时频域独立元分析方法实现超声兰姆波的多模式分离

为方便兰姆波信号分析与模式定征,提出一种将短时傅里叶变换(Short-Time Fourier Transform,STFT)与独立元分析(Independent Component Analysis,ICA)相结合的多模式超声兰姆波识别方法。首先通过STFT将兰姆波时域信号投影至时频域,基于各模式信号在时频域相对独立的特点,利用ICA实现混叠模式分离。根据分离模式时频能量脊提取各模式群速度曲线,进而估计板厚。将方法运用于时域有限差分(Finite-Difference Time-Domain,FDTD)法仿真与钢板实验,分离得到A0、A1和S0三种模式。仿真与实验中平均群速度估计误差约为1.5%和2.0%,板厚估计误差约为0.3%和2.0%。仿真结果表明,在信噪比(Signal-to-Noise Ratio,SNR)不小于0 dB的情况下,时频域独立元分析方法均可实现兰姆波多模式分离、群速度曲线提取及板厚估计。

Ultrasonic Lamb waves multimodal separation through independent component analysis in the time-frequency domain

To facilitate Lamb waves signal analysis and mode characterization,a multimodal Lamb waves identification method combining Short-Time Fourier Transform(STFT) and Independent Component Analysis(ICA) was proposed.Temporal signals were first projected into the time-frequency domain using STFT.Since different modes were relatively independent in the time-frequency domain,ICA was then introduced to realize the multimodal separation.The group velocity curve of each mode was extracted according to its time-frequency energy ridge.Thus,the plate thickness can be further estimated.The proposed method was applied to the Finite-Difference Time-Domain(FDTD) simulation and the steel plate experiment,with A0,S0 and A1 mode being separated.The average error of group velocity estimation in simulation and experiment was about 1.5% and 2.0%,with that of plate thickness estimation being 0.3% and 2.0%approximately.The simulation results show that,as the Signal-to-Noise Ratio(SNR) is not less than 0 dB,the timefrequency independent component analysis method can be used for Lamb waves multimodal separation,group velocity curves extraction and plate thickness estimation.