摘要
随着风电渗透率的增加,电力电子化元件大量接入,风电场表现出的动态特性愈发复杂,传统的基于少量案例、解析的仿真验证方法面临挑战。以深度强化学习为代表的新一代人工智能在多领域的成功应用,为风电场动态参数智能校核提供了借鉴。在双馈风电场等值模型的基础上,基于深度确定性策略梯度(deep deterministic policy gradient,DDPG)算法提出了风电场动态参数智能校核知识学习模型。该模型通过大量的仿真探索并逐步得到风电场动态参数智能校正知识,实现了基于“知识”的风电场动态参数校核。最后,基于某地风电机组实测扰动数据,利用智能体习得的参数校核知识修正风电场动态行为主导参数,并与传统启发式算法进行对比,验证了所提模型的有效性。
As the penetration rate of wind power increases and a large number of power electronic components are connected,the dynamic characteristics exhibited by wind farms become more and more complex.The traditional simulation verification methods based on a small number of cases and analysis are facing challenges.The successful application of a new generation of artificial intelligence represented by deep reinforcement learning in multiple fields provides a reference for the intelligent check of the dynamic parameters of wind farms.Based on the equivalent model of doubly-fed wind farms and the deep deterministic policy gradient(DDPG)algorithm,a knowledge learning model for intelligent check of wind farm dynamic parameters is proposed.The proposed model gradually obtains the intelligent check knowledge of the wind farm dynamic parameters through a large number of simulations and learning,initially realizing the“knowledge”-based check of wind farm dynamic parameters.Finally,based on the measured disturbance data of wind turbines in a wind farm,the parameter check knowledge obtained through intelligent learning are used to correct the dominant parameters of wind farm dynamic characteristics,and the results are compared with traditional heuristic algorithms,which verifies the effectiveness of the proposed method.
作者
周庆锋
王思淳
李德鑫
刘佳琪
李同
ZHOU Qingfeng;WANG Sichun;LI Dexin;LIU Jiaqi;LI Tong(School of Electric Engineering,Northeast Electric Power University,Jilin 132012,China;Electric Power Research Institute of State Grid Jilin Electric Power Co.,Ltd.,Changchun 130000,China)
出处
《中国电力》
CSCD
北大核心
2022年第5期32-38,共7页
Electric Power
基金
国家重点研发计划资助项目(2018 YFB0904500)
国家电网有限公司科技项目(18-GW-05)。
关键词
风电场
主导参数
参数智能校正
深度强化学习
wind farm
dominant parameters
intelligent parameter correction
deep reinforcement learning
作者简介
周庆锋(1995—),男,硕士,从事人工智能及电力系统数值仿真研究,E-mail:284598457@qq.com。
