Nowadays, some defects exist in the power system such as power shortage during the peak time and power imbalance. The problem can be alleviated effectively when peak-valley TOU (time of use) tariff is implemented. Furthermore, the user classification is the primary task of implementing peak-valley TOU (time of use) tariff. In this paper, a power user classification strategy of multi-view clustering is proposed. It does the cluster analysis for the power data from the points of load fluctuations indicators and real-time load curves. Clustering results of load fluctuations indicators can be drawn on the basis of the characteristics of users' load fluctuations indicators in every business. Clustering results of real-time load curves can be drawn on the basis of the characteristics of users' real-time power consuming in every business. The test results show that the strategy is able to classify the power users efficiently and accurately.