How acdeena1994/adaptive-nilm-with-human-in-the-loop-machine-learning works
This repository implements a single-point NILM system: ESP32 firmware sends electrical measurements to a Flask backend, which detects events and identifies appliances; results are persisted and surfaced in a dashboard where users can label events. The README describes the overall stages and feedback loop. The sampled backend excerpt confirms database setup, event detection, appliance matching, and feedback storage, but omitted sections limit verification of specific routes and runtime call relationships.
Main entry points: main.py Flask Ingestion · index.html Web Dashboard · esp32_code.ino ESP32 Sensing
Components 12
Sensing
- ESP32 Sensing
esp32_code.ino
Backend Analytics
- Flask Ingestion
main.py - Validation Controls
main.py - Event Detection
main.py - Appliance Matching
main.py
Storage and Learning
- Event and Data Store
main.py - Appliance Signatures
main.py - User Feedback
main.py - Adaptive Learning
main.py
Dashboard
- Web Dashboard
templates/index.html
Other components
- Electrical Load
- Dashboard User
Connections 14
- Electrical Load → ESP32 Sensing: measures power
README.md - ESP32 Sensing → Flask Ingestion: posts readings
README.md - Flask Ingestion → Validation Controls: validates requests
README.md - Flask Ingestion → Event Detection: detects events
README.md - Event Detection → Appliance Matching: analyzes events
README.md - Appliance Matching → Appliance Signatures: matches signatures
README.md - Flask Ingestion → Event and Data Store: stores results
README.md - Appliance Matching → Event and Data Store: stores predictions
README.md - Event and Data Store → Web Dashboard: visualizes results
README.md - Dashboard User → Web Dashboard: views and labels
README.md - Web Dashboard → Flask Ingestion: submits labels
README.md - Flask Ingestion → User Feedback: stores labels
README.md - User Feedback → Adaptive Learning: updates signatures
README.md - Adaptive Learning → Appliance Signatures: improves matching
README.md