Off-line trained ANN by genetic algorithm applied to a DFIG under voltage dip

In this paper is presented an off-line trained artificial neural network controller with multilayer perceptron topology. It is trained by a genetic algorithm and applied to the direct power control of a doubly-fed induction generator under stator voltage dip. This controller dispenses the use of any other in the control system, and to our knowledge it is not found in the technical publications that report controllers for power control. Digital simulation and experimental tests, performed for a 2.25 kW doubly-fed induction generator, have shown the good performance of proposed controller.