Research on Cooperation Between Wind Farm and Electric Vehicle Aggregator Based on A3C Algorithm

• Published in: IEEE access Vol. 9; pp. 55155 - 55164
• Main Authors: Pan, Yang, Wang, Weiye, Li, Yanbin, Zhang, Feng, Sun, Yanting, Liu, Dunnan
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2021; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: A3C algorithm; Algorithms; Alternative energy sources; Charging; charging service pricing; Cooperation; Economics; Electric power systems; Electric vehicle aggregator; Electric vehicles; Electrical loads; Energy consumption; Flexibility; Income; Income distribution; Machine learning; Model testing; Power consumption; Power markets; Pricing; Profitability; Reinforcement learning; Renewable energy sources; Renewable resources; Schedules; Vehicle-to-grid; wind farm; Wind farms; Wind power; Wind power generation; Wind turbines
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2021.3071803

As renewable energy sources such as wind are connected to the grid on a large scale, the safe and stable operation of the power system is facing challenges and the demand for flexibility is becoming increasingly prominent. In recent years, with the advancement of Vehicle-to-Grid (V2G) technology, electric vehicles (EVs) have become a non-negligible flexibility resource for the power system and an emerging path to solve the renewable energy consumption problem. To address the problem of wind farms' difficulty in making profits in the power market, this paper considers the cooperation between wind farms and EV aggregators and uses the levelable characteristics of EVs charging load to ease the anti-peak characteristics of wind power. Given this, this paper proposes a cooperation mode between the wind farm and the Electric Vehicle (EV) aggregator, constructs a cooperation income and income distribution model, and solves the model using the Asynchronous Advantage Actor-Critic (A3C) reinforcement learning algorithm. Finally, based on the simulation analysis of historical data, the following conclusions are drawn: (1) the cooperation between the wind farm and the EV aggregator can effectively mitigate the negative impact of the anti-peak characteristics of wind power on profitability and achieve an increase in overall economic benefits; (2) the income distribution based on the Shapley value method ensures that the respective income of the wind farm and the EV aggregator increase after cooperation, which is conducive to the promotion of the willingness of both parties to cooperate; (3) the A3C reinforcement learning algorithm is applied to solve the model with good convergence to achieve fast and continuous intelligent pricing decisions for EV aggregators, thus optimizing the charging schedule of EVs promptly.