Department of Mechanical Engineering,Sharif University of Technology
Welding is an important manufacturing process that can be automated and optimized. This
paper focuses on the development of a robotic arc welding system, wherein a three-degreeof-
freedom Selective Compliance Assembly Robot Arm (SCARA) is interfaced to a Gas Metal
Arc Welding (GMAW) process. The entire system is composed of a robot linked to a GMAW
system. Set points are derived using the desired mass and heat input, along with the weld speed.
The stick-out and the current of the welding process are controlled using an Adaptive Neural
Network Controller (ANNC). The trajectory of the robot or the weld prole is also controlled by
implementing a Mixed Fuzzy-GA Controller (MFGAC) on a three-axis SCARA robot. The system
is, then, analyzed and the results show adequate improvement in the eciency and performance
of the proposed controller in welding a curvilinear weld prole.