Nugget and corona bond size measurement through active thermography and transfer learning model

• DOI: 10.1007/s00170-024-14096-4
• Publication Date: 2024-07-10T10:01:42Z
• Authors: Luca Santoro; Valentino Razza; Manuela De Maddis
• Keywords: Spot welding; SIGNAL (programming language); Transfer of learning

Abstract Resistance spot welding (RSW) is considered a preferred technique for joining metal parts in various industries, mainly its efficiency and cost-effectiveness. The mechanical properties of welds are pivotal ensuring structural integrity overall assembly performance. In this work, the quality attributes resistance welding, such as both nugget corona bond sizes, assessed by analyzing thermal behavior joint using physical information neural network (PINN). Starting from signal phase gradient amplitude maps, convolutional (CNN) estimates size nuggets bonds. CNN architecture based on Inception V3 architecture, state-of-the-art that excels image recognition tasks. This study suggests adopting new methodology automatic RSW control analysis.