基于MIC-ResNet-LSTM-BP的短期电力负荷预测  

Short-Term Power Load Forecasting Based on MIC-ResNet-LSTM-BP

作　　者：简定辉 李萍 黄宇航 梁志洋 JIAN Ding-hui;LI Ping;HUANG Yu-hang;LIANG Zhi-yang(School of Physics and Electronic-Electrical Engineering,Ningxia University,Yinchuan 750021,China)

机构地区：[1]宁夏大学物理与电子电气工程学院,宁夏银川750021

出　　处：《计算机仿真》2024年第4期75-79,499,共6页Computer Simulation

基　　金：宁夏自然科学基金项目(2021AAC03073)。

摘　　要：电力能源的合理调度是关系民生的重要问题,而合理的电能调度离不开精准的负荷预测。为有效提高负荷预测精度,提出一种基于MIC-ResNet-LSTM-BP的短期电力负荷预测方法来预测未来1天和3天的负荷。首先,采集6维负荷特征数据,利用最大信息系数(MIC)分析各影响因素与负荷的关联程度从而进行特征选择;其次,采用残差网络(ResNet)对数据进行特征提取;然后,将重构数据输入到长短时记忆网络(LSTM)挖掘数据时序特征;最后,采用Dropout层增加模型泛化能力,通过改进BP神经网络学(BPNN)习数据特征并利用Adam优化器训练模型。将以上模型与BPNN、KNN、LSTM、LSTMBPNN作对比实验,有力验证了上述模型在负荷预测领域的精准性。Reasonable dispatch of electric energy is an important issue related to people’s livelihood,and reasonable electric energy dispatch is inseparable from accurate load forecasting.In order to effectively improve the accuracy of load forecasting,this paper proposes a short-term power load forecasting method based on MIC-ResNet-LSTM-BP to improve the accuracy of load forecasting for the next 1 day and 3 days.First,the 6-dimensional load characteristic data were collected,and the maximum information coefficient(MIC)was used to analyze the correlation between each influencing factor and the load,so as to carry out feature selection;secondly,residual network(ResNet)was used to extract features from the data;then,the reconstructed data was input into the short and long duration memory network(LSTM)to mine the temporal features of the data;finally,the Dropout layer was used to increase the generalization ability of the model,and the data features of BP Neural Network Science(BPNN)were improved and the Adam optimizer was used to train the model.Compared with BPNN,KNN,LSTM and LSTM-BPNN,the accuracy of this model in the field of load prediction is verified.

关 键 词：最大信息系数 负荷预测 残差网络 长短时记忆网络 神经网络 

分 类 号：TM743[电气工程—电力系统及自动化]