MULTI-AGENT SMART METER DASHBOARD SYSTEMS

– The backend is a FastAPI microservices application backed by four purpose-selected database engines: PostgreSQL 15 for relational metadata and billing, InfluxDB 2.7 for time-series telemetry, Qdrant for vector similarity search, and Neo4j 5 for grid topology. A five-agent retrieval-augmented generation (RAG) AI subsystem uses a locally hosted large language model (LLM) for conversational energy advisory, cascade fault analysis, and hallucination-free bill forecasting. A Next.js 15 frontend with role-differentiated portals and WebSocket push delivers a responsive interface to consumers, industrial clients, and grid administrators. Evaluation on a 72-hour continuous run demonstrates sub-500 ms anomaly alert propagation, 100% fault detection recall, and a monthly bill prediction MAPE of 2.3%.