A New Exponential Forgetting Algorithm for Recursive Least-Squares Parameter Estimation

Abstract: This paper develops a new exponential forgetting algorithm that can prevent so-called the estimator windup problem, while retaining fast convergence speed. To investigate the properties of the proposed forgetting algorithm, boundedness of the covariance matrix is first analysed and compared with various exponential and directional forgetting algorithms. Then, stability of the estimation error with and without the persistent excitation condition is theoretically analysed in comparison with the existing benchmark algorithms. Numerical simulations on wing rock motion validate the analysis results.