A Study on Modeling of CO₂ Weld Seam Forming Quality

Abstract

CO2welding technology is widely used nowadays, because the work environment is very bad, weld automation technology is urgently needed. To control the weld quality automatically, weld sensors should be first used to obtain information that could reflect the weld quality. This paper used arc and visual sensors to obtain the electrical and weld pool image of CO2weld process, and signal processing method was used to obtain the signal features of the information. Then neural network method was used to model the process, experiment results showed that the method could effectively predict the weld seam forming.