A Data-Driven Management Strategy of Electric Vehicles and Thermostatically Controlled Loads Based on Modified Generative Adversarial Network
Yuechuan Tao, Jing Qiu, Shuying Lai
Abstract
As part of the transportation system electrification and decarbonization, electric vehicles (EVs) have been experiencing a steady growth in some countries in recent years. However, the uncoordinated charging of EVs brings negative effects to the power grid. Therefore, a new intelligent energy management strategy needs to be involved in the transportation electrification progress to address the variability of flexible loads and energy storage in the active distribution network. In this article, we have proposed a coordinated dispatch strategy of EVs and thermostatically controlled loads (TCLs) based on a modified generative adversarial network (GAN). TCLs are utilized to complement the limits of EVs’ driving behaviors. EVs are modeled as battery energy storage systems (BESSs), and TCLs are modeled as virtual energy storage systems (VESSs). Machine learning is integrated into a bilevel optimization problem to determine the steady-state power dispatch and the energy storage control of the VESS. The proposed method is verified on the IEEE 33-bus system. Based on the simulation results, it can be concluded that the proposed data-driven method can outperform the conventional model-based method in terms of accuracy. Also, the modified GAN helps the training process to be less affected by the missing data. Comparative studies are conducted to show that the coordinated dispatch strategy of EVs and TCLs can maintain the voltage stability in the distribution system by compensating for the drawbacks of each other.