Research On Intelligent Navigation Algorithm Based On Multi-sensor Information Fusion

The multi-sensor information fusion technology is the processing ofmulti-source data,thus more accurate and more comprehensive understanding anddescription of the measured object,in order to make the right judgments,to providethe robustness of the system.The main research work of the paper is the research ofvehicle navigation algorithm based on the theory of multi-sensor informationfusion.At present,based on navigation technology becoming mature gradually,andenormous demands for it in all fields of society,integrated navigationtechnology hasbecome a main trend in its development.In this paper,especially in consideration ofhigh precision low cost,GPS receiver,electronic compass and gyroscope are selectedto construct vehicle-mounted GPS/DR integrated navigation system.Research onIntelligent Navigation Algorithm in the applicationof integrated navigationsystem,compares the precision of extended Kalman filter,Particle filter and theImproved particle filter.The major contents of the paper were summed up as follows:Firstly,the paper introduces the development of vehicle navigation technologyand multi-sensor information fusion technology.Through the study of the domesticvehicle navigation technology,it is shown that the navigation based on multi-sensorinformation fusion is a research direction in future.Secondly,it studies the basic positioning principle and error sources of GPSsystem and DR system,then discusses the strengths and weaknesses of GPSor DRalone positioning by experiment,and identifies the superiority of the GPS/DRintegrated positioning.Then according to the actual situation of system,establishedthe mathematical model of the integrated navigation system.Again,the paper mainly studies multi-sensor information fusion algorithm.In the thesis,we analyses and designs EKF fusion model,writing the matlab simulationprogram based on the figure.But in comparison with classical kalmanfiltering,extended kalman filtering technology is more suitable for dealingwithnonlinear navigation systern.However, the extended kalman filter, we mustlinearize theintegrated navigation system, resulting in the filtering error.Finally, based on the EKF in the linearization process prone to erroraccumulation problem,the paper proposed particle filtering algorithm,andimprovednavigation accuracy.However,a common problem with PF is thedegeneracy phenomenon,in order to solve this problem,this paper has made theimprovementto the algorithm:firstly,the paper generate the importance density usingEKF with full use of the observation information,so that it is closer to the truestate;secondly,set a threshold to measure the degeneracy of the algorithm,so that it ismore reasonable for the choice of particle with power value.The simulationresultsshow that,the improved particle filter to optimize the system performanceofthe filter.