Intelligent neural network control system for gas metal arc welding

Abstract

A neural network control system for keeping arc stability and decreasing the spatter during gas metal arc welding have been created. The characterization and relationship between arc sound and arc stability was studied. Tree kinds of neural network control constructions were presented. After simulated the static and dynamic performance in welding processes, it can be found that the error back propagating model neural network have better properties. The factors affecting the simulating results and the dynamic response quality have also been analyzed.