Uncertainty analysis of varying inertia estimation based on Bayesian method

• Bayesian inertia estimation method (BIEM), which combines model-driven and data-driven approaches is proposed to estimate inertia and evaluate estimation uncertainty. • BIEM represents inertia in the form of a probability distribution and has learning capabilities. • Event onset time is identified from the perspective of probability. Power system inertia is important for maintaining power system frequency stability. Because of the integration of renewable energy sources (RESs) and the complexity of power system frequency response, the power system inertia response becomes intricate and uncertainties in inertia estimation increase. The Bayesian inertia estimation method (BIEM) is proposed to estimate power system varying inertia and quantify estimation uncertainty from a Bayesian statistical and modeling perspective. Firstly, Bayesian linear regression with respect to time is applied to frequency which is measured at an event such as a generator trip or load change. Then the estimated inertia constant can be represented by a normal distribution, and the distribution can be further updated through Bayesian learning. Finally, performance and uncertainty of estimation can be evaluated by the proposed indices. The influence of power imbalance magnitude and frequency spatial distribution can also be analyzed. Moreover, the proposed method’s validity can be demonstrated by implementing Kundur’s four-generator two-area system in Simulink and the IEEE 39-bus New England system in the DIgSILENT PowerFactory.