
Broadcast Power Environment Smart Monitoring Platform
AI-Native 3-tier Power Environment Monitoring for Broadcast & Telecom Sites
| Architecture | 4 layers: Field → Edge → Platform → AI |
| Deployment | 1 HQ + 6 Regional + 50+ Sites |
| Collection | 265+ parameters across 8 monitoring domains |
| Edge | RK3506 industrial gateway (10×RS485, 12×DI/DO) |
| Data store | MySQL semi-sync + TDengine + Redis |
| Visualization | Three.js 2.5D BIM, thermal heatmap |
| AI | Qwen2.5-7B local + Qwen2.5-VL + DH-Claw workflow |
| Protocols | Modbus RTU / SNMP v3 / ONVIF / BACnet / OSDP |
| Uplink | MQTT TLS 1.3 / HTTPS mTLS / gRPC |
| Downlink | WebSocket / OAuth 2.0 / GB/T 28181 |
| Availability | ≥99.95% (annual downtime < 4.38 h) |
| Response | P1 alarm ≤30 s; failover ≤60 s |
| Accuracy | AI root-cause ≥80%; alarm precision ≥95% |
| Power | ~154 items: PDU, ATS, UPS, battery, generator, transformer |
| Environment | ~44 items: T/H, leak, AC, lighting, weight |
| Security | ~23 items: video, access (national crypto), IR, fire link |
| Meteorology | ~19 items: weather station, quake, lightning, wind |
| Energy | ~14 items: PUE, cabinet power, time-of-use, carbon |
| Capacity | ~6 items: rack space, power, weight, port usage |
| Network | ~7 items: switch, MQTT broker, collector online, link |
| Fire | ~5 items: smoke, heat, gas, water pressure, emergency light |



