Research On Adaptive Kalman Filter And Fault-Tolerant Algorithm Used In In-Vehicle Integrated Navigation System

Integrate navigation is the main form of In-Vehicle Navigation System. This paper mainly does research on SINS/OD and SINS/GPS. Integrate navigation often uses Kalman filter which requires the correct system model and measurement noise covariance. However, the environment is so complex that we have to do research on adaptive Kalman filter to adapt to the change of environment. Meanwhile, it often contains outliers in odometer and GPS surveying. So, this paper also does some research on inspecting and correcting outliers.SINS/OD uses the velocity integrate form. The observed quantity of the SINS/OD is the velocity's differences between SINS and odometer, and the SINS/OD also adopts a closed kalman filter to estimate and correct the errors. In order to adapt to different environment, this paper adopts innovation sequence to estimate the measurement noise covariance of odometer on line. Base on the analysis of velocity surveying and how outliers occur, to the question of the observation outliers of the odometer, SINS/OD inspect outliers based on the property of the innovation sequence and assign weight to correct it. Experiment shows that fault-tolerant adaptive filter can restrain observation outliers efficiently, that improve the precision of integrated system.SINS/GPS adopts the tightly coupled form, that is, using pseudorange and pseudorange rate as observations. This paper adopts adaptive Kalman filter with forgetting factor. The vehicle motion constraints enhance the observation of the integrated navigation when satellite is outage.The dimension of measurement vectors is huge in tightly coupled form. In this paper, a sequential Kalman filter is proposed to reduce the computational complexity. Some experiments are carried out and the results show that the proposed algorithm advances the performance of SINS/GPS integrated system. Because of the various error resources in GPS surveying and the influence of the environment, it often contains outliers in pseudorange and pseudorange rate. Base on the studies on chi-square test, this paper adopts dual-states chi-square test to inspect GPS outliers and design a method which efficiently resistant to outliers in sampling data.In order to confirm the stabilization of SINS/OD and SINS/GPS integrated system, this paper analyses it base on the theory of PWCS and finds out some useful conclusion.