Research On The Filter Algorithm For GNSS/INS Integrated Navigation

Integration between the Global Navigation System(GNSS) and Inertial Navigation System(INS) can provide continuously and high precision navigation information(position, velocity and attitude) by combining the advantages of them. This dissertation mainly focuses on the filter theory in GNSS / INS integrated navigation system which incluses the principle of Inertial navigation system and GNSS / INS integrated navigation system 、GNSS positioning mode and the analysis of signal propagation error, the filtering algorithm of in GNSS / DR integrated navigation, the adaptive filtering algorithm about the error of function model and stochastic model in the GNSS / INS tightly integrated navigation system, robust filtering algorithm and the sequential filtering algorithms and so on. The main work and research results are summarized as follows:(1)To sum the principle of processing data about the GNSS / INS integrated navigation system, introduced the basic principles of INS navigation 、 GNSS navigation、positioning mode of GNSS and analyzed the error of signal propagation.(2)Introduce the basic principle of the standard Kalman filter, adaptive Kalman filter algorithm and robust Kalman filter model, and compare the different of the extended Kalman filter(EKF), adaptive and robust EKF algorithm in GPS / DR integrated navigation.(3)For the more observations information on multi-sensor, and low computational efficiency, this dissertation proposes an adaptive Kalman filter algorithm based on Mahalanobis distance, a judging index is defined as the square of the Mahalanobis distance of the innovation vector, which is found to be Chi-square distributed. Using the method to address the random model error and function model error in the GNSS / INS integrated navigation. The results show that the adaptive fading Kalman filter based on Mahalanobis distance not only to guarantee the accuracy but also to improve computational calculated efficiency.(4)When the observation has some gross errors, using the Kalman filter algorithm to address any noise errors may occur filtering divergence and other issues. This dissertation establishes the robust factor based on the Mahalanobis distance based on the difference between the observation and prediction, and the value of the square of the Mahalanobis distance obeys chi-square distribution. Through a method of hypothesis and testing based on statistics to detect some outliers, and to adjust the outlier through a robust factor which establishes of Newton Iterative Method to reduce the impact of outliers on the dynamic model. To test effectiveness of the algorithm in GNSS / INS tightly integrated navigation system, the results of the experiment indicate that the robust algorithm shows well robust performance and high efficiency in computations.(5)When the observation vector has many types of observation elements, using one robust factor to adjust entire observation may happen the problem of eliminate true observation and receive incorrect observation. A robust Kalman filter based on Chi test with sequential measurement update is proposed. First to de-correlate the matrix of R through the Cholesky decomposition. Then establish the Mahalanobis distance based on the difference between the individual observation vector and prediction, through a method of hypothesis and testing based on statistics to detect some outliers, and to adjust the outlier through a robust factor.This approach can not only handle outlier in part or even individual measurement channel, but also further improve the accuracy especially when a novel ordering strategy in processing the measurement elements is adopted. But the accuracy improvement should be attributed to the higher computer load.