基于双树复小波系数层间相关性的非线性超声谐波提取方法

Feature Extraction Method for Nonlinear Ultrasonic Harmonic Based on Dual-Tree Complex Wavelet Scale Coefficients Correlation

非线性超声信号具有非平稳性、非线性和高次谐波信噪比低的特点,为提高非线性超声无损检测技术对缺陷的表征能力,提出一种基于双树复小波系数层间相关性结合软阈值滤波算法的超声谐波提取方法。首先采用双树复小波将信号分解为基频和二倍频等不同频带的分量;由于各分量存在一定程度的频率混叠,利用小波系数层间相关性对各分量信号滤波,消除频率混叠并得到修正后的细节子波;然后结合软阈值算法对修正后的小波系数进一步降噪;最后将滤波降噪后的各分量系数重构,即可实现对非线性超声信号中基波和二次谐波信号的提取。实验结果表明,该算法滤波效果良好,有效地提取了二次谐波信号,提高了非线性超声检测结果的准确性和鲁棒性。

Nonlinear ultrasound signal is non-stationary,nonlinear and has a low signal-to-noise ratio.In order to improve the ability of nonlinear ultrasonic nondestructive testing(NDT)to characterize defects,an ultrasonic harmonic extraction method based on double tree complex wavelet coefficients interlayer correlation and soft threshold filtering algorithm was proposed.Firstly,the signal was decomposed into fundamental frequency and diploid frequency components by using the double tree complex wavelet.Because each component has a certain degree of frequency aliasing,the wavelet coefficients were used to filter the signals of each component,eliminate the frequency aliasing and get the modified detail wavelet.Then,a further de-noising process was performed on the amended wavelet coefficients using threshold filtering algorithm.Finally,the fundamental wave and second harmonic can be extracted by reconstructing signal from the processed wavelet coefficients of different components.Experimental results prove that the proposed method can effectively extract second harmonic,which improves the accuracy and robustness of the testing result.