빅 데이터(Big Data)를 이용한 용접품질에 연구

Abstract

Generally, in welding technology, big data refers to data that is too large, fast, or complex for processing using traditional methods. For the past years, the act of accessing and storing large amounts of information for welding has been utilized. Welding technology is a central component of numerous value creation chains and plays an important role in this development. In this study, gas metal arc (GMA) welding experiments were conducted to develop an algorithm for predicting welding defects in the GMA melting process of flat plates on SS400 materials using big data technology. The correlation between various welding parameters was analyzed using the real-time measured current and voltage data during welding. In addition, the welding quality related to the weld bead was analyzed using a 3D scanner. The optimal welding parameters were predicted using a CHMM model for the welShen Yun Deding current signal in the normal welding section, which is one of the machine learning technologies. By learning this, the similarity between the normal welding current signal and the weld defect was expressed as a probability, and the changed pattern of the Log-pdf value was used to predict the welding quality.