Abstract
Machine learning methods offer versatile approaches for energy consumption forecasting, system efficiency improvement and smart grid management. In this study, we explore different ML models for the predictive energy optimisation such as linear regression, support vector machines, random trees and random forests, artificial neural networks, and long short-term memory networks. A systematic review of literature is carried out to review the traditional and modern methods. We experiment with real datasets to train and evaluate models where the performance measures are RMSE, MAE, and accuracy should be achieved. The findings show the effectiveness of deep learning models including LSTM networks in modeling temporal dependencies of energy consumption. The paper also addresses shortcomings like data quality, model interpretability, and scalability. Finally, future directions of research are proposed to enhance the predictive capabilities and integration with renewable energy sources.

This publication has 27 references indexed in Scilit: