Research On Target Tracking Algorithm Based On Iterated Extended Kalman Filter

Target tracking technology is widely used in military and civilian fields,and filtering algorithms are an important part of it,so it has received extensive attention and research.The iterated extended Kalman filter(IEKF)is a filtering algorithm with superior performance.IEKF has a simple structure,high filtering accuracy,and strong robustness.This thesis focuses on corresponding research on the IEKF algorithm.The main work is as follows:(1)Several typical nonlinear filtering algorithms are analyzed,mainly including the extended Kalman filter(EKF)algorithm,the unscented Kalman filter(UKF)algorithm,and the particle filter(PF)algorithm.Then we describe the principle,advantages and disadvantages of the algorithm.Finally,the performance of the above filtering algorithm are evaluated by numerical simulation.(2)Aiming at the problems of the filtering accuracy degradation and even the filtering divergence of the IEKF in the target tracking application in the case of model mismatch and noise time-varying,an adaptive IEKF algorithm based on multiple fading factors is proposed.The algorithm uses the IEKF algorithm as the basic framework,and obtains a more accurate estimated value of innovation covariance through a limited memory innovation covariance estimator based on normal distribution,thereby improving the accuracy of the fading factors.According to the estimated covariance,the multiple fading factors are distributed to adjust each filtering channel of the one-step prediction covariance and it can enhance the anti-interference ability of the algorithm.Then,judge the filtering anomaly according to the chi-square test principle and introduce the fading factors only when the system is abnormal.It makes the introduction time of fading factors more reasonable.Finally,the parameter variables are constructed using the information of the radial distance and azimuth between the target and the observing station,and the threshold value is preseting according to the relative positioning error relationship.The relationship realizes the adaptive control of the number of iterations of the IEKF algorithm.The improved algorithm is applied to numerical simulation experiments of target tracking under model mismatch,system noise time-varying and measurement noise time-varying.The experimental results show that the improved algorithm effectively improves the filtering accuracy and its robustness is further improved.(3)Aiming at the shortcomings of IEKF algorithm in the case of non-Gaussian noise,an adaptive IEKF algorithm combined with particle filter is proposed.The algorithm uses the adaptive IEKF algorithm to generate the importance density function.Since the adaptive IEKF algorithm can adjust the number of iterations in real time according to the latest measurement information to obtain a higher accuracy state estimation,the generated importance density function will be closer to the real posterior probability density distribution,and then the PF algorithm is used to obtain the filter estimation value.The improved algorithm is applied to nonlinear system simulation experiment under non-Gaussian noise and target tracking numerical simulation experiment under flicker noise.The experimental results show that the improved algorithm has better filtering performance.