系统噪声不可忽略时的气动力参数辨识方法对比  

Comparison of the methods for identifying the aerodynamic parameters when system noise cannot be ignored

作　　者：刘庆[1] 左军毅[2] 李雷[1] LIU Qing;ZUO Jun-yi;LI Lei(Aircraft Flight Test Technology Institute,CFTE,Xi'an 710089,China;School of Aeronautics,Northwestern Polytechnical University,Xi'an 710072,China)

机构地区：[1]中国飞行试验研究院飞机所,陕西西安710089 [2]西北工业大学航空学院,陕西西安710072

出　　处：《飞行力学》2019年第1期11-16,共6页Flight Dynamics

基　　金：国家自然科学基金资助(61473227;11472222);航空科学基金资助(20151353018);航天技术支撑基金资助(2014-HT-XGD);中央高校基本科研业务费资助(3102015ZY001);陕西省自然科学基金资助(2015JM6304)

摘　　要：面向系统噪声不可忽略时的气动力参数辨识问题,对比了基于期望最大化的方法(EMBM)、滤波误差法(FEM)和输出误差法(OEM)三种辨识算法的性能。首先,分析了三种辨识算法在基本处理框架、算法结构以及执行细节等方面的差异;然后,通过算例对其在计算时间、对系统噪声的适应能力、对参数初值的敏感度以及收敛性能4个方面进行了对比。结果表明,除了计算时间略有增加外,EMBM的性能明显优于FEM和OEM。For the problem of the aerodynamic parameter identification with the system noise considered,three identification algorithms including expectation maximization based method( EMBM),filter error method( FEM) and output error method( OEM) were compared.Firstly,the differences of three identification algorithms were analyzed in terms of basic processing framework,algorithm structure and execution details.Then,the performances in four aspects including calculation time,adaptability to system noise,sensitivity to initial values of parameters and convergence performance were systematically compared by an example.The results show that except for the increased computing time,the EMBM significantly precedes FEM and OEM on performance.

关 键 词：参数辨识 状态估计 非线性滤波 气动力模型 

分 类 号：V212.1[航空宇航科学与技术—航空宇航推进理论与工程]