A New Adaptive Kalman Filter Based on Interval Type-2 Fuzzy Logic System

摘要

Due to the noise-sensitive characteristic of Kalman filter, the accuracy of state estimation is dramatically decreased and the divergence of the filter is even caused. In this paper, a new adaptive Kalman filter, namely interval Type-2 Fuzzy Logic-based Adaptive Kalman Filter (T2FL-AKF), is designed to overcome the problem. Based on the ratio of the actual value of the residual covariance to its theoretical value, the proposed T2FL-AKF online adjusts the measurement noise covariance by utilizing an interval type-2 fuzzy logic. Then this adjustment changes the value of the filter gain such that the state estimate is corrected. Extensive simulations were performed to validate the effectiveness of the proposed T2FL-AKF in terms of estimation accuracy. Experimental results have shown that the average variance of the proposed T2FL-AKF can be up to 10.3% lower than that of the benchmarking schemes SKF and T1FL-AKF. At last, a suggestion of future research is also mentioned.