A Bayesian Approach for Online Inertia Estimation of Synchronous and Nonsynchronous Generators
Kai Liu, Yijun Xu, Wei Gu, Jiacheng Ge, Shuai Lu, Lamine Mili, Chao Shen
Abstract
To enhance the frequency stability in modern power systems with low inertia, several techniques for virtual inertia have been devised. Therefore, for comprehensive online inertia monitoring today, it is imperative to consider both actual and virtual inertia. To address this problem, this article introduces a method for simultaneously estimating inertia and virtual inertia in power system dynamic models using a Bayesian inference-based layered adaptive importance sampling (LAIS) technique. This approach does not have any linear assumptions, rendering it well-suited for complicated power system models dominated by power electronics. Its two-layer design that integrates parallel Markov chain Monte Carlo (MCMC) in the upper level and multiple importance sampling (MIS) in the lower level ensures highly accurate estimations even with bad Bayesian priors. Simulation results underscore the method’s outstanding performance in online inertia estimation for both synchronous and nonsynchronous generators.