基于CWPA模型的最优平滑算法性能及其应用  

Performance and application of optimal smoothing algorithms based on CWPA model

作　　者：程建华[1,2] 陈岱岱[1] 赵琳[1] 王冰玉[1] Rene Landry 

机构地区：[1]哈尔滨工程大学自动化学院,哈尔滨150001 [2]魁北克大学ETS学院

出　　处：《中国惯性技术学报》2014年第6期-,共7页Journal of Chinese Inertial Technology

基　　金：国家自然科学基金(61374007;61104036);中央高校基本科研业务费专项资金(HEUCFX41309)

摘　　要：针对线性高斯系统的平滑问题,分析了RTS固定区间平滑与双滤波器固定区间平滑两种算法,提出了一种滤波存储数据更少的RTS平滑新算法。结合平面内的运动追踪问题,基于二维CWPA模型,仿真分析了卡尔曼滤波、RTS固定区间平滑以及双滤波器平滑算法的估计性能。仿真结果表明,两种固定区间平滑算法的估计效果等效,精度均优于卡尔曼滤波,对于实际问题中固定区间平滑算法的选用具有一定的参考价值。最后,结合双滤波器结构提出了一种基于双平滑器的舰载武器惯导传递对准精度评估方法,结果表明新方法相比于单一的平滑算法,可以获取更优的综合平滑性能,特别提升了水平姿态对准误差的平滑估计性能。Aiming at the smoothing problem in linear Gaussian systems, the RTS smoother and two-filter smoother were analyzed. A novel RTS smoother was proposed which need less filtering data to be stored than that of traditional one. By taking into account the in-plane motion tracking problem, a continue Wiener process acceleration(CWPA) model was introduced. Based on the CWPA model, the estimation performances of Kalman filter, RTS smoother and two-filter smoother were studied by simulation. The results show that the estimation effects of the two smoothers are equivalent, and both are better than that of Kalman filter. Finally, according with the framework of two-filter smoother, a two-smoother-based transfer alignment accuracy evaluation approach for airborne slave inertial navigation system was presented. Compared with the conventional schemes in which only single smoother is adopted, this new approach can get better overall smooth performance. In particular, it improves the smoothing-based estimation performance of horizontal attitude misalignment.

关 键 词：最优平滑 卡尔曼滤波 RTS平滑器 双滤波器平滑器 CWPA模型 传递对准精度评估 

分 类 号：U666.1[交通运输工程—船舶及航道工程]