Machine Learning based Design of Energy Management System using Non-Intrusive Load Monitoring Strategy

Abstract

In this work, a Non-Intrusive Load Monitoring (NILM) system is designed for smart homes based on smart energy meters. The proposed solution simplifies the monitoring process by using a single set of sensors, in contrast to conventional systems that call for several sensors. From different appliances and their combinations, crucial electrical information, such as voltage, current, active power, reactive power, apparent power, power factor, and frequency was gathered. Proposed intelligent data analysis employing machine learning methods, particularly Artificial Neural Networks (ANN) and Deep Neural Networks (DNN), is the key innovation. These methods accurately categorize various appliance kinds, with DNN performing best in real-time circumstances. Additionally, there is a built-in email alert system that activates whenever there are odd electrical surges. The streamlined user interface, offers accurate identification of individual appliances and combinations in real-time forecasts based on past data. Future developments include a larger appliance data-set, improved categorization methods, load forecasting, and the creation of a smartphone app for real-time energy consumption data and automatic control, among other things. In a net-shell, this study offers an effective, single-sensorbased solution for load monitoring in smart homes, representing a substantial development in NILM technology. This technology claims to improve resource utilization and energy management in both domestic and commercial settings.