Real-time estimation of time-varying inertia for non-synchronous devices using streaming dynamic mode decomposition

With the increasing integration of renewable and non-synchronous resources, the declining level of inertia poses a threat to the stability of power systems. This paper delves into the mechanism of time-varying inertia within mainstream non-synchronous devices and proposes a real-time estimation technique to estimate the constant and time-varying inertia associated with them. By applying the numerical integration method, this technique utilizes observations and their integrals over a continuous timeframe to formulate a system of equations, enabling the estimation of inertia at any given moment. Based on the linear operator analysis, the relationship between the aforementioned system of equations and linear operators is investigated to downscale the original high-dimensional system and extract the inertia. A significant addition is the introduction of the streaming dynamic mode decomposition (sDMD) algorithm, which updates linear operators based on snapshots composed of observations. Its update merely necessitates a small snapshot length, resulting in efficient data storage utilization and space conservation. Case studies on mainstream non-synchronous devices under various controls demonstrate its effectiveness and accuracy for real-time estimation of time-varying inertia. The remarkable advantage of this technique over the existing state-of-the-art method lies in its ability to operate without setting any hyperparameters.