The Study Of Fusion Filtering Methods For The System With Correlated Noises

In recent years,sensor networks have been widely used in the fields of industrial,agricultural,transportation,military and national defense.Multi-sensor information fusion technology for handling sensor network observations have also been developing rapidly.Multi-sensor network and information fusion technology have become a focal research problem point in the field of information technology.Due to the complexity of the real working environment of the sensors or the targets,the system will inevitably be affected by the correlated noises.The correlations of noises can be summarized as follow:(1)the process noise and the measurement noise are cross-correlated;(2)the measurement noises between different sensors or different time are auto-correlated.These correlations will significantly increase the amount of calculation such that the processing information fusion can be obtained the the optimal state estimation of the target.For the problem of the correlated noise,the existing fusion algorithms have a variety of restrictions,such as the algorithm was only for one kind of correlated noise or a specific step correlation was considered in the study.But the more general scene that exists multiple correlated noises and finite step correlation at the same time still have a lot of problems to deal with.To solve the above mentioned problems,this thesis will be divided into the follow three parts to study:(1)Two types of correlated noise are both limited step correlated in single sensor system.In this part,we mainly adopt the idea of iterative orthogonal transformation to rewrite the observation equation to remove these two kinds of correlated noises.Then the equivalent pseudo measurement will be used as the input of the classic Kalman filter.(2)Two types of the correlated noise are both one-step correlated in Multi-sensor system.A low-dimension filter method based on sequential filtering is proposed.Under the premise of guaranteeing accuracy,it reduce the complexity of matrix operations.And it will also ensure the real-time estimation of the system.(3)Two types of the correlated noise are limited step correlated in Multi-sensor system.The more general noise-related fusion estimation problem is solved by using two methods which are centralized filter algorithm and sequential filter algorithm.These two algorithms can achieve the same accuracy.Compared with the traditional sequential filtering,our algorithm has higher calculation accuracy and superiority.Since the formula of the above three parts are strictly deduced in the sense of linear minimum mean square error,the fusion estimate of system state is optimal.