Research On Error Modeling And Compensation Of MIMU/BD/magnetometer Integrated Navigation System

For low-cost MIMU/BD integrated navigation system, it has poor observability of yaw error and is difficult to realize initial alignment. So MIMU/BD/magnetometer integrated navigation system is built to improve the system's observability of yaw error. Improved robust adaptively Kalman filter is put forward to improve the navigation accuracy in actual system use. The contributions are introduced as follow:Firstly, research on tri-axial magnetometers' error modeling and calibration method is taken. The geomagnetic field belongs to weak magnetic field. Using magnetometers to measure the geomagnetic field is easily disturbed by the external environment. The calibration of magnetometers and data preprocessing of magnetic signals are important parts for efficiently using geomagnetic field for navigation. Based on the theory of magnetic compass, theoretical analysis and Matlab simulation analysis about factors which affect the magnetic yaw error have been made. Improved median filter and wavelet analysis methods are used to process the measurement signals of magnetometers, and test is taken to prove the methods. The parameterized error model of tri-axial magnetometers is built, and the recursive least square(RLS) with forgetting factor algorithm is used to calibrate the tri-axial magnetometers. Simulation and actual test data prove that the algorithm is in good performance.Secondly, the error model of MIMU/BD/magnetometer integrated navigation system is built. Analysis of MIMU error model, BD system observation model and magnetometers observation model is made. The error model of MIMU/BD integrated navigation system and MIMU/BD/magnetometer integrated navigation system is built respectively. For actual system use, outliers and disturbance affect the navigation accuracy. So robust adaptively Kalman filter is applied. The theories of standard Kalman filter and robust adaptively Kalman filter are introduced.Lastly, the integrated navigation system is built and vehicle tests are accomplished. The tests show that adding magnetometers improves the yaw accuracy obviously. And the robust adaptively Kalman filter performs better. The static yaw accuracy is improved from 3.43 to 0.77, and the yaw accuracies when the vehicle moves are improved from 4.97°, 4.61° and 6.36° to 1.19°, 2.90° and 3.72° respectively.