一种检测低信噪比响应信号的广义维纳滤波方法

根据所测动态响应信号拟合系统模型时,一个重要的问题是:在低信噪比的情况下如何提高系统参数的估计精度。采用线性平均方法能适当提高信噪比,但在一般试验条件下,据此所获得的部分系统参数仍具有较大的误差。本文采用了时域广义维纳滤波方法对测取的动态信号进行预处理,以尽可能少的激励次数,从较高的背底噪声中检测出真实信号。文中还说明了适用于工程系统测试中噪声功率估计的滤波器分组方法。以悬臂梁的模态测试及分析为例,检验了该方法的有效性。试验表明,作振型分析时,因测点信号信噪比低而导致数据误差大的问题得以解决,这对实验模态分析是具有重要意义的。所完成的计算机-FFT 谱分析器信号预处理程序有很大的实用价值。

A Generalized Wiener Filtering Method for Detecting the Low Signal-to-Noise Response Signal

When the model of system is fitted according to the dynamic response signalmeasured,there is an important problem:how to increase the estimation accuracyof system parameters in the condidtion of low signal-to-noise ratio.Linearaveraging method can partly increase the signal-noise ratio,but in the general testcondition the system parameters obtained by this method have also unacceptableerror.In this paper,a time-domain generalized wiener filtering method developedby the authors is used to pre-processing the dynamic response signal measured andthe real signal can be detected from the high background noise by as few stimulias possible.The method of grouping filters suitable to the estimation of noisepower in engineering system measurement is also studied in this paper.The modal measurement and analysis of a cantilever beam are taken as anexample to check the effectiveness of this method.The test shows that during theanalysis of vibration mode the data error due to the low signal-noise ratio at themeasuring point is solved and this is significant for experimental modal analysis.This signal pre-processing program developed for computer-spectrum analyzerhas large practical value.