An EV Charging Station Siting Model Based on Machine Learning

Aiming at the problems of the site selection method of EV charging station based on machine learning, such as single site selection factor, large subjective factor, insufficient sample labeling, simple labeling strategy and low location accuracy, a site selection method of EV charging station based on multi-source data fusion was proposed. This method constructed seven basic features from multi-source data, combined with PFAHP and TOPSIS methods, obtained the weight of POI in economic features, and proposed a semi-supervised learning method using mixed similarity automatic allocation markers to extend the training set. Based on the basic features and LogitBoost integration algorithm, an EV charging station siting model was constructed. In this paper, 6 districts in Shanghai were taken as the study area, and 4 sets of schemes and 8 experiments were designed to verify the superiority and applicability of the proposed method. The experimental results show that compared with the single similarity and manual labeling model, the accuracy of the recommended model is increased by 0.6%–2.6%, and the recall rate, F1 and ranking accuracy are all improved. The results show that the method proposed in this paper can better solve the problem of pre-location of EV charging stations, and has certain reference value for other public facilities.