The law of user power consumption cycle consuming is complex and changeable,the correlation between labels is un-stable,the classification accuracy of enterprise power consumption is low,and the time complexity is large.In order to improve the classification accuracy of power consumption cycle labels,a power consumption cycle label classification method considering matching deviation score is proposed.According to the offset of power consumption prediction,the data of each dimension of power users are collected,and the power consumption of users is predicted by calculating the identity attribute weight of power users.The calculation method of matching deviation score is adopted to obtain the time series of power consumption.According to the initial characteristics of power users,the industry labels of power users are preliminarily matched.Combined with the power consumption time series matching process,the matching of power consumption time series is completed.Based on the objective function of user characteristics,the movement of label position in the sorting process is evaluated.Combined with the matching deviation score calculation results of power consumption time series,a power consumption cycle label algorithm is de-signed to achieve the power consumption cycle label.The experimental results show that the proposed method can label the in-dustry according to its power consumption,and improve the industry classification accuracy and the matching degree between labels and industries to more than 90％and 95％.When the number of iterations reaches 30,the time complexity is 2.94 s,which improves the classification accuracy of power consumption cycle labels.
万方数据
维普
