Desenvolvimento de plataforma emuladora de turbina eólica para estudos de algoritmos de MPPT eólicos inteligentes

Abstract

The dynamics of the wind within the wind power small context is problematic since the operating speed of the machine must accompany him in order to extract the maximum wind power. In this paper, we propose the design and development of a wind turbine emulator bench and peak power tracking algorithms models based on artificial neural networks and reinforcement learning. The emulation system aims to allow the algorithms of experimental evaluation previously validated by simulation, since the proposed algorithms aim to achieve a good performance compared to classical algorithms. In addition to literature review, computer simulations were implemented in PSIM and Matlab software, as well as the design, development and validation of emulator bench wind turbine based on DC motor. They present all design steps the emulator converters and charge controller responsible for carrying out the maximum power tracking, as well as all the material necessary for reproduction of the work in the form of appendices. In the emulator bench, two algorithms are proposed in this work: one based on a modification algorithm and disturbs observed through insertion of a neural network that defines the size of the perturbation; already the second is based on recent learning algorithms for enhancing the Actuator-Critical type (CACLA), which had not been used for this purpose