An improved adaptive filtering algorithm with applications in integrated navigation

This paper presents an adaptive filtering algorithm based on random weighting estimation method to improve the Kalman filtering algorithm's accuracy for dynamic navigation positioning. The method involves the concept of fading filtering algorithm. Theories of random weighting estimation and windowing algorithms are proposed for estimating adaptive fading factors based on innovation vectors and estimating adaptively the covariance matrices of observation noises based on residual vectors. The proposed method in this paper provides an effective solution to resist abnormal observation error and system model error. Experimental results show that compared with traditional adaptive filtering estimation, the proposed method can significantly improve navigation positioning accuracy for dynamic navigation system.