Uso de rede neural artificial com função de base radial para controle da velocidade do rotor em turbina eólica

• Autor(a) principal: Silva, Tássio Alessandro Borges da
• Data de Publicação: 2018
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UFPB
• Texto Completo: https://repositorio.ufpb.br/jspui/handle/123456789/14002
• topic: Turbina Eólica; Redes Neurais; Controle do ângulo das pás; Atenuação de Cargas; Wind Turbine; Pitch Control; Neural Network; Load attenuation; Rede neural artificial - Turbina eólica - Sistema; Energia eólica; Cargas - Atenuação; CNPQ::ENGENHARIAS::ENGENHARIA MECANICA

The crescent growth in the wind energy demands more reliable and efficient wind turbines with an increased life span. The wind turbine controllers play a major role in regulating several parameters, but the traditional controllers limit to control the rotor speed, thus ignoring the system dynamics. The techniques for modern control design are applicable in linear systems only, which prevent their use directly in the non-linear system of a wind turbine. The system linearization provides a linearized equation that represents the system dynamics in a small range and allows the use of modern control techniques, but when operating outside the range, the controller cannot perform satisfactorily. The GainScheduling technique consider a larger range by using several linearized systems and their controllers to choose between them when operating in different regions. To choose the gains an Artificial Neural Network with Radial Basis Function is used, this allows overcoming the non-linearity of the system. The Neural Network is trained using the data from stationary states of the wind turbine and controllers gains for each linearized system. Its performance presents an attenuation in the wind turbine components’ oscillations that leads to load mitigation. These loads are evaluated under turbulent wind and wind gust according to the IEC 61400.