Power users' behavior portrait based on information gain and Spearman correlation coefficient

Abstract

With the development of new technologies in power system and the implementation of flexible policies such as demand response, traditional power consumers are gradually turning into prosumers, and their power consumption habits are also evolving and changing. In this paper, the features of power users and the potential value of massive power consumption data can be described and fully utilized by portrait technology. A method of power users' behavior portrait based on information gain and Spearman correlation coefficient is proposed. Firstly, k-means clustering algorithm based on gap statistic is used to analyze the power users' consumption data. Then, considering the effectiveness and redundancy of the feature set, the adaptability evaluation coefficient is introduced. On this basis, the optimal feature subset is obtained by genetic algorithm. Furthermore, quantitative analysis is implemented to characterize the portrait of power users. Several case studies are presented to demonstrate the effectiveness of the proposed method.

Keywords

• information gain
• spearman correlation coefficient
• users' behavior portrait
• clustering analysis
• feature selection
• electricity consumption features