Research On Linearization Filtering Method Of Observation Data Of Nonlinear Observation Equation

In the past few decades,the research of target tracking has made great progress,and has a good prospect and application prospects.Target tracking has been widely used in military,satellite,transportation and other fields.From Kalman Filter(KF),Extended Kalman Filter(EKF)and Particle Filter(PF),to a variety of improved methods based on classical filtering,there are more and more filtering methods for target tracking.Most of the target tracking field uses some classical filtering methods: KF,EKF,Unscented Kalman Filter(UKF),PF and other methods.In this paper,by studying these classical filtering algorithms,a nonlinear dynamic system filtering method combining Kalman filtering is proposed.Because of the complexity of nonlinear dynamic systems,a number of filtering methods,such as EKF,UKF and PF,are proposed for nonlinear dynamic systems.In this paper,the simulation and theoretical analysis of EKF,UKF and PF algorithms are carried out.It is found that these algorithms can effectively filter the nonlinear dynamic systems,but there are some problems,such as the low linearization accuracy of the EKF algorithm and the need to calculate the complex Jacobi matrix.The UKF algorithm is complicated and filtered.The problem of wave divergence or even distortion is that the PF algorithm has the problem of large computation and degeneration of weights.For this reason,this paper proposes an algorithm of Kalman filtering using the local linearization of observation data,which transforms the nonlinear filtering problem into a linear filtering problem.The main research ideas are as follows: for the trajectory generated by the target in the nonlinear dynamic system,the nonlinear observation data(distance and angle)of the trajectory are extracted,and the above distance and angle data are locally linearized;the distance and angle data after linearization are filtered by KF;the distance and angle after the filtering are obtained.The degree is mapped into the target trajectory.In this paper,this method is called New Trajectory Filter(hereinafter referred to as NTF algorithm).Considering the general radar target tracking problem,the NTF algorithm is applied to the constructed radar target tracking scene,and the error and time performance of the EKF algorithm,the UKF algorithm and the PF algorithm are compared.The simulation results show that the proposed NTF algorithm can effectively suppress the filtering divergence and reduce the error of the state estimation,and its performance is estimated.It is obviously superior to the EKF algorithm and the UKF algorithm.The running time of the algorithm is much lower than that of the PF algorithm when the estimation performance is the same.The NTF algorithm can be well applied to the target tracking of a single motion model,but the target is often maneuvered in the actual movement because of the influence of various conditions.So the motion of the target can't be described only by a model,and it is better to adopt different motion models in different stages of motion.The movement of a target.The NTF algorithm can track the target directly without interacting with the model set,and compare it with the interacting multiple algorithm.The simulation results show that the NTF algorithm has higher tracking accuracy and less time consumption compared with the interacting multiple model algorithm.