A Review of Electric Vehicle Charging Demand Forecasting Models by Prediction Horizon: A Multi-Criteria Decision Analysis Approach

Abstract

The rapid increase in the number of electric vehicles (EVs) has raised significant concerns regarding the stability of power grid operations and the effective management of charging infrastructure. In this review paper, we categorize EV charging demand forecasting into three perspectives: short-term, medium-term, and long-term. Subsequently, we compare and analyze the input characteristics essential for enhancing forecasting accuracy. Additionally, we comprehensively examine the features, advantages, and disadvantages of various methodologies employed in recent studies. Specifically, a systematic review was conducted encompassing contemporary research trends, including time-series models, deep learning (DL) approaches, growth-curve models, and scenario-based simulation models. Furthermore, we propose a model selection framework that comprehensively considers forecasting accuracy alongside realistic operational scenarios, such as the data-collection period and the optimal scale of EV charging stations. This framework employs the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), a well-known multi-criteria decision making (MCDM) method technique, to evaluate the relative merits of different forecasting models. Ultimately, this study aims to provide researchers and practitioners engaged in EV charging demand forecasting with clear analytical criteria and practical guidelines for model selection.