Reconfiguração de sistemas de distribuição operando em vários níveis de demanda através de uma meta-heurística de busca em vizinhança variável

• Ano de defesa: 2015
• Autor(a) principal: Possagnolo, Leonardo Henrique Faria Macedo [UNESP]
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual Paulista (Unesp)
• Palavras-chave em Português: Energia elétrica - Distribuição; Sistemas de energia eletrica; Programação heuristica; Electric power distribution
• Link de acesso: http://hdl.handle.net/11449/126489; http://www.athena.biblioteca.unesp.br/exlibris/bd/cathedra/12-08-2015/000844053.pdf

The distribution network reconfiguration problem consists in determining the radial to- pology, that can be obtained by opening and closing sectionalizing switches (normally closed switches) and tie switches (normally open switches), so that an objective is achieved, commonly loss minimization, load balancing, voltage levels improvement or fault isolation. Furthermore, the optimal topology must satisfy operational constraints, such as voltage levels on nodes and current magnitude on circuits. The model for this problem is a mixed-integer nonlinear pro- gramming problem, non-convex and hard to solve by classical optimization techniques, besides, this problem presents the combinatorial explosion phenomenon. This work presents methodol- ogies, based on the variable neighborhood search metaheuristic, to solve the distribution net- work reconfiguration problem with variable demand and fixed topology, which aims in finding only one optimal topology to operate on the various load levels during a period. The considered objective is the reduction of the cost of energy losses. Four variable neighborhood search algo- rithms were developed: Basic Variable Neighborhood Search (BVNS), Variable Neighborhood Descent (VND), Reduced Variable Neighborhood Search (RVNS) and General Variable Neigh- borhood Search (GVNS). All programs were implemented in FORTRAN. The proposed algo- rithms were tested with the 33, 84, 136, 415 and 10477-node systems. The results were com- pared with the best-known solutions presented in specialized literature and the solutions ob- tained from an optimization model, written in AMPL and solved with commercial solver CPLEX