A neural networks based controller for Electron Beam welding power supply

Welding is an unavoidable unit practically for every manufacturing industry. Electron Beam welding (EBW) are very important unit of some specific manufacturing processes where high degree of accuracy and flawless welding is highly desirable like aerospace engineering. The power supply unit (PSU) used for EBW are very important unit, for which high degree of stability and accuracy is a must. Since EBW is absolutely nonlinear system, for better performance, adopting non linear control methods could be a good solution. In the current study a robust adaptive controller based on multi-layer feed-forward neural network is developed for real-time voltage regulation. Simulation shows a better characteristic in terms of settling time and overshoot for the neural controller compared to that of a conventional PI controller based on Ziegler Nichols [1,2,3,4] frequency response tuning method. The controller has the unique advantages of nonlinear mapping and adaptive learning.