設備預測性維護
AI → HumanAI 分析感測器資料,在設備故障發生前預測風險。
5 個節點 · 5 條連接manufacturing
eventagenthumanapi
視覺化
感測器資料串流event
來自機械的震動、溫度與壓力讀值。
↓sequential→ 震動與熱能分析
震動與熱能分析agent
ML 模型偵測異常模式,識別磨損或即將發生的故障。
↓sequential→ 故障機率評分
↓timeout→ 維護團隊告警
故障機率評分system
風險分類:低(低於 20%)、中(20-60%)、高(高於 60%)。
↓conditional→ 維護團隊告警
維護團隊告警api
推送告警至 CMMS 與團隊行動裝置。
↓sequential→ 安排維修
安排維修human
維護負責人審查後安排停機時間窗並指派人員。
uc-equipment-failure-prediction.osop.yaml
osop_version: "1.0"
id: "equipment-failure-prediction"
name:"設備預測性維護"
description:"AI 分析感測器資料,在設備故障發生前預測風險。"
nodes:
- id: "sensor_stream"
type: "event"
name: "感測器資料串流"
description: "來自機械的震動、溫度與壓力讀值。"
- id: "ai_analysis"
type: "agent"
subtype: "llm"
name: "震動與熱能分析"
description: "ML 模型偵測異常模式,識別磨損或即將發生的故障。"
security:
risk_level: "medium"
- id: "failure_probability"
type: "system"
name: "故障機率評分"
description: "風險分類:低(低於 20%)、中(20-60%)、高(高於 60%)。"
- id: "maintenance_alert"
type: "api"
name: "維護團隊告警"
description: "推送告警至 CMMS 與團隊行動裝置。"
- id: "schedule_repair"
type: "human"
subtype: "review"
name: "安排維修"
description: "維護負責人審查後安排停機時間窗並指派人員。"
security:
approval_gate: true
edges:
- from: "sensor_stream"
to: "ai_analysis"
mode: "sequential"
- from: "ai_analysis"
to: "failure_probability"
mode: "sequential"
- from: "failure_probability"
to: "maintenance_alert"
mode: "conditional"
when: "failure_risk > 0.2"
- from: "maintenance_alert"
to: "schedule_repair"
mode: "sequential"
- from: "ai_analysis"
to: "maintenance_alert"
mode: "timeout"
timeout_sec: 60
label: "Escalate if analysis stalls"