Integrated IoT and AI-Driven Smart Refrigerator System

In addition to the considerable economic costs, inefficient electricity usage significantly contributes to environmental pollution. To tackle this issue, we developed an IoT system utilizing various electronic devices, including an ESP8266 microcontroller, a temperature-humidity sensor, an electromechanical switch, a power metering for appliance control, and a Raspberry Pi 4. This system records and stores data on temperature, humidity, door openings, and power consumption. By leveraging the open-source Home Assistant platform, we successfully synchronized these components to achieve our desired outcomes. Using machine learning techniques, such as Long Short-Term Memory (LSTM), we can predict power consumption based on temperature and humidity. We also employed the Linear Regression method to correlate temperature differences with the duration of refrigerator door openings. Finally, we applied the Anomaly Detection and Isolation Forest model to identify significant variations in temperature and humidity when the door is not in use.