Improved Algorithm Of Unscented Kalman Filter And Its Application In GPS/INS Integrated Navigation

Integrated navigation technology has developed rapidly in today's society,and has attracted the attention of more and more countries and research scholars.There are a variety of different combinations of navigation technologies,each with its own advantages and disadvantages.Based on the understanding of the common coordinate system,coordinate transformation,basic principle of INS and information fusion technology in GPS/INS integration.This thesis focuses on the unscented Kalman filter(UKF)algorithm and its application in GPS/INS integrated navigation.An improved unscented Kalman filter algorithm is then proposed.The main contents and contributions of the thesis include:(1)For the UKF filtering with large number of sigma points,a marginalised UKF(MUKF)filtering based on the marginalised unscented transform(MUT)is proposed.The new algorithm is applied to the GPS/INS integrated navigation,and compared with the UKF filtering algorithm.The performance of the new algorithm are analyzed in terms of the positioning accuracy and algorithmic computational efficiency.The results show that the MUKF and UKF are equivalent,while the MUKF could improve the computational efficiency of GPS/INS integrated system.(2)The existence of gross errors will inevitably affect the estimation results.Aiming at the problem that the possible gross error observations may existed in the GPS/INS integrated navigation system,which may affect navigation accuracy,a robust MUKF(RMUKF)algorithm based on adaptive factor is proposed.Therefore,the RMUKF makes the MUKF have the ability to resist gross error.(3)In order to verify the effectiveness of the RMUKF filtering algorithm in practical applications,the RMUKF filtering algorithm is applied to a field test experiment.The vehical experiment was carried out in Qingdao,Shandong Province.The experimental track runs through various terrains of the city,such as high-rise buildings and roads with dense trees on both sides.The results show that a more effective and reliable positioning results could be obtained by RMUKF in contrast to MUKF algorithm.