Spatiotemporal charging demand models for electric vehicles considering user strategies

Abstract

As the number of urban electric vehicles continues to increase, accurate prediction of the electric vehicle (EV) spatial and temporal distribution charging demand is of great importance for safely operating the power grid. Due to the uncertainty and variability of EV user charging and discharging strategies, the strategic factors behind user behavior become the key to influencing whether the charging demand prediction results are reasonable. As a result, this paper proposes a charging demand prediction model based on real-time data from Baidu map that can interpret EV user driving strategies and charging strategies based on the strategy learning capability of generative adversarial imitation learning. This paper first analyzes the correlation between strategy factors and SOC in user charging and discharging data, then describes establishing a 24-hour SOC prediction model for a single vehicle, and finally discusses building a spatiotemporal model of charging demand in the region on this basis. The results demonstrate that, while it can be combined with real-time traffic data, the method has better prediction accuracy and robustness compared with the current mainstream prediction methods and high application value.