Secant-Improved State Estimation Method for Moving Target Tracking Under Video Satellite

Abstract

A video satellite has continuous imaging capabilities, which grants it great potential for tracking and monitoring moving targets. The Extended Kalman Filter (EKF) and Unscented Kalman Filter (UKF) are commonly used in the above process. However, the accuracy of EKF estimation is low, and the computational complexity of UKF estimation is high. To address the contradiction between estimation accuracy and real-time performance in mobile-target state estimation, in this paper, we propose a new Kalman Filter with a secant-approximating nonlinear function. Firstly, the truncation error mechanism in the EKF is analysed here to illustrate the limitation of the EKF in approximating the nonlinear function. Then, the paper recommended a secant method to approximate the nonlinear function, which improved fitting accuracy without excessively increasing computational complexity. In order to improve the robustness of the proposed method, an adaptive selection strategy for correction elements is designed based on the advantageous range of secant approximation. The simulation results show that, in conventional ship motion scenarios, the computational accuracy is comparable to that of the EKF. In constant-power acceleration scenarios, the target positioning accuracy was 28.6% better than that of the EKF, and the computational speed was an order of magnitude greater than that of the UKF.