Study on the Operation Mode of Electric Vehicle-Based Virtual Power Plants Using TGCN-MPC

The surge in electric vehicles (EVs) necessitates effective load management and scheduling for charging stations. This study introduces an optimized scheduling method for an EV Virtual Power Plant (EVPP) using Temporal Graph Convolution Network (TGCN) and Model Predictive Control (MPC). TGCN predicts charging station load by capturing spatial and temporal dependencies, utilizing historical load data and influencing factors. Combined with MPC, it enables real-time, optimized resource allocation based on forecasted loads. Experimental results indicate substantial improvements in load prediction accuracy and scheduling efficiency, enhancing EVPP operational stability and reducing grid fluctuations compared to traditional methods.