The welding power supplies group control system based on T-S model and ANFIS

A group control system of welding power supplies with the combination of fuzzy theory and neural networks is proposed in the paper. The membership function of state variables, inference rules in the system, algorithms for fuzzy inference and defuzzification, etc. are analyzed and devised subsequently. Based on those, adaptive neuro-fuzzy inference system (ANFIS) with Tagaki-Sugeno model is achieved. Finally the completed ANFIS is simulated using Simulink toolbox in Matlab, which is used in Membrane-water-wall welding machine for experimentation. The result shows that the design has the advantage of less regulating time (shortened by 22%) and less overshoot (reduced by 40%), and the better dynamic performance is obtained.

I. Introduction

Presently, the control system of welding power supply has made great progress with the development of digital technology. It brought about the naissance of intelligent power supplies, such as Lintegrai series of EWM Company, Transplus synergic series of Fronius Company [1], and so on. At the same time, some new control arithmetic were presented, for example, waveform control in welding power supplies [2], modeling and estimating for external characteristic of welding source [3], and integrated quality control system for GMA-welding [4]. Those intelligent power supplies and control strategies make a positive effect on reducing impact on welding process due to voltage fluctuating with single source control.