Programação inteira binária aplicada na coordenação da proteção de sistemas elétricos de potência

• Ano de defesa: 2016
• Autor(a) principal: Corrêa, Rafael
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• dARK ID: ark:/26339/001300000218r
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; Engenharia Elétrica; UFSM; Programa de Pós-Graduação em Engenharia Elétrica; Centro de Tecnologia
• Palavras-chave em Português: Coordenação; Otimização; Programação inteira binária; Proteção; Sistemas elétricos de potência; Coordination; Optimization; Binary integer programming; Protection; Electric power systems; CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA
• Link de acesso: http://repositorio.ufsm.br/handle/1/13784

This paper presents an innovative approach to coordinate typical protection devices of transmission, subtransmission and distribution systems – fuse links and functions ANSI 21, 67, 87, 50 and 51. The settings obtained by the optimization process are the following: characteristic curve, time dial and pickup current settings of overcurrent functions; delay of zones 2 and 3 of distance functions; and the rated current of fuse links. The proposed methodology is based on Binary Integer (BIP), which is an exact optimization technique used to solve problems whose variables can only assume the values 0 or 1. This technique was selected because the settings to be determined are discrete, within the available range of digital relays or fuse links. By the BIP approach the mathematical model, which is non-linear and non-convex originally, is linearized, allowing the application of Mixed Linear Integer Programming (MILP) solvers. A pre-processing step is proposed to enable the reduction of variables for each model aiming to reduce the solver runtime. Moreover, this step is to ensure that the protection operates in secure times, without including these time constraints on the models, which could increase the computational effort to solve them. Simulations were performed on three test systems using the MILP solver of CPLEX®. The solver was executed on its default configuration, so the user does not need to select parameters of the algorithm, commonly presented in metaheuristics, which requires experience. From the simulation results can be concluded that the proposed methodology is innovative due to the fact that no other work considered all the discussed settings as variables; it offers good solutions at a low computational time and can be applied at the power system planning; and has the potential to be applied online, considering the use of a supervisory system and the remote communication between a host computer and digital relays, looking toward the smart grids.