Privacy-Preserving Distributed Clustering for Electrical Load Profiling

Electrical load profiling supports retailers in identifying consumer categories for customizing tariff design. However, each retailer only has access to the data of customers it serves. Centralized joint clustering on retailers' union dataset either enables identification more types users that allows design customized retail plans, or informs whether already a sufficiently broad customer base. centralized requires confidential retailers. This may cause privacy issues among retailers, because can not do want share their information with others. To tackle this issue, we propose privacy-preserving distributed framework by developing accelerated average consensus (PP-AAC) algorithm. Using proposed framework, modify several commonly used methods, including k-means, fuzzy C-means, and Gaussian mixture model, provide methods. In way, be achieved local calculations sharing between neighboring without sacrificing privacy. The correctness, property, time-saving feature, robustness random communication failures methods are verified using real-world Irish residential dataset.