模型预测滤波在机载SAR运动补偿POS系统中的应用

 机载合成孔径雷达（SAR）运动补偿用位置姿态系统（POS）的定位精度直接影响SAR成像的效果。为进一步提高POS的导航精度，提出将模型预测滤波（MPF）与扩展卡尔曼滤波（EKF）相结合的方法应用于POS中。该方法不需要假设模型误差为高斯过程，并能够在线实时估计并修正系统模型，有效解决了MPF算法与系统模型不完全兼容的问题。飞行试验结果表明，该方法的收敛速度和滤波精度均明显优于目前工程应用中的KF和EKF，特别是大大提高了POS的定位精度；同时该算法与线性滤波KF的计算量相当，更好地满足了工程应用对导航精度和实时性的要求。

Application of Model Predictive Filtering Method in POS for Airborne SAR Motion Compensation System

Position and orientation system (POS) has been used widely to compensate the motional error of airborne synthetic aperture radar (SAR), and its positioning precision has great effect on SAR’s imaging directly. In order to obtain higher navigating precision, a method of combining model predictive filter (MPF) and extended kalman filter (EKF) is used in POS. In this method, the model error is assumed unknown and is estimated as part of the solution, and the model error is not limited to Gaussian noise characteristics.This algorithm can be implemented on-line to both filter noisy measurements and estimate the state trajectories. The results of flight tests show that this method has better precision and convergence, especially the position one, than those of KF and EKF. Furthermore, it decreases the computation time by reducing the number of states, and satisfies the requirements of positioning precision and real time which are emphasized in project.