临近空间飞行器具有机动特性复杂、运动轨迹多阶段性等特点,在目标跟踪的过程中,易出现由于系统模型误差较大导致跟踪精度降低、滤波发散的问题。针对该问题,在容积卡尔曼滤波的过程中加入衰减因子,通过衰减记忆的方法补偿模型误差;同时,提出了一种实时辨识容积卡尔曼滤波衰减因子的方法,达到自适应跟踪的目的。仿真结果表明:衰减记忆容积卡尔曼滤波算法能够很好地解决模型失配问题,自适应算法实时对衰减因子赋值,避免了衰减因子取值的困难,可以达到更好的跟踪效果。

Near space vehicle has complex maneuvering and multi-phased trajectory characteristics. In the process of target tracking, lower tracking accuracy or filtering divergence problem is prone to happen due to the large model error. For this problem, added attenuation factor is introduced into the cubature Kalman filter to compensate the model error by attenuating memory method. Also a real-time identification method for cubature Kalman filter attenuation factor is proposed to achieve adaptive tracking purpose. The simulation results show that attenuation memory cubature Kalman filter can solve the model mismatch problem, adaptive algorithm will assign attenuation factor in real-time to avoid difficult issues for ranging attenuation factor so as to get a better tracking performance.

TN957.52