小波降噪算法在临界折射纵波残余应力检测中的应用

Application of Ultrasonic Algorithm Based on the Wavelet De-Nosing on the Longitudinal Critical Refraction (LCR) for Residual Stress Detection

利用超声波方法在进行材料及结构的残余应力检测时,影响因素众多。其中由于物体本身材料的不均匀性或存在缺陷,产生的超声信号噪声会影响应力的检测精度。以小波降噪算法为研究对象,对小波分解层数、阈值算法和阈值处理算法等影响小波降噪的三个条件进行综合分析。结果表明:当选用硬阈值算法对小波系数进行处理后重构的信号信噪比值整体比软阈值算法高很多;在选用无偏似然估计阈值和启发式阈值时,整体的降噪效果比使用固定阈值和极大极小阈值时好很多;并且在相同阈值处理和阈值算法下,降噪效果随着均随着分解层数的增加而降低。选用sym8小波,3层小波分解、无偏似然估计阈值和硬阈值处理是最优的降噪参数。

There are manyfactors that would affect the accuracy of residual stress detection with ultrasonic wave method. Especially the distribution uniform or defects of the materials would affect the results greatly. In this paper, the ultrasonic algorithm based on the wavelet de-nosing was taken into consideration, the wavelet decomposition level, threshold algorithm were analyzed. The results showed that the reconstructed signal to noise ratio(SNR) of hard algorithm was higher than the soft algorithm. The whole noise attenuation effect were higher of unbiased likelihood estimation threshold method and heur-sure threshold than the stable threshold and maximum or minimum threshold. The noise attenuation effect decreased with the decomposed layers increased under the same threshold processing and threshold algorithm. The best noise parameter was sym8 wavelet de-nosing, three layers decomposition, unbiased likelihood estimation threshold method and hard threshold processing.