IntelliSocket: An Intelligent Electric Socket with ANN and Autoencoder for Energy Forecasting and Anomaly Detection

This paper presents the development of an intelligent electric socket equipped with IoT (Internet of Things) technology designed for real-time energy tracking, predictive analytics, and enhanced user convenience, addressing the growing need for efficient energy management in residential settings. The voltage and current are measured using the ZMPT101B and ACS712 sensors across two sockets, with an ESP32 microcontroller collecting data from the sensors and submitting it to Google Sheets. Two advanced computational models, namely a neural network-based model (ANN) for energy consumption forecasting and Autoencoder for anomaly detection, are implemented on a computer. The processed data is stored in Firebase, allowing continuous real-time updates and seamless integration with a mobile application. This application allows users to monitor real-time of the energy consumption, control the connected sockets remotely, and receive notifications in case of unusual energy consumption. The outcome of the experiment shows that the ANN model achieves R-squared values of 0.9813 for Socket 1 and 0.9600 for Socket 2, while the Autoencoder algorithm yields a Mean Reconstruction Error (MRE) of 0.0007 for Socket 1 and 0.0003 for Socket 2. The system that was created is accurate and efficient, providing actionable insight to enhance reliability and support sustainable energy management, making it an invaluable tool for residential and small-scale energy optimization.