Wind turbine emulator using an MPPT controller based on neural networks

This paper presents the development of an experimental prototype of an emulator for wind turbines, with controllers based on artificial neural networks. The emulator in question is able to easily represent the dynamic behavior of wind turbines with different characteristics. The application of neural networks in this study focuses on the maximum power point tracking system of a wind turbine, eliminating the wind speed sensoring. The elimination of these sensors is aimed at reducing the problems related to the logistics of maintenance. The network receives as input the mechanical power of the rotor axis and the generator speed and provides as output the estimated wind speed. The wind power generation system consists of a wind turbine connected to the grid through an induction generator and a PWM back-to-back converter. The induction generator is controlled so that the wind turbine always operates at the optimum speed reference provided by the neural network, thus extracting the maximum wind energy. The wind power generation system, the emulator and the neural network were modeled in PSCAD and Mathematica. After that, it have been implemented in an experimental prototype using two F28335 DSP microcontrollers.