Akademik Veri Yönetim Sistemi
6th International Congress on Human-Computer Interaction, Optimization and Robotic Applications, HORA 2024, İstanbul, Türkiye, 23 - 25 Mayıs 2024
Electric resistance spot welding, also known as spot welding, is commonly used in industry for joining metal parts. Rapid and accurate classification to determine whether spot welding is appropriate is necessary to achieve the desired product quality level and ensure the continuity of serial production conditions such as cycle time. In this study, we developed a system that can be easily adapted to both robotic and manual spot welding processes. We supported the visual inspection process of spot welds performed by an operator in non-destructive testing during serial production with our system. With the algorithm we developed, we achieved classification rates of 95% to 98% accuracy. We visualized spot weld images on the surface captured by a camera in a way that operators can easily see these results. Our system demonstrated the ability to easily adapt to different spot welding processes due to its ability to evaluate spot welds within their own unique processes, without requiring special conditions other than basic requirements such as lighting and positioning. The fast operation of our system, alongside its high accuracy rates, indicated its suitability for use in serial production conditions. Thus, we established a spot weld evaluation process that supports operators against the risk of making erroneous evaluations under variable conditions.