太原理工大学学报2025，Vol.56Issue(6)：1118-1123,6.DOI:10.16355/j.tyut.1007-9432.20240403
基于互信息和IndRNN的电动汽车充电负荷预测
Electric Vehicle Charging Load Forecasting Based on Mutual Information and IndRNN
摘要
Abstract
[Purposes]Accurate load forecasting of electric vehicle charging stations is of great sig-nificance to charging stations.In the context of mutual information,an electric vehicle charging load forecasting based on mutual information and IndRNN is proposed.[Methods]First,the historical data of user charging behavior,historical data of electric vehicle charging load,and weather data were collected,and the MRMR algorithm was used to process these data.After processing,appropriate data were selected as input features.Then,the selected feature variables were input into the IndRNN model for training and predicting.Next,IndRNN is compared with traditional long short-term memory network(LSTM)and gated recurrent unit(GRU).It is found that IndRNN can better process and predict longer time series information in electric vehicle charging load prediction,and solve the problem of gradient disappearance and gradient explosion of traditional RNN.[Results]With the ac-tual charging load data of electric vehicles for verification,the experimental results show that the MAPE and RMSE indicators of the proposed method are lower than those of other methods when pre-dicting the charging load of electric vehicles,which verifies the superiority of the proposed method.关键词
充电行为大数据/负荷预测/独立循环神经网络/互信息/最大相关最小冗余Key words
charging behavior big data/load forecasting/IndRNN/mutual information/MRMR分类
信息技术与安全科学引用本文复制引用
吴康妍,王东,马垚..基于互信息和IndRNN的电动汽车充电负荷预测[J].太原理工大学学报,2025,56(6):1118-1123,6.基金项目
山西省自然科学研究面上项目(20210302123131) （20210302123131）
