A Prediction Method of Charging Station Expected Demand Based on Graph Structure

In view of the limitations of the traditional method is unable to predict the demand for electric vehicle charging stations (EVCS), this paper proposes an expected demand prediction method of charging stations based on graph structure. Our method is used to predict the expected demand of any station (in operation or to be built) in the urban charging station network. Different from previous studies, our method is used to transform the network of urban charging stations into a graph structure and build a deep learning model to predict the expected demand of charging stations. First, this method uses multi-graph convolution to capture the irregular spatial relationship between charging stations. In addition, the attention mechanism is introduced to further consider the relationship between features. Finally, the dependencies acquired in the multi-mode are integrated, and the expected demand prediction results are output through the multilayer perceptron. We build a real data set collected from an electric vehicle charging platform in Shanghai to evaluate the proposed method. The experimental results show that the proposed method is feasible, and the prediction accuracy of the proposed deep learning model is better than the baseline models.