Research on EV User Portraits for Vehicle-to-Grid Interaction Based on Behavioral Clustering and Evolutionary Game

Abstract

As the global adoption of electric vehicles (EVs) grows rapidly, Vehicle-to-Grid (V2G) technology has increasingly established itself as a critical mechanism for enhancing grid flexibility and improving user economic returns. This study systematically addresses the inherent coordination challenges between grid operators and end-users within V2G systems by developing an integrated modeling framework. The proposed methodology first applies user behavior clustering techniques to characterize diverse EV user groups' energy storage attributes and behavioral patterns. An optimization objective function is subsequently constructed based on evolutionary game theory, comprehensively incorporating grid operational costs and user-side economic benefits. Extensive validation through simulation experiments confirms the model’s practical efficacy and computational robustness. Specifically, the K-Means (KM) clustering algorithm is employed to construct detailed user portraits, supplemented by Density Sampling (DS) for enhanced data refinement and Training by Iteration (TB) to strengthen the convergence stability of evolutionary strategies. The findings clearly demonstrate that the proposed approach not only accurately identifies distinct user categories but also promotes significant cost reduction for both grid operators and end-users, thereby offering valuable theoretical insights and practical support for the commercialization of V2G technology.