Architecture
Technical architecture
From plant data to operator decisions. Solid blocks are in the current MVP; dashed blocks are future industrial integration.
Data → decision
System layers
Current MVPFuture integration
Data sources
Production SchedulePlant SensorsSmart MeterHeat DemandWeatherElectricity PricesPV ForecastGrid Signals
MVP: simulated plant data + public price/weather data
Energy LIFT data layer
Forecasting models
Production ForecastHeat ForecastSolar ForecastLoad Forecast
Optimization engine
Heat DispatchThermal StorageBatteryGrid ConstraintProduction Constraint
Resilience / re-optimization engine
Explainability agent
Plant operator dashboard
SCADA / EMS integration
LaterFuture industrial integration — closed-loop setpoints
Intelligence stack
How the intelligence is layered
01
Prediction models
Predict demand and operating conditions.
02
Mathematical optimization
Calculates the best feasible operating schedule.
03
Constraint / safety layer
Prevents violation of industrial and grid limits.
04
Explainability agent
Translates technical decisions into understandable operational recommendations.
Rule: the explainability agent must never override the mathematical optimizer or safety constraints. It reads results; it cannot write setpoints.
Backend contract
API endpoints
- POST /api/optimize
- POST /api/scenario
- GET /api/forecast
- GET /api/assets
- GET /api/decisions
- GET /api/performance
Currently served by deterministic mock services. A Python/FastAPI backend replaces them without UI changes.