Multi-dimensional Data Generation Method of Electric Vehicle Charging Behaviors Based on Improved Generative Adversarial Network

Abstract

Electric vehicle (EV) charging behaviors are important for load forecasting, but charging behavior data collection is time-consuming and labor-intensive. To solve this problem, an improved generative adversarial network (GAN) based charging behavior data generation method for EVs was proposed. In this method, convolutional neural network is used as the structure, Wasserstein distance is used as the loss function of the neural network, and conditional labels are added to make the generator learn the probability distribution mapping relationship that meets the corresponding conditions. These methods finally construct the EV charging behavior data generation network. In the end, this paper compares this method with traditional data generation algorithm synthetic minority oversampling technique (SMOTE) and conditional generative adversarial network (CGAN). The result shows that this method can effectively learn the distribution rule of small sample data and effectively expand the data scale.