Privacy-preserving Distributed Probabilistic Load Flow

Mengshuo Jia,Yi Wang,Chen Shen,Gabriela Hug
DOI: https://doi.org/10.1109/tpwrs.2020.3022476
IF: 7.326
2020-01-01
IEEE Transactions on Power Systems
Abstract:In a multi-regional interconnected grid, the probabilistic load flow (PLF) of any region cannot be calculated individually but should consider the uncertainties introduced in other areas. Accordingly, the topologies, loads, and generations of every region are needed. Although the renewable generation data could be assumed as publicly known, some regional independent system operators (ISOs) still would not share important parameters with others. This motivates the development of a privacy-preserving distributed (PPD) PLF method. The challenge is to identify the mapping between the regional flows and uncertain power injections across regions without full information about the entire grid. The main idea of this paper is to respectively calculate the coefficient matrix and constant vector of the mapping: for the former, a PPD accelerated projection-based consensus algorithm is proposed; for the latter, a privacy-preserving accelerated average consensus algorithm is leveraged. Consequently, a PLF method is derived for each ISO to analytically obtain its regional joint PLF in a distributed way without sharing parameters — the key contribution of this paper. Experiments on the 118- and 1354-bus systems demonstrate that this method can generate the same results as the corresponding centralized method, and has satisfactory accuracy compared with frequently used PLF methods.
What problem does this paper attempt to address?