Research On Integrated Navigation Method Based On Adaptive Filtering

The integration of Strapdown Inertial Navigation System(SINS)and Global Positioning System(GPS)can take the advantages of two kinds systems avoiding the speed and position accumulated over time.The in-depth and meticulous research of SINS / GPS integrated navigation method will be of great significance.In this paper,taking SINS / GPS loose combination of navigation as the background,the research focuses on data fusion algorithm of the integrated navigation process,that is,adaptive Kalman filter.We mainly focus on the SageHusa adaptive Kalman filter and the recursive variational Bayesian adaptive Kalman filter.Sage-Husa adaptive Kalman filter is able to estimate the unknown system noise Q or measurement noise R,thereby enhancing the accuracy of the filter.But when the Q value or R value is reduced sharply,filter accuracy decreases or even diverges.The recursive variational Bayesian adaptive Kalman filter does not need the previous time R value when estimating the current time R value,so it has higher stability than the Sage-Husa adaptive Kalman filter.Even so,it cannot estimate the system noise Q value in the filtering process,which will also decline the filtering accuracy.In this paper,an improved variational Bayesian adaptive Kalman filter algorithm is proposed,which is based on the fact that the Sage-Husa adaptive Kalman filter can estimate the Q value under the condition that the system noise Q is unknown.So that the filtering accuracy of proposed algorithm is improved.In this paper,SINS / GSP integrated navigation system model is established,which provides the simulation environment for adaptive filtering in this paper.Then,the Sage-Husa adaptive Kalman filter is simulated in SINS / GPS integrated navigation simulation environment.The simulation results show that the Sage-Husa adaptive Kalman filter is more accurate than the classical Kalman filter in the case that the measured noise R is unknown,but the Sage-Husa Adaptive Kalman filter will diverge when the R changing sharply.Then,the variational Bayesian adaptive Kalman filter is simulated in SINS / GPS integrated navigation system.The simulation results show that the filtering accuracy is higher than that of Sage-Husa filter and classical Kalman filter.Finally,the modified variational Bayesian filter is simulated in SINS / GPS simulation environment.The simulation results show that the accuracy of the improved variational Bayesian adaptive Kalman filter is higher than that of the unmodified variational Bayesian adaptive Kalman filter.In order to verify the advantages of improved filtering algorithm,Sage-Husa adaptive Kalman filter,variational Bayesian adaptive Kalman filter and improved variational Bayesian adaptive Kalman filter are used to deal with SINS / GPS integrated navigation of the actual vehicle data.Experiment results show that the improved variational Bayesian adaptive filtering accuracy is higher than the other two adaptive filters.The improved variational Bayesian adaptive filtering can not only preserves the advantages of variational Bayesian adaptive Kalman filter estimation to estimate the noise R value,but also improves the filtering accuracy when the system with unknown Q,which verifies the advantages of proposed improved filtering algorithm.