Privacy protection method of power metering data in clustering based on differential privacy

Abstract

Power companies can use the power grid big data platform to cluster analysis of power metering data, which can improve the personalized service quality of power grid companies for different users and discover the power stealing behavior of users to protect the interests of power grid companies. However, in the cluster analysis of power measurement data, the privacy information of power users may also be disclosed. To defend the privacy information of power users, the article applies differential privacy technology to cluster analysis of power metering data to avoid power users’ privacy leakage. First, the article presents the attack model that exists in the cluster analysis of power metering data. Then, the article add Laplacian noise to the power metering data to defend against attacks in the cluster analysis of attackers. Next, to enhance the data availability of noise-added power measurement data in cluster analysis, the article limits noise distance based on the results of the cluster analysis. Experiments show that method proposed in article can guarantee the privacy information of power data during the cluster analysis of power metering data, and ensure the data quality of the power metering data after privacy protection.