Electric Vehicle Charging Load Prediction Based on User Portrait

To address the problem that traditional electric vehicle (EV) charging load prediction methods cannot portray users' charging behavior in detail, this paper proposes an EV charging load prediction model based on user portrait. Firstly, the key features are extracted by analyzing the behavior of EV charging users. Secondly, we carry out a multidimensional clustering analysis of key features based on the improved Gaussian mixture model (GMM) algorithm and generate typical user portraits. Finally, a Monte Carlo simulation method is used to build the EV charging load prediction model based on the user portraits. The results show that the user portrait generated by the improved GMM algorithm is more accurate than that of the unimproved GMM algorithm. The performance of the EV charging load prediction model based on the user portraits is better, which can provide a basis for guiding users to charge in an orderly manner and planning the construction of charging infrastructure.