基于VMD的家用电器负荷特征提取与识别  

Load feature extraction and identification method of household appliances based on VMD

作　　者：吴玉刚 迟长春(指导)[1] WU Yugang;CHI Changchun(School of Electrical Engineering,Shanghai Dianji University,Shanghai 201306,China)

机构地区：[1]上海电机学院电气学院,上海201306

出　　处：《上海电机学院学报》2023年第6期317-323,共7页Journal of Shanghai Dianji University

摘　　要：负荷识别是诊断用户用电情况的重要方法,在监测、预警和故障诊断等方面有着重要的意义。针对传统特征提取方法容易出现负荷信息丢失的问题,采用变分模态分解(VMD)算法,将方差贡献率最大的本征模态分量(IMF)对应的能量熵作为特征,结合卷积神经网络(CNN)模型完成负荷识别。实验结果表明:此负荷识别方法识别效果良好,识别率为95.9459%,为电力部门开展相关研究提供了可靠的依据。Load identification is an important method for studying userselectricity consumption,and has important significance in monitoring,early warning,and fault diagnosis.To solve the loss problem of load information in traditional feature extraction methods,the variational mode decomposition algorithm(VMD)is applied to use the energy entropy corresponding to the intrinsic mode component(IMF)with the largest variance contribution rate as a feature,and load identification is completed by combining with convolutional neural network(CNN)model.The experimental results show that the load identification method has a good identification effect,with a recognition rate of 95.9459%,which provides a reliable basis for the power department to carry out related research.

关 键 词：负荷识别 变分模态分解算法 卷积神经网络 

分 类 号：TM561[电气工程—电器]