Abstract: During pulsed eddy current testing, the noise component of the transient signal and one-time scan waveforms can affect the accuracy of quantitative detection of defects. Both the transient and one-time scan waveforms of pulsed eddy signal are analyzed. Wavelet analysis is utilized in pulsed eddy testing. Wavelet threshold denoise is used to process transient pulsed eddy signal and the denoise performances under different threshold denoise methods are compared and analyzed theoretically. Experimental results confirm the efficiency of wavelet threshold denoise method by comparing with traditional low-pass filters. For one-time scan waveforms, the approximate coefficients of wavelet decomposition are used to reconstruct original signal and peak scanning curves are extracted from reconstructed signal. The experimental results show that, soft threshold has better performance than hard threshold in transient signal denoise because soft threshold not only retains real signal characteristics but also restores the peak value and zero time of the real signal. Therefore soft threshold solves the problem in traditional filter. The peak value scanning curve acquired in this paper eliminates shaking and the extremum points become explicit which improve the defects length testing precision.

Key words: Pulsed Eddy Current（PEC）, detection of defects, wavelet analysis

摘要： 脉冲涡流探伤时，瞬态信号和一次扫描波形中的噪声干扰成分会影响缺陷定量检测的准确度。对于该问题，研究了小波分析在脉冲涡流探伤中的两方面应用。对于瞬态信号，利用小波阈值去噪法对其处理，分析了阈值方法的不同对去噪效果的影响并进行实验验证，并与传统低通滤波进行了比较；对于一次扫描波形，利用小波分解后的近似系数重构信号，再提取峰值扫描曲线。实验结果证明：在瞬态信号的去噪中，软阈值法相比于硬阈值法有更好的去噪效果，其可以在保留真实信号特征的情况下去除噪声，并解决了传统低通滤波会损害真实信号特征的问题，还原了信号的真实峰值和过零时间；使用该方法获得的峰值扫描曲线消去了抖动，极值点时间变得明确，提高了缺陷长度的检测准确度。

关键词: 脉冲涡流, 探伤, 小波分析