Computer Simulation of Neural Network Control System for CO2 Welding Process

摘要

In this paper, neural network control systems for decreasing the spatter of CO2 welding have been created. The Generalized Inverse Learning Architecture(GILA), the Specialized Inverse Learning Architecture(SILA)-I & II and the Error Back Propagat Model(EBPM) are adopted respectively to simulate the static and dynamic welding control processes. The results of simulation show that the SILA-I and EBPM have better properties. The factors affecting the simulating results and the dynamic response quality have also been analyzed.