Compensation method for metal defect detection based on deep learning and ultrasonic wave

Ultrasonic nondestructive testing plays a crucial role in evaluating the integrity of metal structures. When there are defects in the tested workpiece, there will be reflection, refraction, diffraction, and other behaviors of sound waves. By analyzing the ultrasonic echo signal, the location and depth of defects can be identified. In previous ultrasonic defect detection, due to changes in the ultrasonic probe, the identified defects may change, affecting the accuracy of defect discrimination. This paper proposes a method for detecting the depth of metal cracks based on deep learning. This method combines finite element models and experimental analysis data to compensate for the depth of metal cracks, improving the accuracy of defect identification. Finally, an ultrasonic experimental study on metal components was conducted to verify the effectiveness of this method.