方案对比：
┌────────────────┬──────────┬──────────┬──────────┐
│ 方案           │ 实时性    │ 语义理解  │ 决策能力  │
├────────────────┼──────────┼──────────┼──────────┤
│ 点对点接口      │ 中        │ 无       │ 无       │
│ ESB 集成       │ 中-高     │ 弱       │ 无       │
│ 数据仓库       │ 低 (T+1)  │ 弱       │ 有限     │
│ Coomia DIP    │ 高 （秒级） │ 强       │ 内建     │
└────────────────┴──────────┴──────────┴──────────┘

ObjectType: Equipment
  description: "核心业务实体 - 工厂设备"
  properties:
    - id: string (PK)
    - name: string
    - type: enum
    - status: enum [Active, Inactive, Pending, Archived]
    - priority: enum [Low, Normal, High, Critical]
    - metadata: dict
  computed_properties:
    - risk_score: float
    - health_index: float
    - trend: enum [Improving, Stable, Declining]

ObjectType: SensorReading
  description: "传感器时序数据"
  properties:
    - id: string (PK)
    - source_system: string
    - timestamp: datetime
    - value: float
    - unit: string
    - quality_flag: enum [Good, Suspect, Bad]
  time_series: true
  retention: "365d"

ObjectType: MaintenanceOrder
  description: "维护工单"
  properties:
    - id: string (PK)
    - type: enum
    - status: enum [Draft, Submitted, InReview, Approved, Rejected, Completed]
    - requester: string
    - start_time: datetime
    - end_time: datetime
    - severity: enum [Low, Medium, High, Critical]

ObjectType: FailurePrediction
  description: "故障预测结果"
  properties:
    - id: string (PK)
    - analysis_type: string
    - input_data: dict
    - result: dict
    - confidence: float [0-1]
    - model_version: string

ObjectType: AlertRule
  description: "告警规则配置"
  properties:
    - id: string (PK)
    - source_id: string
    - target_id: string
    - relation_type: string
    - weight: float
    - evidence: list[string]

Relations:
  - Equipment -> generates -> SensorReading
    cardinality: 1:N
    description: "设备产生传感器数据"

  - Equipment -> triggers -> MaintenanceOrder
    cardinality: 1:N
    description: "设备触发维护工单"

  - SensorReading -> analyzedBy -> FailurePrediction
    cardinality: N:1
    description: "传感器数据被分析引擎处理"

  - FailurePrediction -> impacts -> Equipment
    cardinality: N:M
    description: "预测结果反馈到设备"

  - Equipment -> linkedVia -> AlertRule
    cardinality: N:M
    description: "设备关联告警规则"

  - MaintenanceOrder -> resolvedBy -> FailurePrediction
    cardinality: N:1
    description: "维护工单通过预测分析得到方案"

┌──────────────────────────────────────────────────────┐
│                    应用层                              │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐           │
│  │ 设备看板  │  │ 维护报告  │  │ 移动端   │           │
│  └────┬─────┘  └────┬─────┘  └────┬─────┘           │
│       └─────────────┼─────────────┘                  │
│                     │                                 │
│  ┌──────────────────┴─────────────────────┐          │
│  │         Ontology 语义层                  │          │
│  │   Equipment --- SensorReading --- MaintenanceOrder│
│  │       |           |           |                    │
│  │   FailurePrediction ------- AlertRule             │
│  │   统一模型 / 统一查询 / 统一权限                      │
│  └──────────────────┬─────────────────────┘          │
│                     │                                 │
│  ┌──────────────────┴─────────────────────┐          │
│  │   数据接入： CDC | API | Stream | Batch  │          │
│  └────────────────────────────────────────┘          │
└──────────────────────────────────────────────────────┘

from ontology_sdk import OntoPlatform

platform = OntoPlatform()

# 查询高风险设备及传感器数据
entities = (
    platform.ontology
    .object_type("Equipment")
    .filter(status="Active")
    .filter(priority__in=["High", "Critical"])
    .include("SensorReading")
    .include("MaintenanceOrder")
    .order_by("updated_at", ascending=False)
    .limit(100)
    .execute()
)

for entity in entities:
    print(f"设备： {entity.name} | 健康评分： {entity.health_index}")

    # 检查异常传感器数据，自动触发故障预测
    recent_bad = [d for d in entity.sensorreadings
                  if d.quality_flag == "Bad"]
    if len(recent_bad) > 5:
        platform.actions.execute(
            "ExecuteFailurePrediction",
            target_id=entity.id,
            analysis_type="anomaly_detection",
            parameters={"window": "24h"}
        )

rules:
  - name: "高风险设备告警"
    trigger: Equipment.risk_score > 80
    actions:
      - alert: critical
      - action: Escalate(severity=Critical)

  - name: "健康趋势恶化"
    trigger: Equipment.trend == "Declining" AND priority in [High, Critical]
    actions:
      - alert: warning
      - action: ExecuteFailurePrediction(type=root_cause)

  - name: "传感器数据异常"
    trigger: SensorReading.quality_flag == "Bad" count > 10/hour
    actions:
      - alert: warning

  - name: "维护工单自动升级"
    trigger: MaintenanceOrder.severity == "Critical"
    actions:
      - action: Escalate(severity=Critical)
      - notification: sms -> on_call

from intelligence_plane.models import PredictionModel
from datetime import timedelta

class FailurePredictionModel(PredictionModel):
    def __init__(self):
        super().__init__(
            name="failureprediction_v2",
            input_type="Equipment",
            output_type="FailurePrediction"
        )

    def predict(self, entity, context):
        history = (
            context.ontology.object_type("SensorReading")
            .filter(source_id=entity.id)
            .filter(timestamp__gte=context.now - timedelta(days=90))
            .order_by("timestamp")
            .execute()
        )
        features = self.extract_features(history)
        prediction = self.model.predict(features)
        return {
            "level": prediction["level"],
            "confidence": prediction["confidence"],
            "factors": prediction["contributing_factors"],
            "actions": prediction["recommended_actions"]
        }

首年 ROI = (220 - 55) / 55 * 100% = 300%
三年 ROI = (220*3 - 55 - 15*2) / (55 + 15*2) * 100% = 676%