Author(s): ZHANG Xiao-Juan, LI Guo-fu, ZHANG Shuang, LIU Jie, National Ocean Technology Center
Pages: 43-48
Year: 2019 Issue:  2
Journal: Journal of Ocean Technology
Keyword:  extended Kalman Filter; unscented Kalman filter; cubatrue Kalman filter; underwater target tracking;
Abstract: Aiming at the improvement of non-linear tracking accuracy of underwater target tracking, the target maneuver model is assumed to be a constant speed motion model, under the Bayesian framework, because the extended Kalman filter tracking method carries out Taylor expansion of the model at the estimated point, ignoring the second-order and higher-order terms, leading to model errors. The root mean square of filtering errors of extended kalman, unscented Kalman and cubature Kalman under the interference of Gauss noise are compared,and the running time of the three methods are also given in this paper. The simulation results show that the tracking accuracy of the cubature Kalman filter is higher than that of the unscented Kalman filter, which is higher than that of the extended Kalman filter in the nonlinear system of 5 state dimensions. A simulation example is provided for adaptive modeling and filtering of non-linear measurement systems, such as sonar for offshore targets. This paper also provides a basis for the next step of filter algorithm improvement.