Nonlinear Model Identification and Adaptive Control of Variable Speed Wind Turbine Using Recurrent Neural Network

The best configuration for generating electricity energy form a variable-speed wind energy conversion system (WECS) is using double-output induction generator (DOIG). Controlling this system in order to optimum performance on maximum extracting power from wind in each speed were attracted the attention of many researchers. This kind of generators use a rectifier and inverter know as static Kramer drive (SKD) and changes on the firing angle of the inverter can control the operation of the generator. Achieving above purpose is difficult because the behavior of this system under classic controller is very time variant and nonlinear and need to an adaptive controller is presented. With regard to high capability of neural network in control subject, in this paper one structure of this kind of networks for controlling wind energy conversion system was proposed. This controller uses recurrent neural network based on approximation of non-linear autoregressive moving average (NARMA) model. Feasibility and effectiveness of controller are demonstrated by simulation results. Different cases, such as applying a distinct disturbance, applying noise to system and Parameters variations and uncertainties of the system in order to study the ability of proposed controllers, were considered.