Maroua-a/Smart-home-app

Smart Home App PFE 2026 # Smart Home – Energy Management Application with AI ## Project Overview This is a full-stack smart home energy management system developed as a final year project (PFE) at USTHB. It combines a Deep Q-Network (DQN) reinforcement learning agent with user clustering to provide real-time energy saving recommendations and device control. The system consists of: - A Flask backend exposing a REST API, managing the DQN agent, clustering, and MQTT communication. - A React Native frontend (Expo) for mobile control and visualisation. - Optional integration with Home Assistant via MQTT discovery. --- ## Key Features | Feature | Description | |---------|-------------| | User authentication | Registration, login, password reset | | Dashboard | Overview of rooms, devices, weather, active devices, and savings | | Room control | Turn devices on/off, adjust intensity (brightness, temperature) | | AI recommendations | DQN agent suggests actions (turn off, reduce) based on state and user preferences | | Energy analytics | Compare Normal vs Eco consumption, view history per day/week/month | | User preferences | Set preferred temperature and brightness per room; parent preferences override child preferences | | Multilingual | French and English (i18n) | | MQTT bridge | Bidirectional communication with Home Assistant | | User clustering | Classifies users as Economical, Normal, or Energy-intensive based on behaviour | --- ## Technology Stack ### Backend - Python 3.9+ - Flask + Flask-CORS - SQLite - PyTorch (DQN) - Scikit-learn (K-Means clustering) - Paho-MQTT - Requests (OpenWeatherMap API) ### Frontend - React Native + Expo - React Navigation (Stack + Bottom Tabs) - i18n-js - Expo Vector Icons ### IoT Integration - Home Assistant (via MQTT auto-discovery) - Mosquitto MQTT broker --- ## Project Structure ``` smart-home-app/ ├── backend/ │ ├── app.py # Main Flask application │ ├── dqn_agent.py # DQN agent implementation │ ├── clustering.py …