System frame and algorithm research based on adaptive information fusion for the integrated navigation

Due to the uncertainty of the application circumstance, it is difficult to describe the noise statistics of integrated navigation system accurately. This may cause the general Kalman filter instability and even divergence. The common solving method is to estimate covariance matrix Q and R directly by adaptive filter. But the problem is that the increase of equations will cause heavy calculation, thus real-time cannot be ensured. This paper presents a new method to estimate states by ARMA model adaptive parameters identification to solve gain K, at the same time research an anti-saturation algorithm for the error covariance. Computer simulation was carried out according to the analysis and design of the INS/GPS integrated navigation system based on information fusion, and the results show this method is very useful for improving the accuracy and calculation speed of the system.