摘要
针对时变参数的在线辨识问题,提出了基于CZT的变时间窗时变参数在线辨识方法,通过滑动时间窗以及基于方差阈值的重启设计,提高了现有频域辨识方法对时变参数的辨识速度,并且有效避免了数值微分对噪声的放大问题。最后,基于飞机系统的执行机构故障,对该方法进行仿真验证,并与传统辨识方法进行对比,说明了该方法对时变参数有较强的辨识能力,验证了该方法的有效性。
For online identification problem of time-varying parameter,an online identification method is proposed based on sliding time window design and Chirp-Z transform.By the way of sliding time window design and CZT restarting design once the variance is above the designed threshold,the convergence speed of traditional identification method is effectively improved by using this method.Besides,the proposed identification is based on the frequency domain,which can also avoid the noise caused by the numerical differentiation.The simulation on the aircraft aerodynamic system is implemented to verify the efficiency of proposed method.
作者
钟鸿豪
白文艳
黄万伟
Zhong Honghao;Bai Wenyan;Huang Wanwei(Beijing Aerospace Automatic Control Institute,Beijing 100854,China)
出处
《航天控制》
CSCD
北大核心
2019年第6期19-23,共5页
Aerospace Control
关键词
CZT
变时间窗
时变参数
在线辨识
Chirp-Z transform
Adaptive time window
Time varying parameter
Online identification
