Data Processing On Vibration Test And Modal Parameter Identification

The vibration signal processing and modal parameter identification take important parts in model experiments,vibration testing and structural health monitoring.With the development of digital signal processing theory and vibration test equipment,vibration signal processing and modal parameter identification technology has achieved significant improvement.However,due to the small amplitude response of large scale civil engineering structures under ambient excitation and the noise which comes from instruments and equipments,it is still difficul to get valuable information from the recorded signals.Take the integration of acceleration for example,it offen drift away from the baseline and far deviateed from the true displacement time series.This paper summarizes the previous work of the past researches and focus on the integral of acceleration,spectral refinement,transfer function refinement,frequency domain modal parameter identification.Then clarify some misunderstanding.At last,personal opinion and innovation have been made.The main conclusions are as follows:(1)The integration algorithm of the frequency domain and time domain has litter difference between each other.Their difference can be fitted by a linear trend.The biggest advantage of frequency-domain integration is that it can filter out the long-cycle trends durnning integration and has the ability to filter out the specified frequency(such as current frequency interference).(2)Durnning the integration of acceleration,it offen interfered by the long-cycle trends.Polynomial fitting,high-pass filter,the EMD methods are able to eliminate the trend to some extent.But the best is the high-pass filtering algorithm which very sensitive to low frequency while not sensitive to high frequency.(3)The ability of CZT refinement algorithms and the FFT spectrum of continuous refinement algorithms are the best without increasing the sampling data.Phase compensation FFT algorithm and ZoomFFT algorithms need to increase D-fold(refinemented number)sampling data while zooming spectrum.Compared to the refinement results by direct FFT,ZoomFFT algorithm does not improve the frequency resolution.The CZT algorithm and the FFT spectrum of continuous refinement algorithm can be any degree of refinement of the spectrum.However the largeset zoom factor of ZoomFFT is fixed.(4)The FFT spectrum of continuous refinement algorithm and CZT transform algorithm are the best compare to the other algorithms which get smoother transfer function and fitting the unrefinemented transfer function better.The refined transfer function comes from ZoomFFT is rougher than CZT algorithm and could not get exact amplitude of the transfer functuon at the half band power.Also the ZoomFFT's largest zoom factor is limited.(5)The refinement of transfer function and improved Levy algorithm can both improve the accuracy of modal parameter identification result.However the combined algorithms can get better results.