焊接缺陷信号的小波去噪方法研究

针对摩擦焊接头‘伪结合’缺陷信噪比低，难以准确检测的特点，利用平移不变量小波方法对缺陷信号进行去噪，该方法通过多次对所分析的信号进行循环平移，再将平移后的信号进行软（硬）闽值小波消噪处理，然后将消噪后的信号进行逆平移，最后平均所获得的结果。实验证明：该方法能够有效地消除软阈值小波去噪过程中出现的pseudo—Gibbs（伪吉布斯）现象，提高了信噪比，使去噪结果更加逼近于真实信号。该方法在摩擦焊检测信号降噪处理方面具有广泛的应用前景。

A Better Method for Detecting Friction Welding Defect

Pseudo bonding often occurs in friction welding and greatly reduces the strength of friction weld. Traditional detection method finds it very difficult to detect pseudo bonding reliably as the signal's SNR (signal-to-noise ratio) is very low. This paper aims at providing a better detection method that can raise the SNR of the detection signal. In the full paper, we explain our better detecting method in much detail; here we give only a briefing. Our better method includes two ways of utilizing wavelet transform: (1) translation invariant wavelet transform; (2) soft thresholding wavelet transform. In the limited amount of simulation experiments performed by us, we find that using translation invariant wavelet transform can give somewhat higher SNR of detection signal than that obtained with using soft thresholding wavelet transform. Simulation experimental results show preliminarily that our better method can detect pseudo bonding reliably.