基于深度强化学习与改进Jensen模型的风电场功率优化  

Power Optimization of Wind Farms Based on Improved Jensen Model and Deep Reinforcement Learning

作　　者：王冠朝 霍雨翀 李群 李强 WANG Guanchao;HUO Yuchong;LI Qun;LI Qiang(Department of Electrical Engineering,Nanjing University of Science and Technology,Nanjing 210094,China;Electric Power Research Institute of State Grid Jiangsu Electric Power Co.,Ltd.,Nanjing 211103,China)

机构地区：[1]南京理工大学电气工程系,江苏南京210094 [2]国网江苏省电力有限公司电力科学研究院,江苏南京211103

出　　处：《中国电力》2025年第4期78-89,共12页Electric Power

基　　金：国家电网有限公司科技项目(攻关团队项目)(含多构网型变流器的中远海风电场经柔直并网主动频率支撑关键技术,5108-202218280A-2-241-XG)。

摘　　要：风电场的功率捕获能力受多种因素的制约。为最大化风电场的功率输出,并应对尾流效应和湍流风速的影响,提出一种基于深度强化学习的风电场控制方案。该方案结合有模型与无模型的控制方法,并整合至基于Actor-Critic架构的深度确定性策略梯度强化学习网络中。在控制精度方面,采用改进的Jensen尾流模型,通过考虑时间延迟,提升了尾流效应的精确性,并有效捕捉了风电场长期功率输出。仿真结果表明,相比于传统单纯的有模型或者无模型方法,所提方法有效提升了风电场的最大功率输出,同时在保证控制精度的基础上,显著降低了训练时间和计算资源消耗,从而提升了控制策略的整体性能。The power capture capability of wind farms is constrained by various factors.To maximize the power output of wind farms and address the impacts of wake effects and turbulent wind speeds,this paper proposes a wind farm control scheme based on deep reinforcement learning.This scheme combines both model-based and model-free control methods and integrates them into a deep reinforcement learning deep deterministic policy gradient network with an Actor-Critic architecture.In terms of control accuracy,Jensen wake model consider time delay is adopted to enhance the precision of wake effects and effectively captures the long-term impact on the wind farm's power output.Simulation results show that,compared to traditional model-based or model-free methods,this scheme significantly increases the maximum power output of the wind farm while maintaining control accuracy,and significantly reduces training time and computational resource consumption,thereby improving the overall performance of the control strategy.

关 键 词：风电场控制 最大化风能捕获 深度强化学习 无模型控制 有模型控制 神经网络 

分 类 号：TM614[电气工程—电力系统及自动化]