Research On Adaptive Square CKF Method For Target Tracking

Target tracking is one of the indispensable processing module in the modern fire control,command and control system,linking to data processing.its essence is a hybrid system state estimate problem,Namely using the discrete sensor observation to estimate the goals contimuous state,filtering the random noise and solving the target motion elements.Maneuvering target tracking,target dynamic models are usually modeled in the Cartesian coordinate system,however,the radar measurements are expressed in polar form,including the distance,azimuth,elevation between the target and radar,thus,the target tracking becomes a nonlinear estimation problem.The nonlinear filtering algorithms which are usually used by us are extended Kalman filter algorithm(EKF)or unscented Kalman filter(UKF),however,EKF has to calculate the Jacobin matrix,has a low precision one-order linearization and can not be applied to the strongly nonlinear systems,which limites its application.The UKF does not have rigorous mathematical derivation,its requires filtering matrix factorization and inverse operation during the iterative process,which is diffcult to ensure the positive definiteness of the state covariance matrix form,and can not guarantee the availability of the RMS,and when the state dimension is higher,the filtering accuracy is lower,all of these limit its application.Cubature Kalman filter(CKF)is a new nonlinear filtering method,which uses the integral criterin of non-product form,its core idear is translate the aim integraion to the form of polar coordinates,and can be extended to the square root cubature Kalman filter(SCKF),which avoids the matrix decomposition and inversion operation and improves the filtering stability.with the development of the science and technology,the maneuvering target becomes more complicated,when the statistical properties of the target noise model is unknow or inaccurate or the state in one directin is mutation,SCKF algorithm can not be used to inhibit the loss of precision or divergent.The following aspects are studied to solve the problem.(1)Maneuvering target motion modeling(2)Decribes several common target tracking filter method and focus on the square root CKF algorithms.(3)Studies the Sage-Husa algorithms and derives the adaptive square root CKF filtering methods,which can be applied to nonlinear systems,for the unknown of state noise and measurement noise statistical properties.(4)The strong tracking algorithm is studied to solve the problem that the theoretical model and the actual movement dismatch bacuse of the target state mutaion.and the squareroot CKF is applied to the framework of the strong tracking algorithm and prove its feasibility.(5)The interactive multi-model algorithms is studied to solve the problem that the target run in more one model,and the square root CKF algorithm is used to the interactive multi-model algorithm.