Optimization of welding process with neural networks

Within the context of integrated design, we propose a new approach for off-line programming of welding robots by interfacing a CAD modeler (geometric database) and an artificial intelligence system (welding database). The CAD system, used to design pieces to be assembled allows us to automatically generate welding trajectories and extract the assembly features needed to determine welding parameters. Using these features, we propose a new approach to generate automatically welding parameters, in GMAW process, with neural networks. We have chosen to use backpropagation neural networks because this;approach integrates database and modeling aspects. Moreover a neural nets based system is easily improvable, if can enlarge his field of application using new experimental welding data. The proposed method is able to determine the welding process and the welding wire to use and then to compute the welding parameters. We present In this paper the system we have developed with neural networks, the results we obtain and the possibilities of the method.