Applicability comparison of different algorithms for ambient signal based load model parameter identification

Abstract

The time-varying characteristic of the power load can be tracked with the ambient signal based load model parameter identification method proposed recently. Meanwhile, the small fluctuations of the ambient signal bring higher requirement for optimization algorithm as well. The interior-point algorithm (IPA) and the activeset algorithm (ASA) are two typical traditional nonlinear constrained optimization algorithms, while the differential evolution algorithm (DEA) is an excellent heuristic search algorithm. Besides, the grid search algorithm (GSA) is a violent search algorithm, which is also an effective method considering the fact that the real load model parameters are unknown in practical power system. Based on the review of the basic idea of the four algorithms, the applicability of the four algorithms is comparatively assessed for ambient signal based load model parameter identification. Their performance is compared using ambient data from both simulation and field phasor measurement unit (PMU) in China Southern Power Grid considering the identification accuracy, calculation time, and the robustness to measurement error and algorithm parameters. Some general conclusions are drawn from the analysis for several cases.