Clustering plays an important role in the big data analysis of power grid customers. With the promulgation of China’s Data Security Law, how to give consideration to data privacy and clustering quality in power customer data clustering analysis has become a difficult point to be solved. The existing k-means clustering method based on differential privacy is difficult to balance data privacy and clustering quality. The method of adding noise to data distance is proposed. By extracting the data distance and adding noise satisfying differential privacy constraint to the distance value, a noise matrix is constructed to realize the privacy protection of the data distance. Furtherly, the kq-means clustering method based on noise matrix is designed, the concept of k nearest neighbor is introduced, and the clustering division strategy is designed. The data records are distributed to the expected intervals of several nearest central points, which reduces the clustering error caused by the accumulation of differential noise in the process of multiple iterations.Our solution can achieve both privacy and clustering accuracy of power grid customer data.