = 合格品 / 总产量
    quality = getQuality(deviceId, startTime, endTime)
    
    // OEE = 可用率 × 性能率 × 质量率
    oee = availability * performance * quality / 10000
    
    insert into oee_data values (deviceId, now(), availability, performance, quality)
    
    return oee
}

def getAvailability(deviceId, startTime, endTime) {
    runTime = getRunTime(deviceId, startTime, endTime)
    plannedTime = endTime - startTime
    return runTime * 100.0 / plannedTime
}

def getPerformance(deviceId, startTime, endTime) {
    actualOutput = getActualOutput(deviceId, startTime, endTime)
    theoreticalOutput = getTheoreticalOutput(deviceId, startTime, endTime)
    return actualOutput * 100.0 / theoreticalOutput
}

def getQuality(deviceId, startTime, endTime) {
    return calculateQualityRate(deviceId, startTime, endTime)
}

三、实时统计

3.1 时间窗口统计

python 复制代码 
// 多时间窗口统计
def multiWindowStats(deviceId) {
    now = now()
    
    return dict(STRING, ANY, [
        ["hourly", calculateHourlyStats(deviceId, now - 3600000, now)],
        ["daily", calculateDailyStats(deviceId, now - 86400000, now)],
        ["weekly", calculateWeeklyStats(deviceId, now - 604800000, now)]
    ])
}

def calculateHourlyStats(deviceId, startTime, endTime) {
    return select device_id,
                  count(*) as production,
                  sum(qualified) * 100.0 / count(*) as quality_rate
           from production_stream
           where device_id = deviceId
           and timestamp between startTime and endTime
           group by device_id
}

3.2 滑动窗口统计

python 复制代码 
// 滑动窗口统计
def slidingWindowStats(deviceId, windowSize = 3600000) {
    now = now()
    startTime = now - windowSize
    
    return select device_id,
                  count(*) as production,
                  avg(production) as avg_production,
                  sum(qualified) * 100.0 / count(*) as quality_rate
           from production_stream
           where device_id = deviceId
           and timestamp between startTime and now
           group by device_id
}

3.3 实时聚合引擎

python 复制代码 
// 实时KPI聚合
share table(1:0, 
    `time_window`kpi_id`kpi_value,
    [TIMESTAMP, STRING, DOUBLE]) as kpi_realtime

// KPI计算引擎
def kpiCalculateEngine() {
    while (true) {
        now = now()
        
        // 计算各KPI
        production = getProductionCount(now - 60000, now)
        qualityRate = getQualityRate(now - 60000, now)
        oee = getOEE(now - 60000, now)
        
        // 写入实时表
        insert into kpi_realtime values (now, "production", production)
        insert into kpi_realtime values (now, "quality_rate", qualityRate)
        insert into kpi_realtime values (now, "oee", oee)
        
        sleep(60000)
    }
}

submitJob("kpi_engine", "KPI计算引擎", kpiCalculateEngine)

四、趋势分析

4.1 趋势数据

python 复制代码 
// KPI趋势表
share table(1:0, 
    `time_window`kpi_id`kpi_value`trend,
    [TIMESTAMP, STRING, DOUBLE, STRING]) as kpi_trend

// 趋势计算
def calculateTrend(kpiId, periods = 7) {
    data = select * from kpi_realtime
           where kpi_id = kpiId
           order by time_window
           limit periods
    
    if (data.rows() < 2) {
        return "stable"
    }
    
    first = data.kpi_value[0]
    last = data.kpi_value[-1]
    
    if (last > first * 1.05) {
        return "up"
    } else if (last < first * 0.95) {
        return "down"
    } else {
        return "stable"
    }
}

4.2 同比环比

python 复制代码 
// 同比环比计算
def calculateYoY(kpiId) {
    now = now()
    current = getKpiValue(kpiId, now - 86400000, now)
    lastYear = getKpiValue(kpiId, now - 365*86400000, now - 364*86400000)
    
    if (lastYear == 0) {
        return 0.0
    }
    
    return (current - lastYear) * 100.0 / lastYear
}

def calculateMoM(kpiId) {
    now = now()
    current = getKpiValue(kpiId, now - 86400000, now)
    lastMonth = getKpiValue(kpiId, now - 60*86400000, now - 59*86400000)
    
    if (lastMonth == 0) {
        return 0.0
    }
    
    return (current - lastMonth) * 100.0 / lastMonth
}

4.3 预测分析

python 复制代码 
// 简单移动平均预测
def predictKpi(kpiId, periods = 7) {
    data = select * from kpi_realtime
           where kpi_id = kpiId
           order by time_window desc
           limit periods
    
    if (data.rows() == 0) {
        return 0.0
    }
    
    return avg(data.kpi_value)
}

五、指标告警

5.1 告警规则

python 复制代码 
// KPI告警规则
kpiAlertRules = table(
    ["production_low", "quality_low", "oee_low"] as rule_name,
    ["production", "quality_rate", "oee"] as kpi_id,
    [500, 90, 80] as threshold,
    ["<", "<", "<"] as operator,
    [2, 1, 1] as alert_level
)

// 检查KPI告警
def checkKpiAlerts() {
    alerts = array(STRING, 0)
    
    for (rule in kpiAlertRules) {
        value = getLatestKpi(rule.kpi_id)
        
        if (eval(string(value) + rule.operator + string(rule.threshold))) {
            alerts.append!(rule.rule_name + ": " + string(value))
        }
    }
    
    return alerts
}

5.2 告警推送

python 复制代码 
// KPI告警推送
def pushKpiAlert(kpiId, value, threshold) {
    message = "KPI告警: " + kpiId + " 当前值 " + string(value) + " 低于阈值 " + string(threshold)
    
    // 记录告警
    insert into alert_log values (now(), "system", kpiId, 2, value, message)
    
    // 推送
    print(message)
}

六、可视化展示

6.1 KPI看板数据

python 复制代码 
// KPI看板
def getKpiDashboard() {
    now = now()
    
    return dict(STRING, ANY, [
        ["production", getLatestKpi("production")],
        ["quality_rate", getLatestKpi("quality_rate")],
        ["oee", getLatestKpi("oee")],
        ["production_trend", calculateTrend("production")],
        ["quality_trend", calculateTrend("quality_rate")],
        ["oee_trend", calculateTrend("oee")]
    ])
}

def getLatestKpi(kpiId) {
    return exec last(kpi_value) from kpi_realtime where kpi_id = kpiId
}

6.2 KPI排名

python 复制代码 
// 设备KPI排名
def getKpiRanking(kpiId, limit = 10) {
    return select device_id,
                  kpi_value,
                  rank() over order by kpi_value desc as rank
           from device_kpi
           where kpi_id = kpiId
           limit limit
}

七、实战案例

7.1 完整KPI监控系统

python 复制代码 
// ========== 生产KPI监控系统 ==========

// 1. 创建数据表
share streamTable(100000:0, 
    `product_id`device_id`timestamp`quantity`qualified,
    [SYMBOL, SYMBOL, TIMESTAMP, INT, BOOL]) as production_stream

share table(1:0, 
    `time_window`kpi_id`kpi_value,
    [TIMESTAMP, STRING, DOUBLE]) as kpi_realtime

share table(1:0, 
    `device_id`timestamp`availability`performance`quality`oee,
    [SYMBOL, TIMESTAMP, DOUBLE, DOUBLE, DOUBLE, DOUBLE]) as device_oee

// 2. 产量聚合引擎
share table(1:0, 
    `time_window`device_id`production_count`qualified_count,
    [TIMESTAMP, SYMBOL, LONG, LONG]) as production_agg

productionEngine = createTimeSeriesEngine("production_engine", 60000,
    <[count(*) as production_count,
      sum(qualified) as qualified_count]>,
    production_agg, `timestamp, `device_id)

subscribeTable(, "production_stream", "production_agg", -1, productionEngine, true)

// 3. KPI计算任务
def kpiCalculateTask() {
    while (true) {
        now = now()
        
        // 计算产量
        production = exec sum(production_count) from production_agg
                      where time_window > now - 60000
        
        // 计算良品率
        quality = exec sum(qualified_count) * 100.0 / sum(production_count)
                  from production_agg
                  where time_window > now - 60000
        
        // 写入实时KPI
        if (production.size() > 0) {
            insert into kpi_realtime values (now, "production", production[0])
        }
        if (quality.size() > 0) {
            insert into kpi_realtime values (now, "quality_rate", quality[0])
        }
        
        sleep(60000)
    }
}

submitJob("kpi_calculate", "KPI计算", kpiCalculateTask)

// 4. 模拟数据
def generateMockProduction() {
    while (true) {
        data = table(
            "P" + string(rand(1000, 10)) as product_id,
            take(1..10, 10) as device_id,
            take(now(), 10) as timestamp,
            take(1, 10) as quantity,
            rand([true, true, true, true, false], 10) as qualified
        )
        production_stream.append!(data)
        sleep(5000)
    }
}

submitJob("mock_production", "模拟数据", generateMockProduction)

// 5. KPI看板接口
def getKpiDashboard() {
    return select * from kpi_realtime order by time_window desc limit 10
}

addFunctionView(getKpiDashboard)

print("生产KPI监控系统启动完成")

八、总结

本文详细介绍了DolphinDB生产指标监控：

  1. KPI定义：产量、良品率、OEE、能耗
  2. 指标计算：实时计算、批量计算
  3. 实时统计：时间窗口、滑动窗口、聚合引擎
  4. 趋势分析：趋势计算、同比环比、预测分析
  5. 指标告警：告警规则、告警推送
  6. 可视化展示：KPI看板、KPI排名

思考题：

  1. 如何设计灵活的KPI定义系统？
  2. 如何保证KPI计算的实时性？
  3. 如何实现KPI的自动预警？

参考资料

相关推荐
SkyWalking中文站5 小时前
认识 Horizon UI · 15/17：用模板定制控制台
运维·监控·自动化运维
笨鸟先飞的橘猫1 天前
skynet监控实践
监控·skynet
SkyWalking中文站2 天前
认识 Horizon UI · 11/17：运行时规则与实时调试
运维·监控·自动化运维
SkyWalking中文站4 天前
认识 Horizon UI · 6/17：Trace 探索器
运维·监控·自动化运维
SkyWalking中文站5 天前
认识 Horizon UI · 5/17：3D 基础设施地图
运维·监控·自动化运维
SkyWalking中文站6 天前
认识 Horizon UI · 1/17：SkyWalking 新一代可观测性控制台
运维·前端·监控
亲亲小宝宝鸭11 天前
前端性能监控：web-vitals
前端·性能优化·监控
kyriewen18 天前
前端错误监控最全指南：捕获 JS 异常、Promise 拒绝、资源加载失败，附上报代码
前端·javascript·监控
热门推荐
01GitHub 镜像站点022026 年 AI 编程工具终极横评：Cursor vs Claude Code vs Copilot vs Windsurf03AI科技热点日报 | 2026年07月01日04【AI】2026 年具身智能模型和世界模型总结052026 AI 编程工具终极实战指南：Cursor vs Claude Code vs Copilot，开发者该怎么选？062026 年 AI 大模型 & AI 编程工具实战全总结072026年6月AI大模型全景报告：GPT-5.6、Claude Opus 4.8、Gemini 3.5，中美AI三足鼎立谁主沉浮？082026 AI 编程工具选型横评：Cursor / Claude Code / Trae / Copilot 到底选谁（建议收藏·避坑版）09Agnes AI 免费 API 接入指南：文本、生图、生视频，一套接口全免费102026 国产 AI 大模型横评：DeepSeek、通义千问、Kimi、文心一言、星火、豆包谁更能打？