Objectives Unscented Kalman filter (UKF) is an effective method to solve the nonlinear filtering of global navigation satellite system/strapdown inertial navigation system (GNSS/SINS) integrated navigation system. UKF shows good performance in Gaussian noise, but its performance will be seriously degraded in non-Gaussian noise, especially when the system is interfered by some heavy tail pulse noise. In order to improve the robustness of UKF against heavy tail pulse noise, we propose a maximum correntropy UKF (MCUKF) algorithm for GNSS/SINS integrated navigation system.

Methods First, based on the mechanical arrangement equations of SINS and the output information of GNSS, a nonlinear system model for the GNSS/SINS integrated navigation system was established, the key feature of this model is that the state equation is nonlinear while the measurement equation is linear. Then, a prior estimation of the state and its covariance matrix is obtained by using the unscented transform (UT) of UKF. Finally, based on the establishment of the maximum correntropy criterion, a statistical linear regression model was established using the prior estimation model and the measurement equation. Subsequently, the solution equation of the state vector and the posterior estimation of the state and its covariance matrix were solved, and the fixed-point iterative implementation steps of the MCUKF algorithm were designed.

Results The simulation results show that the performance of UKF is slightly better than MCUKF in Gaussian noise. Compared with UKF, MCUKF with a core bandwidth of 5 can improve position accuracy by 13.4% and velocity accuracy by 10.3%. Compared with adaptive Kaman filter, MCUKF can improve position accuracy by 8.8% and speed accuracy by 7.5%.

Conclusions The experimental results show that the filtering performance of MCUKF is obviously better than that of UKF at small core bandwidth, which can improve the filtering accuracy of integrated navigation system in complex environment.