Recursive Parameter Estimation for Wireless Power Transfer Systems with Non-PE Data

The Recursive Least Squares algorithm with exponential forgetting (RLS-EF) is capable of effectively tracking the time-varying parameters of WPT system caused by factors such as coil spacing. However, under non-persistent excitation (Non-PE), the RLS-EF algorithm suffers from ill-conditioning, consequently causing the undesirable estimator windup. To address the parameter identification problem for WPT systems under Non-PE conditions, an improved RLS algorithm with Adaptive Variable Direction Forgetting is proposed in this paper. The effectiveness of the proposed method was validated through numerical simulations and experiments conducted in actual time-varying WPT scenarios, resulting in more stable tracking of time-varying parameters.