摘要
在区域用电量短期预测研究中,普遍存在数据质量差、非线性特征复杂及传统模型动态适应性不足的问题,为此提出一种基于CNN-LSTM的区域用电量短期预测方法。首先,对用电数据展开均值插补,填补缺失值,检测并修正用电数据中的异常值,提高用电数据整体质量;其次,采用EMD算法分解用电序列获得IMF分量,通过K-means聚类算法对IMF分量展开聚类分析,构造新的时序序列;最后,将时序序列输入CNN网络中,获取区域用电特征图,并将其作为LSTM网络的输出,输出区域用电量短期预测结果。实验结果表明,所提方法可有效提高用电数据精度,获得的用电量预测结果与实际结果高度拟合。
In the short-term forecasting of regional electricity consumption,poor data quality,complex nonlinear characteristics,and insufficient dynamic adaptability oftraditional models are common problems.Therefore,a CNNLSTM-based short-term forecasting method for regional electricity consumption is proposed.First,mean interpolation is performed on electricity consumption data to fill missing values,detect and correct outliers in the data,and improve the overall quality of the electricity consumption data.Second,the EMD algorithm is used to decompose the electricity consumption sequence to obtain IMF components.Cluster analysis of the IMF components is performed using the Kmeans clustering algorithm to construct a new time series.Finally,the time series is input into a CNN network to obtain regional electricity consumption feature maps,which are used as the output of the LSTM network to output shortterm regional electricity consumption forecasts.Experimental results demonstrate that the proposed method can effectively improve the accuracy of electricity consumption data,and the obtained electricity consumption forecasts are highly consistent with the actual results.
作者
方智淳
肖吉东
楼霞薇
吴昊天
FANG Zhi-chun;XIAO Ji-dong;LOU Xia-wei;WU Hao-tian(Marketing Service Center,State Grid Zhejiang Electric Power Co.,Ltd.Hangzhou Zhejiang310000,China)
出处
《计算机仿真》
2026年第1期97-100,376,共5页
Computer Simulation
基金
国网浙江省电力有限公司科技项目(5211YF24000B)。
关键词
聚类算法
用电量预测
仿真
EMD algorithm
K-means clustering algorithm
CNN network
LSTM network
Electricity consumption forecast
