Inspired by Envision (远景) AI Energy System visit. Adds 4 major feature modules:
- AI Prediction (AI预测): PV/load/COP forecasting, self-consumption optimization
- Energy Strategy (策略优化): TOU pricing, peak/valley strategy, weather integration
- Carbon Asset (碳资产): CCER, green certificates, targets, benchmarks, reports
- AI Operations (AI运维): device health scoring, anomaly detection, predictive maintenance
20 new models, 4 services, 58 API endpoints, 23 frontend pages.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
System Management:
- System Settings page with 8 configurable parameters (admin only)
- Audit Log page with filterable table (user, action, resource, date range)
- Audit logging wired into auth, devices, users, alarms, reports API handlers
- SystemSetting model + migration (002)
Device Detail:
- Dedicated /devices/:id page with 4 tabs (realtime, historical trends, alarm history, device info)
- ECharts historical charts with granularity/time range selectors
- Device name clickable in Devices and Monitoring tables → navigates to detail
Email & Scheduling:
- Email service with SMTP support (STARTTLS/SSL/plain)
- Alarm email notification with professional HTML template
- Report scheduler using APScheduler for cron-based auto-generation
- Scheduled report task seeded (daily at 8am)
UI Enhancements:
- Dark mode toggle (persisted to localStorage, Ant Design darkAlgorithm)
- Data comparison view in Analysis page (dual date range, side-by-side metrics)
- i18n framework (i18next) with zh/en translations for menu and common UI
- Language switcher in header (中文/English)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Major additions across backend, frontend, and infrastructure:
Backend:
- IoT collector framework (Modbus TCP, MQTT, HTTP) with manager
- Realistic Beijing solar/weather simulator with cloud transients
- Alarm auto-checker with demo anomaly injection (3-4 events/hour)
- Report generation (PDF/Excel) with sync fallback and E2E testing
- Energy data CSV/XLSX export endpoint
- WebSocket real-time broadcast at /ws/realtime
- Alembic initial migration for all 14 tables
- 77 pytest tests across 9 API routers
Frontend:
- Live notification badge with alarm count (was hardcoded 0)
- Sankey energy flow diagram on dashboard
- Device photos (SVG illustrations) on all device pages
- Report download with status icons
- Energy data export buttons (CSV/Excel)
- WebSocket hook with auto-reconnect and polling fallback
- BigScreen 2D responsive CSS (tablet/mobile)
- Error handling improvements across pages
Infrastructure:
- PostgreSQL + TimescaleDB as primary database
- Production docker-compose with nginx reverse proxy
- Comprehensive Chinese README
- .env.example with documentation
- quick-start.sh deployment script
- nginx config with gzip, caching, security headers
Data:
- 30-day realistic backfill (47K rows, weather-correlated)
- 18 devices, 6 alarm rules, 15 historical alarm events
- Beijing solar position model with seasonal variation
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
