基于SSA-BiGRU-CNN神经网络和波动数据修正的电动汽车短期负荷预测模型  

A Short-Term Load Forecasting Model for Electric Vehicles Based on the SSA-BiGRU-CNN Neural Network and Fluctuation Data Correction

作　　者：张钰声 曹敏 雷宇 李龙 ZHANG Yusheng;CAO Min;LEI Yu;LI Long(Electric Power Research Institute of State Grid Shaanxi Electric Power Co.,Ltd.,Xi’an710005,Shaanxi,China)

机构地区：[1]国网陕西省电力有限公司电力科学研究院,陕西西安710005

出　　处：《电网与清洁能源》2025年第2期67-74,共8页Power System and Clean Energy

基　　金：陕西省自然科学基础研究计划(青年项目)2022JQ-534;国网陕西省电力有限公司科技项目5226KY220002。

摘　　要：为提高区域级电动汽车负荷预测精度,考虑了历史负荷数据自身的内在联系以及天气因素所带来的波动影响,提出一种基于麻雀搜索算法的双向门控循环单元(bidirectional gaterecurrentunit,BiGRU)-卷积神经网络(convolutional neural network,CNN)的电动汽车短期负荷预测模型。构建BiGRU-CNN模型,并应用麻雀搜索算法(sparrowsearch algorithm,SSA)对BiGRU神经网络参数进行优化;利用BiGRU神经网络充分学习历史负荷数据的前、后向联系,采用CNN对历史负荷数据进行局部优化,并通过全连接层进行预测;考虑到天气数据内部规律性不强,采用BiGRU-CNN神经网络对天气数据所带来的负荷波动进行误差预测和修正。以陕西某区域电动汽车充电站为例,分别预测预见期为4 h和24 h的电动汽车负荷,实验结果表明,所提模型无论在工作日还是双休日都具有很高的预测精度,验证了所提方法的有效性。To improve the accuracy of regional electric vehicle(EV)load forecasting,a short-term load forecasting model for EVs based on the Sparrow Search Algorithm(SSA)-optimized Bidirectional Gated Recurrent Unit(BiGRU)and Convolutional Neural Network(CNN)is proposed.This model takes into account the intrinsic relationships within historical load data and the fluctuation impacts caused by weather factors.The BiGRU-CNN model is constructed,and the SSA is applied to optimize the parameters of the BiGRU neural network.The BiGRU neural network is used to fully learn the forward and backward relationships in historical load data,while the CNN is employed to perform local optimization on the historical load data,and predictions are made through fully connected layers.Considering the weak internal regularity of weather data,the BiGRU-CNN neural network is used to predict and correct the load fluctuations caused by weather data.Taking an EV charging station in Shaanxi as an example,the EV load is forecasted for a lead time of 4 hours and 24 hours.The experimental results show that the proposed model has high forecasting accuracy on both weekdays and weekends,verifying the effectiveness of the method.

关 键 词：电动汽车 负荷预测 双向门控循环单元 卷积神经网络 麻雀搜索算法 

分 类 号：TM744[电气工程—电力系统及自动化]