Analysis of Ultrasonic Propagation Characteristics of Metal 3D Printed Products and Defect Classification through Transfer Learning

With the increasing growth of the 3D printer market, research is being conducted to manufacture parts using 3D printers rather than conventional cutting and casting methods. Consequently, current studies are focusing on improving the manufacturing speed and accuracy of 3D printers to create components faster at lower costs. However, due to the characteristics of the stacked structure, defects such as pores and cracks may occur during the manufacturing process. Such defects can damage equipment assembled with various parts, increasing the possibility of human and material damage. Therefore, in this study, the ultrasonic propagation characteristics of cutting and additive products are compared, and the ultrasonic propagation characteristics of each lamination angle of 3D printer products are analyzed. Finally, the characteristics of the defect specimen are compared to detect defects, and an analysis is done to determine the presence or absence of defects through transfer learning, which is a commonly employed deep learning technique.