基于深度强化学习算法的V2G充换电供需两侧调度策略优化  

Optimization of V2G charging and swapping power supply and demand sides scheduling strategy based on deep reinforcement learning algorithm

作　　者：尹昊 YIN hao(Ningxia Architectural Design and Research Institute Co.,Ltd.,Yinchuan 750002,China)

机构地区：[1]宁夏建筑设计研究院有限公司,宁夏银川750002

出　　处：《电气应用》2024年第6期8-15,共8页Electrotechnical Application

摘　　要：提出了一种基于深度强化学习(DRL)算法的V2G充换电供需两侧调度策略。该策略采用了最新的深度Q网络(DQN)算法,以最小化电网运行成本和新能源汽车用户成本为目标。针对传统方法难以处理高维状态空间和动作空间的问题,使用DQN算法对策略进行端到端的训练与优化,并通过引入双重学习机制和优先级回放策略提高学习效率。以某城市电网实际运行数据为基础,构建了包含100辆新能源汽车的V2G模型实验平台,并对调度策略在不同负荷曲线和新能源渗透率下的性能进行了测试。实验结果表明,与经典的规则化控制方法相比,深度强化学习算法显著降低了供需不平衡,提升了能量调度策略效率,降低了系统运行成本,电网需求侧峰值削减率平均提高了15%,电网稳定性得到了保障。A scheduling strategy for V2G charging and discharging that accounts for both supply and demand,based on the Deep Reinforcement Learning(DRL)algorithm is proposed.This strategy adopts the latest Deep Q-Network(DQN)algorithm,aiming to minimize both the operational costs of the electric grid and the costs for new energy vehicle users.Addressing the issue traditional methods face when dealing with high-dimensional state and action spaces,the DQN algorithm is used for end-to-end training and optimization of the strategy,while a dual learning mechanism and priority replay strategy are introduced to improve learning efficiency.For this study,actual operational data from a city’s electric grid was used to construct a V2G model experimental platform that included 100 new energy vehicles.The performance of the scheduling strategy under different load curves and new energy penetration rates was tested.Experimental results show that,compared to classical rule-based control methods,the deep reinforcement learning algorithm significantly reduced supply and demand imbalances and improved the efficiency of the energy scheduling strategy,lowering system operational costs.The average peak shaving rate on the demand side of the grid increased by 15%,ensuring grid stability.

关 键 词：深度强化学习 V2G技术 电力系统 能量优化调度 新能源汽车 

分 类 号：U491.8[交通运输工程—交通运输规划与管理] TM73[交通运输工程—道路与铁道工程] TP18[电气工程—电力系统及自动化]