Research On Information Fusion Methods For INS/GNSS/Odometer Vehicle Integrated Navigation System

The navigation and positioning of vehicles has become an indispensable part of daily life.With the promoting of vehicle navigation and positioning accuracy,in many cases,a single navigation system has been unable to meet the navigation needs of vehicles,such as INS and GNSS.To improve the accuracy of the navigation results,the main method should be used more than two single navigation systems at once.At present,INS/GNSS is the most commonly method.Not only can effectively overcome the shortcomings but make full use of advantage.At the same time,the vehicle odometer is conveniently for obtaining the speed information of vehicles.Therefore,this paper aimed at determine the fusion method of navigation information by the INS/GNSS/Odometer vehicle navigation System.Firstly,this paper research on the principle of INS,GNSS and odometer,and what causes these errors.Establishes the error models of three navigation systems,and the advantages and disadvantages of their navigation and positioning performance are analyzed.Secondly,use the federal filtering method to divide the integrated navigation system into two subsystems,and establishes the integrated navigation system equation which based on INS,and then establishes the measurement equations for the two subsystems separately.And simulate the trajectory of the car which move under different motion transitions.Using the inertial navigation solution algorithm to obtain the speed,position and attitude information of the inertial navigation system.Thirdly,this paper research on the information fusion algorithm which used in integrated navigation system.Since the vehicle navigation system is a nonlinear system,it needs to use non-linear filtering algorithm to fusion the information.This paper research on the Kalman Filter(KF)the Unscented Kalman Filter(UKF)which improved the KF,then combined the Interacting Multiple Model(IMM)algorithm with the UKF algorithm,and simulate the information fusion results.Analyze the navigation system's performance that in different measurement noise.Then process the navigation data which obtained by the subsystem by the federal IMM-UKF algorithm,and simulate the navigation information of the INS/GNSS/Odometer integrated navigation system.Finally,because of the GNSS signals are susceptible to environmental occlusion and external interference,this paper put forward a method on error correction that using artificial neural networks when GNSS signals are out of lock.Optimize the initial weight and threshold of the BPNN by the GA.And train the network when the GNSS signal is valid and predict the navigation information through the trained network when the GNSS signal is out of lock,use this method can complete the error correction effectively..