Time Synchronization And Information Distribution In Multi-source Information Fusion

The continuous development of science and technology has led to an increase in the number of navigation systems.Different navigation systems have their own advantages and inherent defects.INS navigation solutions have accumulated time error characteristics,and GNSS signals are vulnerable to obstruction and interference,which makes it unable to provide continuous navigation.Therefore,it is necessary to use the advantages of each navigation system for data fusion in order to maximize navigation accuracy and robustness,minimize complexity,and optimize processing efficiency.This paper introduces the basic principles of KF and EKF filtering that are widely used in the field of navigation.The basic principle of INS navigation is described and the characteristics of INS navigation error which is accumulated over time are verified by simulation.The GNSS single-point navigation positioning solution and the filtering-based navigation positioning solution mathematical model are deduced.The simulation verifies that the navigation solution obtained by the filter form is more smooth and accurate compared to the single-point navigation positioning solutionThis paper studies the problem of time synchronization between systems in the field of information fusion,and uses INS/GNSS integrated navigation as an example to model its existing time synchronization problems.The reasons for the time synchronization problem between the systems and their impact on the state estimation are analyzed.For the loosely coupled INS/GNSS Integrated navigation based on the coupling of the location domain,comparing and analyzing the performance of navigation solutions which are obtained with or without compensating time synchronization errors.Whether there is a speed after compensating the time synchronization error is used as a measurement input to influence the navigation solution performance.For tightly coupled INS/GNSS integrated navigation based on the coupling of the distance domain,the performance of the navigation solutions which ate obtained with or without compensating the time synchronization error is analyzed by using the model simulation.The solution to the problem of time synchronization between INS/GNSS systems is also applicable to other integrated navigation systems.This dissertation studies the information distribution methods of the widely used decentralized federated filtering in the field of information fusion.This paper expounds the importance of information distribution method to the federal filter,introduces the existing information distribution methods and their design concepts,and points out the characteristics of federated filtering based on fixed scalar,dynamic scalar,and dynamic vector information distribution methods.On this basis,a dynamic vector information distribution method based on state observable order,system observable matrix singular value and system state covariance matrix is designed.This method can effectively improve the estimated accuracy of federated filtering and further improve the navigation performance of the system.In order to weaken the influence of system dynamics model error,this paper designs a robust self-adaptive federated filter information distribution algorithm.Finally,simulation results show that the proposed algorithm can effectively resist the influence of dynamic model error,and can improve the stability of entire federated filtering system.