SUSTAINABLE ENERGY MANAGEMENT IN SMART GRIDS USING MACHINE LEARNING

Abstract

This research paper investigates an approach for sustainable energy management in smart grids using machine learning by forecasting renewable energy generation using Long Short-Term Memory (LSTM) networks. An energy management system is developed that collects real-world time-series data from renewable sources (solar and hydro) and then applies data pre-processing techniques, including dropout regularization and normalization, to build an LSTM model that predicts generation trends. To validate the accuracy and sustainability impact of the model, performance metrics such as Mean Squared Error (MSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) are determined. This specific model enables switching between renewable sources and conventional grid supply, which is based on predicted generation demand. The results indicate enhanced resource utilization, reduced dependency on fossil fuels, and support for smart grid automation.