With the constant adjustment of the industrial structure in China, the users’characteristics are changing, and the users’electricity behavior gradually develops into individuation.Firstly, the discrete wavelet transform is used to extract the characteristics of user load data. Secondly, the improved fast density peaks clustering algorithm is used to cluster the users into load groups with different power consumption behaviors, and then the time distribution characteristics of the load groups are analyzed.The mutual information method is used to analyze the correlation between electricity consumption data, economy, temperature, industry key indicators and so on, and the key influencing factors are extracted. Finally, the simulation results of a typical user in an industry in a province verify the effectiveness of the proposed method.