SolVisionCase Study

AI Laser Welding Defect Detection

Automating Laser Welding Inspections

The art of laser welding

Laser welding uses a highly concentrated heat source to quickly melt, vaporize, cool off, then solidify materials to join them together. Relatively easier to set up, it provides rapid and precise seam welding for even heat-resistant materials, as well as improved efficiency and quality compared to traditional welding.

AI deep learning technology for laser welding defect detection

Quality indicator: welding weave patterns

Two main types of laser welding are heat conduction welding and keyhole (deep penetration) welding, and they generate different weld characteristics. As welding positions and patterns are ever-changing, the number of possible variations makes it challenging for rule-based systems to recognize welding weave patterns or locate missed welding spots and other production errors. Therefore, welding defect inspections are usually performed manually, and are one of the causes of inconsistent welding quality.

Intelligent AI welding inspections

SolVision is an advanced visual inspection system that uses sample images to train an AI model to recognize welding inconsistencies. The system’s Classification and Segmentation tools can accurately identify welding rates and missed welds, as well as detect the number and distribution of welding weave patterns using deep learning.