Positioning Algorithm Of GNSS Based On Improved Particle Filter

The Global Navigation Satellite System(GNSS)is playing an increasingly important role,and it is of great significance to further improve the positioning accuracy and real-time performance of GNSS.The main work of this thesis is as follows:This thesis first introduces the basic principles of GNSS pseudorange positioning and Unscented Kalman Filter(UKF)algorithm,and establishes a corresponding model based on the application scenarios of GNSS positioning solution.Then the common error sources in the process of pseudorange measurement and the errors caused by the positioning results are analyzed,and the corresponding correction model is given.This thesis introduces the principle of the standard particle filter(Particle Filter,PF)algorithm,and analyzes its disadvantages such as particle degradation,reduced particle diversity,and large amount of calculation.Four resampling methods and the importance of different selections are studied.The effects of four resampling methods and the selection of different importance functions on the performance of particle filter are studied.This thesis introduces the implementation process of particle filter positioning solving,and proposes two improved algorithms of particle filter.Part Resampling Unscented Particle Filter(PRUPF)based on unscented Kalman filter uses UKF to guide particle sampling,which is to improve the resampling method on the basis of Unscented Particle Filter(UPF),partition the particles for incomplete resampling.The simulation results show that compared with UPF,PRUPF has improved computational efficiency.The Mean Shift Particle Filter(MSPF)based on the mean shift merges the Mean Shift(MS)algorithm with the PF,and uses the MS algorithm to optimize the particle set to make the particle distribution closer to the real state distribution.The simulation results show that MSPF reduces the number of required particles and improves calculation efficiency,but its filtering accuracy is lower than PRUPF.Finally,this thesis uses actual data to test the positioning performance of UKF,UPF,PRUPF and MSPF algorithms in two different observation environments.The test results show that the UKF has the best positioning performance in a static and open urban environment;in a dynamic partial occlusion environment in the city,the observation noise does not completely obey the Gaussian distribution,the positioning accuracy of UKF is lower than UPF,PRUPF and MSPF,while PRUPF and MSPF are Both real-time and accuracy are better than UPF.