Artificial Neural Network and Optimized control for Resistance Spot Welding System

摘要

Research work carried out in this paper is to compare performance analysis of optimization and artificial intelligence control on current spike reduction by means of magnetization level control in the primary winding on the medium frequency direct current (MFDC) welding transformer for resistance spot welding system (RSWS). Spike reduction in primary current of a welding transformer is a major criterion for uninterrupted operation of spot-welding system. Spike generation in spot welding system is due to the unequal impedance of secondary circuits of transformer and different characteristics of diodes at the load end cause the protection system of welding system to be switched-off. The current control technique is a piecewise linear control technique that is inspired from the DC-DC converter control algorithms to register a novel current spike reduction method in the MFDC spot welding applications. Conventional control systems like PI controller and Hysteresis controller are implemented in the previous research but those controllers were facing a problem like high ripple content, spike reduction in the current signal is not in desired limits, etc. advancement in the control area of spot-welding system has been carried out with artificial intelligence techniques. Here in this paper, the two controllers from two different environment chosen in order to reduce the spike in the primary current of welding transformer. This paper analyses the performance of Artificial Neural Network (ANN) controller and Optimized controller in the view of spike reduction in the current, total harmonic distortion (THD), percentage ripple in welding current, rise time and settling time. Above mentioned controllers are implemented in Matlab/Simulink software environment with 220 kVA welding transformer and results are tabulated