Residential user load curve classification method considering data imbalance

Abstract

Due to the diversity and randomness of users' electricity consumption behaviors,the imbalance of load data classes is increasingly obvious. Traditional load curve classification technologies have become ineffective to deal with the im-balanced class problem of data. Therefore,an algorithm combing improved K -means with long short term memory (LSTM) neural network and convolutional neural network (CNN) classification model is proposed. Firstly,to improve the classifica-tion accuracy of the K -means on imbalanced data,a method of relative k-nearest neighbor density peaks (RKDP) based on the density peak clustering algorithm (DPC) is proposed to select the initial clustering centre of K -means. Secondly,in order to improve the performance of RKDP- K -means in processing high-dimensional load data,an au-to-encoder based on LSTM is used to extract load characteristics from high dimensional data,and com-bined with RKDP- K -means to obtain accurate load profiles labels. Finally,based on LSTM neural network and CNN,load characteristics were extracted to construct load curve classification model to realize the classification of large-scale load curve. Different algorithms are employed to classify Ireland smart meter data set and London load data set. The results show the proposed algorithm is more effective and practicable in large-scale load curve classification.

Keywords

• load curve classification
• imbalanced data
• improved k -means
• auto-encoder
• long short term memory (lstm) neural network
• convolutional neural network (cnn)