Gps/ins Integrated Navigation Data Fusion Algorithm Research

Due to the strong complementarities among the various subsystems in theintegrated GPS/INS navigation system, combined by Global Positioning System (GPS)and Inertial navigation system (INS), it has a very broad application prospect and playsan important role, and it is also the focus and emphasis of research in the field of highaccuracy and reliability of navigation systems. In the fields of GPS/INS IntegratedNavigation System, the integration of various information echnologies, especially datafusion filter technology in the integrated navigation system, is one of the maindirections of the GPS/INS Integrated Navigation System.This dissertation briefly introduces the development of satellite navigation systemand inertial navigation system, and analysises the research status of the integratednavigation system data fusion algorithm. Besides, introduce the transformation betweenthe common coordinate system and coordinate system, describe the basic principles ofGPS and INS. Based on these, this dissertation simulates the two tracks of vector,researches the data of the inertial navigation system, navigation and positioning data,and generates the data source for further study of data fusion algorithm.Firstly, research the GPS/INS navigation system mathematical model,mathematical model of the loose combination of navigation system position, velocitymeasurements. Then the integrated navigation fusion algorithm is analyzed, mainly forthe extended Kalman filter algorithm, unscented Kalman filter algorithm. Compared thesimulation results and performance of the above algorithms.Due to the particle filtering existing in particle degradation problem, proposed twoways to avoid the problem, selecting the appropriate importance of the probabilitydensity function and re-sampling, while the re-sampling would cause particledegeneracy. Considering those problems, it put the importance on selecting theprobability density functions and improving the algorithm, and a new interactivemulti-mode particle filter algorithm is come up. The new algorithm takes IMM-UKFalgorithm to create importance of the probability density function, and comparing toother improved algorithm, it is closer to the posterior probability of the probability density function. Comparing to unscented Kalman filtering Algorithm,the estimateaccuracy of proposed algorithm is better.