Multi-sensor Under-observed Generalized System Information Fusion Incremental Kalman Filter

With the general application of generalized system theory in mechanical computer-aided modeling,aerospace and other fields,the problem of generalized system state estimation has also received extensive attention.When it comes to state estimation,Kalman filtering is thought of,because this method is relatively well-developed,and it is also convenient to rely on computer simulation experiments.There is also a problem that needs to be considered to improve the estimation accuracy.Then the information fusion theory rises in response to the proper time and conditions.In the study of the multi-sensor information fusion state estimation problem,the weighted observation fusion algorithm with the help of the Kalman filter algorithm is a globally optimal distributed fusion algorithm,and the amount of calculation is small.This algorithm can further improve the estimation accuracy of the system.Although the Kalman filter is quite perfect,it still has shortcomings that have not been resolved.For example,the Kalman filter algorithm can only work normally under the conditions of accurate model parameters and white noise.In practice,it is difficult to guarantee such conditions,which will cause unknown errors in the system.The Kalman filtering algorithm has no ability to eliminate such systematic errors.Therefore,for the state estimation problem of under-observed singular systems with unknown parameters,this dissertation conducts related research on the incremental Kalman estimation algorithm and fusion algorithm for under-observed singular systems based on incremental observation equations,and the contents are as follows:Based on an incremental equation of linear discrete singular systems,two canonical incremental Kalman estimation algorithms for under-observed singular systems and two canonical fusion incremental Kalman estimation algorithms for multi-sensor systems are deduced respectively.Aiming at the estimation problem of the incremental observation equation with colored noise,the under-observed generalized incremental Kalman estimation algorithm with colored noise and the multi-sensor under-observed generalized fusion incremental Kalman estimation algorithm with colored noise are proposed.Simulation examples are given for the above algorithms to prove their practicability and superiority.