摘要
针对大规模电动汽车参与电网调峰的调度问题,文章提出一种考虑用户参与度的电动汽车集群优化调度策略。其采用多层感知器神经网络预测电力负荷并获取负荷峰谷差;利用大量车辆信息并考虑车主意愿训练基于卷积神经网络的电动汽车集群,对车辆调峰参与情况进行集群分类,并快速确定参与调峰的总电量;构建同时考虑调峰效果和用户经济收益的优化目标,采用改进粒子群算法优化集群参与调峰的功率。最后,以县级地区为例,在Matlab平台对所提出的方法进行验证,各类电动汽车集群分类结果准确率均高于90%,同时通过电动汽车集群充放电降低了负荷峰谷差,证明了电动汽车集群方法及削峰填谷调度方案的有效性。
To solve the scheduling problem of large-scale electric vehicles participating in peak shaving,this paper presents a clusterbased optimal scheduling strategy for electric vehicles considering user participation.Multilayer perceptron neural network is adopted to predict power load and obtain peak difference.The electric vehicle cluster classification network based on convolutional neural network is trained by using a large number of vehicle information and considering vehicle owner intention to classify the vehicle peak shaving participation,and quickly determine the total electricity involved in peak shaving.Considering both peak shaving effect and user economic benefits,an improved particle swarm optimization algorithm is proposed to optimize the power participating in peak shaving.Taking the county area as an example,the proposed method is verified on Matlab platform.The accuracy of all kinds of electric vehicle clusters is higher than 90% and the peak difference between peak and valey load is reduced through the charge and discharge of electric vehicle clusters,which can verify the effectiveness of the electric vehicle clustering method and the scheduling scheme of peak shaving and valley filling.
作者
韩妍
丁惜瀛
程锟
李晓东
HAN Yan;DING Xiying;CHENG Kun;LI Xiaodong(Shenyang University of Technology,Shenyang,Liaoning 110870,China)
出处
《控制与信息技术》
2021年第6期51-56,共6页
CONTROL AND INFORMATION TECHNOLOGY
基金
国家电网公司科技项目(2019YF-01)。
关键词
电动汽车集群
负荷预测
卷积神经网络
削峰填谷
优化调度
多层感知器神经网络
electric vehicle clusters
load prediction
convolutional neural network
peak cutting and valley filling
optimal scheduling
multilayer perceptron neural network