Defect Recognition in Ultrasonic Blind Zone Based on Hilbert Marginal Spectrum

In the ultrasonic defect detection of metal, the small defect signals near the surface are submerged in the surface or bottom echo signals, and there is a blind zone of defect detection. To solve this problem, a method based on Hilbert marginal spectrum is proposed to identify metal near surface defects. Hilbert-Huang transform is used to analyze ultrasonic signals and eliminate the interference of non-stationary signals. The feature signal is integrated on the time axis to obtain the actual frequency distribution of the signal to identify the defects, and support vector machine (SVM) is used for cross experiment verification. The near surface micro defects were identified by 304 stainless steel test block. The results show that the Hilbert-Huang transform can effectively extract the features of ultrasonic signals and reduce the influence of blind areas. The average accuracy of defect identification using Hilbert marginal spectrum is 98.94%, which provides a new method for identifying small defects near the surface of metal.