A systematic review and meta-analysis of machine learning, deep learning, and ensemble learning approaches in predicting EV charging behavior

Machine learning (ML) and deep learning (DL) have enabled algorithms to autonomously acquire knowledge from data, facilitating predictive and decision-making capabilities without explicit programming. This transformative potential has reshaped industries by utilizing data-driven insights. ML and DL models have found extensive application within the domain of electric vehicle (EV) charging predictions. These techniques effectively forecast EV charging behavior, considering variables such as charging station location, time of day, battery state of charge, EV owner behavioral patterns, and weather conditions. This study aims to comprehensively evaluate ML and DL applications in forecasting EV charging behavior while introducing a systematic categorization, a notable gap in current literature. A comprehensive dataset, selected from both the Web of Science and the Scopus database, sourced from Elsevier Journal, was thoughtfully chosen to cover relevant research studies for the purpose of achieving this goal. Furthermore, our research emphasizes the significance of model evaluation and explores the usefulness of commonly employed ML and DL techniques within forecasting approaches, including Short-Term Load Forecasting (STLF), Medium-Term Load Forecasting (MTLF), and Long-Term Load Forecasting (LTLF) to ensure precise predictions. Within this framework, the selected publications are classified based on methodology, research focus, objectives, publication year, geographic origin, and research outcomes. While both ML and DL techniques exhibit substantial potential in predicting EV charging behavior and mitigating challenges posed by the rising adoption of EVs, our analysis demonstrates that ensemble learning techniques surpass them in terms of predictive performance.