An Adaptive Kalman Filter for SINS/GNSS Integrated Navigation with Inaccurate Process Noise Covariance Matrix Coefficient

Adaptive state estimation with inaccurate process noise covariance matrix (PNCM) has always been a challenge in the SINS/GNSS integrated navigation due to the small magnitude of the PNCM. In this paper, this problem is partially solved by estimating the coefficient of the PNCM. The nonadjacent state transition model is firstly established to get a larger equivalent PNCM, whereby the coefficient can be exactly estimated by using the variational Bayesian method. A sample screening technique by dimension is then proposed to further improve the estimation accuracy of the PNCM coefficient. Simulations and semi-physical experiments validate that the proposed filter can achieve more accurate estimates of the PNCM and navigation error state than the existing state-of-the-art adaptive Kalman filters if the power spectral density ratio of the gyroscope to the accelerometer is known.