Aiming at the problem of user privacy leakage caused by data analysis on power dispatching operation data, this paper proposes a differential privacy analysis method for the power consumption pattern of users. We use K-means clustering method on the original time series data to obtain the optimal number of the power consumption patterns of users. We propose a differential privacy clustering analysis method, which adds Laplacian noise when generating clusters and calculating the centroid within clusters, and then, de-noises the centroid data within clusters by smoothing function to improve the data availability. The algorithm is tested on real dataset REDD. The experimental results show that the proposed algorithm not only achieves privacy protection, but also improves the availability of privacy data when analyzing the operation data of power dispatching.