Neural network based self-organized fuzzy logic control in arc welding process

Abstract

A neural network based fuzzy logic control algorithm is set up for welding process control. This control strategy embeds a feedforward neural network into a fuzzy logic controller with the advantages of learning ability of the neural network as well as the explanatory ability of the fuzzy logic system. It can regulate the parameters of a membership function using sample data and generate fuzzy logic rules automatically without consulting the experts. Computer simulation shows that it is feasible for weld geometry regulation.