Research On Smart Kalman Filtering Algorithm On Observable Degree Theory

In general,the target tracking is regarded as the maneuvering target state estimation.Kalman filtering(KF),as the core technology of maneuvering target tracking,its estimation performance directly determines the tracking performance,and the estimation precision and velocity largely depends on the degree of system observability,therefore,the high observable degrees(ODs)of each state components are the prerequisite of high tracking performance.In many practical engineering,the state model,observation model or noise statistical feature are partly known or even unknown.The filtering performance will be decreased or even diverged if directly filtering with inaccuracy model parameters.However,Current adaptive Kalman filtering(AKF)cannot be used to evaluate the filtering performance though it can be used to solve this problem and suppress the filtering divergence problem and it cannot determine the performance optimized degree after adaptive filtering.Therefore,the system OD which can be used to characterize the filtering performance is introduced to select the adaptive factor,however,there has no unified analytical relations between the current OD methods and filtering performance,and the external noise is not considered in the most OD definitions.In order to solve these problems,the common OD analysis methods are taken as the theory basis,the internal relation between OD and filtering is revealed,and the OD method based on Kalman filtering is redefined,finally the smart Kalman filtering method on OD analysis is defined by filtering convergence theorem.(1)The revelation of the correlation between OD and filtering accuracy.Four typical comparative methods of OD were explored,and their principles and characteristics were analyzed and compared concisely.Finally,the computation methods based on estimation error covariance(EEC)and singular value decomposition(SVD)are respectively analyzed,and the correlation between OD and filtering performance is analytically demonstrated.(2)The analysis of observable degree method based on Kalman filtering.From the perspective of linear parameter estimation,a basic observablility matrix by WLS method is constructed,and then an optimal OD is obtained through adjusting the basic OD with the aid of the relation between OD and KF accuracy.Finally,the local/state OD(LOD)and global/system OD(GOD)are redefined.(3)The smart Kalman filtering on observable degree analysis.The smart correction link is taken as the overall filtering framework with the smart OD be the basis.Thus the smart adjust factor can be optimized and selected through the supplementary condition of filtering convergence theorem and iterative analysis method.Finally,the SKF with performance evaluation based on OD can be defined.