Utilização de redes neurais artificiais para rastreamento de máxima potência de sistemas fotovoltaicos parcialmente sombreados

• Autor(a) principal: Silva, Júlia Rosenthal Caetano
• Data de Publicação: 2019
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Repositório Institucional da UFU
• Texto Completo: https://repositorio.ufu.br/handle/123456789/27863; http://doi.org/10.14393/ufu.te.2024.5039
• topic: Geração de energia fotovoltaica; Photovoltaic power generation; Engenharia Elétrica; Electrical engineering; Conversores de corrente elétrica; Electric current converters; Simulação (Computadores); Computer simulation; Redes Neurais (Computação); Neural Networks (Computer science); CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA::SISTEMAS ELETRICOS DE POTENCIA::CONVERSAO E RETIFICACAO DA ENERGIA ELETRICA

Large demands for electricity, coupled with growing environmental concern, welcomes new researches and developments of alternative, clean and renewable sources of electricity. The sun, as direct conversion of electricity, has become increasingly used as the main choice in distributed generation in small consumer units. Other means of obtaining electric energy, such as wind, tidal, hydroelectric and biomass sources, require very specific factors for its implementation, and is not feasible for residential use. Advances in legislation with government programs that facilitate the use of power grid connections, such as the “Luz Para Todos” program, have become popular the use of this type of technology in Brazil. Many citizens who previously had no electricity now have a small photovoltaic power plant that serves basic sanitation needs such as the use of refrigerators and night lighting. It is within this scope that this research follows, with the main focus on the control of maximum power recognition. The use of an Artificial Neural Network tries to optimize the generation, in this case, in the specific situation of partial shading of the photovoltaic generator. The computational study was done using MATLAB/Simulink software with all the com- ponents assembled separately, allowing the individual study of each one. The components evaluated are the photovoltaic generation system, the voltage-up converter, the control of the maximum power point and its comparison with methods long established in the market, the inverter and the connection to the power grid. Tests were carried out on a proposed Artificial Neural Network and comparisons were made with other maximum power control methods of photovoltaic systems. The results were satisfactory with great advantage over other methods.