Research On Maneuvering Target Tracking Technology Under Different Noise Conditions

With the continuous enhancement of target mobility in modern electronic warfare,the requirements of target tracking performance are increasing and the maneuvering target tracking has also become a popular research subject in the field of target tracking.In addition,it is assumed that the noise in the tracking system model obeys different distributions,which will affect the tracking filter and further affect the performance of maneuvering target tracking algorithm.Therefore,based on the airborne radar,this thesis does research in the maneuvering target tracking technology under Gaussian noise and non-Gaussian noise respectively.The concrete research contents are as follows.(1)The maneuvering target tracking problem without clutter interference under Gaussian noise is studied.Due to the randomness and unpredictability of the target maneuvering,it is impossible to build the only one motion model that always be matched with the motion state of the maneuvering target in tracking process.To solve this problem,the classical interactive multiple model(IMM)algorithm is studied.Then the IMM algorithms based on extended Kalman filtering(IMM-EKF)and square root cubature Kalman filtering(IMM-SRCKF),respectively,are investigated because of the nonlinear system.Finally,the simulation experiments are implemented and the experimental results verify that both algorithms can realize the tracking of maneuvering targets.Considering the tracking performance and the execution efficiency of the algorithms,IMM-EKF algorithm is more suitable for practical engineering applications.(2)The maneuvering target tracking problem with clutter interference under Gaussian noise is studied.Considering the problem of how to interconnect the measurements in tracking gate and the existing trajectory in clutter,the probabilistic data association(PDA)algorithm is introduced.It is combined with IMM algorithm to solve the maneuverability problem,and then the IMM-PDA algorithm based on EKF is proposed due to the nonlinear system.Two groups of simulation experiments with different clutter densities are set to validate the algorithm.The experimental results display that IMM-PDA algorithm based on EKF can track maneuvering targets in clutter.When the target is maneuvering,the algorithm has strong adaptability and good tracking performance.(3)The problem of maneuvering target tracking under time-invariant and non-Gaussian noise is studied.To handle the difficult problem that the process noise and measurement noise have heavy tail characteristic during radar tracking of maneuvering targets,the nonGaussian noise is modeled as student’s t distribution determined by the degree of freedom and scale matrix,and the student’s t extended Kalman filtering(STEKF)algorithm is elicited which can be applied for nonlinear systems.Combined with the IMM algorithm,the IMMSTEKF algorithm is proposed and verified.The simulations show that the IMM-STEKF algorithm can track the maneuvering target stably under time-invariant and non-Gaussian noise.Then,the comparison experiments with IMM-EKF and IMM-SRCKF algorithms are set.The results show that the tracking performance of IMM-EKF and IMM-SRCKF algorithms will decrease under the condition of time-invariant and non-Gaussian noise,and the IMM-STEKF algorithm has higher tracking accuracy.(4)The maneuvering target tracking problem under time-varying and non-Gaussian noise is studied.To deal with the problem that non-Gaussian noise is time-varying due to the uncertainty of outlier and the randomness of target maneuvering during the actual radar detection,the interactive multiple model-variational Bayesian(IMM-VB)is proposed in which process and measurement noise are non-Gaussian noises.The algorithm is improved on the basis of the IMM-VB algorithm with time-varying and non-Gaussian noise only for measurement.It uses VB method to estimate the parameters-degree of freedom and scale matrix in student’s t distribution obeyed by non-Gaussian process and measurement noise to minimize the KL divergence of the approximate joint posterior probability density and the real joint posterior probability density,so that the student’s t distribution under the estimated parameters is closer to the student’s t distribution obeyed by the real noise at each instant.Then,it is combined with the IMM algorithm and the experiment is conducted.Based on the experiment results,it is proved that IMM-VB can realize the adaptive tracking of maneuvering targets under time-varying and non-Gaussian noise.Finally,the comparative experiment is executed by using IMM-STEKF and IMM-VB algorithm.Theoretical analysis and simulation results indicate that the IMM-VB algorithm can reduce the maneuvering target tracking error under the condition of time-varying and non-Gaussian noise.