Research On Algorithm Of Information Fusion And Fault Detection For SINS/Multi-satellite Integrated Navigation System

Inertial navigation system (INS)/multi-satellite integrated navigation system is a relatively perfect system due to the navigation information of the two systems are complementary. In order to further improve the accuracy and reliability of the integrated navigation system, the algorithm of information fusion and fault detection for SINS/multi-satellite integrated navigation system is researched in this dissertation.Considering the disadvantages of the general federated kalman filter and the characteristics of SINS/multi-satellite integrated navigation system which measurement noise is time-varying, a double-adaptive federated filter algorithm is put forward in this dissertation. The algorithm can adjust the measurement noise on-line without knowing the statistic characteristics of systematic noise, meanwhile, the information distribution coefficient can be adjusted adaptively according to the geometry dilution of precision (GDOP). The simulation results of SINS/GPS/Galileo /BDS integrated navigation system show that the algorithm can improve the precision of the integrated navigation system effectively.Due to the multi-information navigation system on the airplane often works in the kinematic disturb environment, a filtering algorithm based on kinematic disturb is researched. The arithmetic uses geometry dilution of precision to adjust the measurement noise and uses the innovation of kalman filter to whole control the state covariance matrix, meanwhile, the information distribution coefficient can be adjusted adaptively according to the covariance matrix of each subsystem. The algorithm solves the problem that innovation cannot determine the adaptive factor of observed value and status parameter at the same time. The simulation results prove the solution is feasible.In order to improve the reliability of multi-information integrated navigation system, a dual-fault detection approach which fits for federated filter is proposed. Residual Chi-square testing method of the federated filtering structure is used to check the hard fault, and the optimum global state estimate of k-m is used to construct the moving residual testing function and to check the soft fault, meanwhile, the information distribution coefficient can be adjusted adaptively by the soft fault testing function. SINS/Galileo/BDS simulation results verify that the algorithm has high fault detection sensitivity to navigation system's hard fault and soft fault, and can further improve the integrated navigation system's reliability.