Accelerated Generalized Correntropy Interior Point Method in Power System State Estimation

摘要

Classical Weighted Least Squares (WLS) is a well-known and broadly applicable method in many state estimation problems. In power system networks, WLS is particularly used because of its stability and reliability in the cases that measurement noise are Gaussian. Nowadays, with the use of renewable energy sources and the migration to smart grids WLS is no more appropriate because the noises are far from being Gaussian. Recently, a novel state estimation algorithm denoted Generalized Correntropy Interior-Point method (GCIP) was presented that can deal with measurements contaminated by gross errors. Under that conditions, the superiority of GCIP is confirmed in a variety of tests. This paper presents an improved GCIP in terms of computational efficiency. The main computational burden of GCIP arises from a large dimension matrix of the correction equation. By looking into the structure of the data, a new arrangement for this matrix with lower order is presented that helps to reduce computational time remarkably. The efficiency of new method was tested with different IEEE benchmark systems.