Sage–Husa Algorithm Based on Adaptive Double Forgetting Factors

Abstract

In order to address the issues of insufficient filtering accuracy and filtering divergence that have been observed in the Sage–Husa algorithm when applied to nonlinear system state estimation, an adaptive double forgetting factor-based Sage–Husa algorithm is proposed. This algorithm builds upon the Sage–Husa algorithm with forgetting factors by introducing double forgetting factors and adaptively adjusting them using a windowing method combined with an exponential form. On the basis of ensuring the semi-positive definiteness of the process noise covariance matrix and the positive definiteness of the observation noise covariance matrix, a covariance matching technique is employed to determine whether the measurement noise statistical characteristics need to be re-updated. The results of the simulations demonstrate that the proposed algorithm enhances the accuracy of filtering and exhibits strong effectiveness and feasibility.

Keywords

• nonlinear system
• Sage–Husa
• adaptive double forgetting factors
• windowing
• exponential form