基于多智能体Q学习算法的能源互联园区协调调度  

Coordinated scheduling of energy interconnected parks based on multi-agent Q-learning algorithm

作　　者：黄文杰[1] 崔雪[1] 陈君[1] 饶云杰 HUANG Wenjie;CUI Xue;CHEN Jun;RAO Yunjie(School of Electrical Engineering and Automation,Wuhan University,Wuhan 430072,China)

机构地区：[1]武汉大学电气与自动化学院,湖北武汉4730072

出　　处：《武汉大学学报（工学版）》2022年第11期1141-1148,共8页Engineering Journal of Wuhan University

基　　金：国家自然科学基金面上项目(编号:51977160)。

摘　　要：针对能源互联园区中各主体利益诉求不同以至于难以协调调度的问题,提出了基于多智能体Q学习算法的能源互联园区协调调度方法。首先,将能源互联园区主体划分为5个智能体;其次,构建了各智能体的决策模型,并确立了以智能体之间利益均衡为目标的目标函数;最后,以某能源互联园区为算例进行仿真,利用Q学习算法进行求解,在能源供应商、可再生能源服务商、电动汽车收益最大以及园区能源服务商成本最低时,达到利益均衡点,得到园区各设备和负荷的出力、需求分布,即为园区协调调度下的最优均衡运行策略。算例结果表明,所提方法的有效性和实用性可为未来能源互联园区的调度提供参考。Aiming at the problem that it is difficult to coordinate the scheduling due to the different interest demands of various subjects in the energy interconnection park, a coordinated scheduling method of energy interconnection park based on multi-agent Q-learning algorithm is proposed.Firstly,the main body of the energy interconnection park is divided into five agents: energy suppliers, energy service providers, renewable energy service providers, electric vehicles and users in the park. Secondly, the decision-making model of each agent under constraints is constructed, and the objective function aiming at the balance of interests among agents is established. Finally, taking an energy interconnection park as an example, the Q-learning algorithm is used to solve the model.When the benefits of energy suppliers,renewable energy service providers and electric vehicles are the largest and the cost of energy service providers in the park is the lowest,the interest equilibrium point is reached,and the output and demand distributions of various equipment and loads in the park are obtained,which is the optimal balanced operation strategy under the coordinated dispatching of the park. The example results show that the proposed method is effective and practicable, and can provide a reference for the scheduling of energy interconnection parks in the future.

关 键 词：能源互联园区 协调调度 多智能体 Q学习算法 

分 类 号：TM743[电气工程—电力系统及自动化]