抄録
This paper is describe d about the current status and future development of in-process monitoring technology, which is a key technology for welding work toward the realization of smart factory. Molten pool sensing technology and bead shape measurement are generally used as monitoring technology for quality information during welding. Machine learning such as deep learning has recently been applied to these measurements. Although temperature measurement of weld is difficult in real time and has few applications, it is an indispensable technique for confirming the mechanical properties of welds. On the other hand, the laser ultrasonic method has the advantage of being able to directly detect the internal weld defects during welding, and a practical system mounted on the welding robot has been developed. In addition, a fully autonomous welding robot system can be constructed by fusing the prediction technology such as penetration shape and temperature distribution by numerical simulation technology using a physical model of heat source and molten pool formation with these sensing information. Furthermore, it will be effective to build a welder support system using these sensing technologies in collaboration with humans.
