Estudo e desenvolvimento de um sistema de energia baseado em célula a combustível para injeção de potência na rede elétrica

• Ano de defesa: 2012
• Autor(a) principal: Avelar, Henrique José
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; BR; Programa de Pós-graduação em Engenharia Elétrica; Engenharias; UFU
• Palavras-chave em Português: Fontes alternativas de energia; Células a combustível de membrana polimérica; Inversores; Paralelismo de inversores com a rede; DSP; Identificação de sistemas; Modelamento; Energia - Fontes alternativas; Células a combustível; Alternative energy sources; Proton exchange membrane fuel cell (PEMFC); Inverters; Grid connected; Digital control; System identification; Modeling; CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA
• Link de acesso: https://repositorio.ufu.br/handle/123456789/14306; https://doi.org/10.14393/ufu.te.2012.56

This work presents the study and development of an alternative source of energy, composed primarily of a fuel cell as the primary organ of generation, a full-bridge converter lift and a single-phase PWM inverter. This source can operate in standalone mode, feeding local loads, or connected to the grid. An electric model for the fuel cell is proposed, which considers changes in temperature as a function of current at its terminals. This model was developed from tests on a 1,2 kW fuel cell, using the identification by the least squares method with the application of pseudo-random binary signals and optimization techniques. To study the parallel connection with the electric grid is also presented a state space model of the PWM inverter connected to the grid, using active power frequency (P×ω) and reactive power voltage (Q×V) droops control scheme, including an additional feedback from the active power variation on the inverter phase, which allows the analysis of the dynamics and stability of the inverter connected to the grid. Simulation and experimental results are presented to validate the models.