An identification-based state estimation method for a class of non-linear systems

Abstract

In this paper, an online state estimation method for unmeasured states appearing linearly in deterministic non-linear systems is developed. The method is based on a recursive least-squares identification algorithm. The proposed method possesses some superior properties, mainly its simplicity and better estimation performance, compared with other typical estimation methods. The effectiveness of the method is demonstrated on case studies with a comparison to the popular extended Kalman filter (EKF). Although the method is presented for state estimation, it can be easily modified for time-varying parameter estimation.