一种基于递推最小二乘的频域辨识的新算法

A New Arithmetic of Frequency Identification Based on Least Squares

研究系统传递函数频率特性的准确测量问题,对于飞行器器件而言,系统的传递函数往往比较复杂,主要体现在传函阶次较高,传函中系数之间的量级相差很大。同时,不同输入信号之间的频率量级差别也很大。应用一般的算法所辨识出的结果往往不太理想。研究了如何利用频域响应数据辨识飞行器器件的传递函数,提出了一种系统频域辨识的新算法。算法在递推最小二乘的基础上,对辨识数据进行迭代修正处理,以提高结果在低频部分的拟合度。通过仿真,验证了算法的有效性。

The concept of frequency identification is originated from measurement of frequency characteristics of system transfer function. And for aircraft component, its transfer function is often complex, mainly showing the high order, huge difference in magnitude of its coefficient and the wide band of data provided. Therefore, results identified with traditional algorithms are not good enough. This paper studied how to identify transfer function of aircraft compo nent using frequency domain response data, and presented a new algorithm for system frequency identification. On the basis of the recursive least squares, the data to be identified were amended by iterating to improve the fitting effect in the low frequency part. The validity of this algorithm was substantiated by simulation experiment.