Study On Integrated Navigation Technology Based On SINS/Beidou

When coping with practical engineering problems like stringent environment and carrier's high dynamic maneuver,compared with single navigation system,integrated navigation system delivers more prominent performance;while,filter effect of the system with information fusion as its core has direct impact on system's stability and precision.Since the integrated navigation system usually appears strong non-linear in reality,the filter fitting strong non-linear system,theoretically,is the best option for information fusion.This article is aimed at information fusion in SINS/BeiDou based integration navigation system.It boils down to the followings:1.Study on how BeiDou and SINS operate;combining the practical engineering,it analyzes where the errors come from,and followed by their errors model establishment.2.Study on combination pattern of SINS and BeiDou;by probing into operation principles of loose coupling and tight coupling,the article designs an information fusion model of SINS/BeiDou tight coupling system.3.Study on non-linear filter algorithm;we probe into theory basis of EKF,PF,APF,RPF and how they are implemented.Through analyzing both defects of particle filter and current methods of improvement,the article puts forward regularized particle filter algorithm based on adaptive likelihood distribution(ALD-RPF).It suggests increasing the common part of likelihood and prior distribution for the purpose of enlarging the effective sample size so as to improve the state estimation.Simulation comes next in non-static growth model(UNGM)as to mentioned five algorithms and with performance comparison in all dimensions.4.In simulation system of SINS/BeiDou tight coupling,EKF,RPF and A LD-RPF are respectively deployed to accomplish system information fusion.Th en the filter performances of three particle filters are under comparison and an alysis.The simulation results show that RPF and ALD-RPF outperform EKF i n terms of filter precision,while ALD-RPF,in stability and precision,outperfo rms RPF that is with no adaptive likelihood function.