The Application Research Of Improved Particle Filter Algorithm To Integrated Navigation

At the end of 1980s, as the widely study of GPS technology, it was used in the system of vehicle location and navigation soon. While, there are still some existing problems with GPS location and navigation system, such as low dependability and sometimes not enough higher precision, on the other hand, the location precision of vehicle GPS receiver is also affected by low Secondary planet signal and road environment. Dead Reckoning system, which is made up of gyroscope and odometer, has better independence. Unfortunately, since it does not have good enough location precision after a little long time, it cannot be used independently. GPS/DR combination system can bring into full play advantage of theirs, especially for the DR system made up with piezoelectricity crystal speed gyroscope and odometer, has many advantages, such as small, high precision and low cost etc. GPS/DR combination orientation system is the direction of development of city vehicle navigation.Firstly,this project has studied GPS/DR vehicle location system, and then introduces particle filter theory based on the analysis of non-linear filter. Particle filter is a Mont Carlo algorithm based on recursive propagation. A particle set, which is randomly sampled from probability function and has corresponding weights, is introduced to approach the posterior distribution. Therefore it can handle nonlinear and non-Gaussian problems without any limits. Though the particle filter has so many advantages, there are still some existing problems, such as sample impoverishment, low performance and bad real-time performance.Since the above disadvantage of particle filter, some improved algorithms, such as auxiliary particle filter, U-particle filter and Strong Tracking particle filter has been presented in this dissertation. These algorithms have been analyzed and illuminated separately, and finally are applied to integrated navigation model. The simulation result show that the reliability of these algorithms is greatly increased compared to the general particle filter algorithm.