基于变分贝叶斯的星载雷达非线性滤波

Spaceborne radars play an important role in early warning defense systems because of their unique advantages such as wide detection range, long distance and all-weather surveillance capability. Due to the high-speed movement of the platform and the strong nonlinear observation function, high-accuracy target tracking for spaceborne radars is difficult. In this paper, we propose a variational Bayes-based nonlinear filtering method, which transforms the nonlinear state estimation problem into an optimization problem. The analytical solution is obtained via a closed-loop iteration manner. Moreover, a pitch angle estimation method is presented using the a priori information of target height. Simulation results show that, compared with the extended Kalman filter, unscented Kalman filter, and the converted measurement Kalman filter, the proposed variational Bayes-based nonlinear filtering method achieves the best estimation accuracy.