A Kalman and Fading Memory Cofilter for Uncertain Systems Based on Self-Perception Mechanism

Abstract

A cofilter by collaboration of Kalman filter and fading memory filter improves the filter estimation performance for uncertain systems. Specifically, the influence function is utilized to quantify the influence of uncertainty on estimation performance, forming the self-perception mechanism. Then, the cofilter takes the Kalman filter as the robust lower bound and the fading memory filter as the robust upper bound and adjusts the robust parameters based on the self-perception mechanism to form an adaptive robust filter. The advantage of the proposed cofilter is that it resists uncertainty while reducing performance loss. The performance of the adaptive robust filter is analyzed theoretically using the Riccati equation and the Lyapunov equation. Furthermore, one numerical example simulation, one practice-oriented 1-degree of freedom (1-DoF) torsion simulation, and one water tank experiment are given as an illustration of the efficiency of the proposed adaptive robust filter.