| Summary: | 碩士 === 國立雲林科技大學 === 電機工程系 === 106 === In this paper, we use an unscented Kalman filter for nonlinear motions in an interacting multiple model algorithm to track maneuvering target. The unsented transform (UT) is a method for calculating the statistics of a random variable which undergoes a nonlinear transformation. Consider propagating a random variable through a nonlinear function, and assume that the nonlinear function has mean and covariance. To calculate the statistics of the nonlinear function, we form a matrix of sampling points with corresponding weights. These sampling points are propagated through the nonlinear function and the mean and covariance are approximated using a weighted sample mean and covariance of the posterior sampling points. Through combining the new mean and covarianc of these new sampling points, the state estimation of the nonlinear system can be obtained to achieve accurate nonlinear maneuvering target tracking.
When the measured value of the estimator becomes inaccurate due to the influence of high Gaussian distribution noise, the unscented Kalman filter based interacting multiple model (IMM-UKF) is less affected by noise interference than the Kalman filter based interacting multiple model (IMM-KF). Although IMM-UKF is not comparable to the anti-interference ability of the unscented particle pilter based interacting multiple model (IMM-UPF) for Gaussian noise, IMM-UKF owns streamlined computing architecture and computing process. The completion time of the IMM-UKF is much shorter than that of the IMM-UPF which increases the simulation time due to the number of particles. IMM-UKF is more suitable for applications that do not have sufficient time. According to the simulation results of this paper, the estimated time which spent on the IMM-UKF is abou 35 times faster than that of the IMM-UPF.
|
|---|
