Abstract

Purpose: This study investigates the application of supervised learning and reinforcement learning algorithms for the optimization of smart grid systems, focusing on improving efficiency and reducing operational costs. The research aims to evaluate the performance of these algorithms using the IEEE standard dataset, with a specific emphasis on metrics such as root mean square error (RMSE) and cost reduction.

Design: A mathematical design is employed in this study, utilizing machine learning models to predict and optimize energy distribution within a smart grid framework. The supervised learning approach is applied to forecast grid demand, while reinforcement learning is utilized to optimize decision-making processes. The study leverages the IEEE standard dataset, which provides real-world power system data, and evaluates the algorithms based on RMSE for forecasting accuracy and operational cost reduction.

Findings: The results demonstrate that both supervised learning and reinforcement learning algorithms significantly improve the efficiency of energy distribution, with the reinforcement learning model achieving notable reductions in operational costs while maintaining accuracy. The RMSE values indicate a high level of predictive accuracy, with reinforcement learning outperforming supervised learning in the long-term optimization of grid operations.

Originality: This research contributes to the growing body of knowledge on the application of machine learning in smart grid optimization, specifically in energy distribution management. By combining supervised learning and reinforcement learning, the study provides a novel approach to optimizing grid operations, with practical implications for both energy providers and consumers.

Keywords: Smart grid, machine learning, supervised learning, reinforcement learning, IEEE standard dataset, optimization, RMSE, cost reduction.