Modified Gain Fuzzy Kalman Filtering Algorithm

Abstract

Recently, we proposed a fuzzy Kalman filtering algorithm (FKF algorithm) which is formulated by embedding a set of parallel Kalman filters inside a fuzzy inference mechanism (FIM)⁽1)-(4).This paper proposes a version of FKF algorithm that directly determines the gain of each of the parallel Kalman filters.It fuzzifies both the modeling uncertainty of the dynamic system and the quality of each measured data to determine the Kalman gains.This can largely reduce the number of the parallel Kalman filters inside the FKF algorithm and the computation requirement.Monte Carlo simulations are conducted for observation and comparison.By adjusting the Kalman gains directly, the computation load is highly reduced with only slightly change in the filter performance.