Algoritmo de busca dispersa aplicado ao problema de fluxo de potência ótimo considerando o desligamento de linhas de transmissão
Carregando...
Arquivos
Data
2019-10-25
Autores
Garcia, André Mendes
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
O principal objetivo deste trabalho é a implementação de uma metodologia que, utilizando a meta-heurística de busca dispersa (BD) resolva o problema de fluxo de potência ótimo (FPO) considerando o desligamento de linhas de transmissão (OTS) para a redução dos custos de ope-ração. Com o objetivo de avaliar o potencial da meta-heurística, o algoritmo de BD foi imple-mentado para otimizar funções multimodais restritas, metodologia denominada BD-FMR, e para resolver o problema de FPO, metodologia denominada BD-FPO. Foram realizados testes com onze problemas de funções multimodais restritas disponíveis na literatura especializada, utili-zando a metodologia BD-FMR, sendo que os resultados obtidos são comparáveis com os me-lhores resultados disponíveis na literatura. O problema de FPO foi resolvido pela metodologia BD-FPO utilizando três sistemas teste de 6, 14 e 57 barras, sendo que os resultados não foram satisfatórios quando comparados com as soluções do modelo exato do problema obtidas pelo solver KNITRO. Entretanto, o algoritmo BD-FPO serviu de base para a implementação da me-todologia principal deste trabalho. Por fim, a metodologia BD-OTS foi implementada em lin-guagem de programação C/C++, com a utilização de recursos de programação paralela através da biblioteca OpenMP. Neste trabalho a formulação utilizada para representar a operação da rede considera o modelo AC (corrente alternada), que consiste em um problema de programa-ção não linear inteira mista (PNLIM) devido a presença de variáveis discretas relacionadas com o estado de operação de uma linha, com a posição dos taps dos transformadores e com o estado de operação dos compensadores shunt. Na metodologia BD-OTS, a meta-heurística gerencia as variáveis inteiras e binárias, fornece para o modelo matemático essas variáveis na etapa do cál-culo da função objetivo, transformando o modelo em um problema de programação não linear (PNL). O problema de PNL é então resolvido pelo solver KNITRO, utilizando o método de pontos interiores, o qual retorna para a meta-heurística todas as demais variáveis contínuas do problema. Foram utilizados onze sistemas teste, dois sistemas de 14 barras, três de 30 barras, dois de 57 barras, dois de 118 barras e dois de 162 barras. As soluções do problema de OTS pela metodologia BD-OTS foram comparadas com resultados das soluções destes mesmos sis-temas obtidos na literatura e pelos resultados calculados pelo solver KNITRO. A metodologia BD-OTS se mostrou uma ferramenta robusta e promissora, pois apresentou soluções factíveis em todos os sistemas e melhores resultados em nove sistemas. Dois resultados obtidos pela BD-OTS apresentaram valores de função objetivo praticamente iguais aos calculados pelo solver, e melhores que os da literatura. Além disso, as soluções obtidas pela BD-OTS forneceram topo-logias dos sistemas totalmente conexas, sem barras desconectadas das topologias dos sistemas, enquanto as soluções de OTS pelo solver apresentaram topologias desconexas.
The main objective of this work is the implementation of a methodology that, using the scatter search meta-heuristic (SS) solves the problem of optimal power flow (OPF) considering trans-mission switching (TS) to reduce the operation costs. In order to evaluate the potential of the meta-heuristic, the SS algorithm was implemented to optimize constrained multimodal func-tions, a methodology called BD-FMR, and to solve the OPF problem, a methodology called BD-FPO. Eleven constrained multimodal problems available in the specialized literature were solved using the BD-FMR method, and the results obtained are comparable with the best results available in the literature. The OPF problem was solved by the BD-FPO methodology using three test systems with 6, 14, and 57 buses, and the results were not satisfactory when compared to the solutions of the exact formulation of the problem obtained by the KNITRO solver. How-ever, the BD-FPO algorithm served as the basis for the implementation of the main method of this work. Finally, the BD-OTS method was implemented in the C/C ++ programming lan-guage, using parallel programming resources through the OpenMP library. In this work, the formulation used to represent the operation of the grid considers the alternating current (AC) model, which consists of a mixed-integer nonlinear programming (MINLP) problem due to the presence of discrete variables related to the operation state a line, transformer tap position and the operating state of the shunt compensators. In the BD-OTS method, the meta-heuristic man-ages the integer and binary variables, providing these variables to the mathematical model in the objective function calculation step, transforming the model into a nonlinear programming (NLP) problem. The NLP problem is then solved by the KNITRO solver using the interior point meth-od, which returns to the meta-heuristic the values of all the continuous variables of the problem. Eleven systems, two 14-bus, three 30-bus, two 57-bus, two 118-bus, and two 162-bus systems were used in the tests. The solutions of the TS problem by the BD-OTS method were compared with the results of the solutions of these same systems obtained in the literature and the results calculated by the KNITRO solver. The BD-OTS method proved to be a robust and promising tool as it obtained feasible solutions for all systems and better results for nine systems. Two re-sults obtained by the BD-OTS algorithm presented objective function values practically equal to those calculated by the solver, and better than those in the literature. In addition, the solutions obtained by BD-OTS provided topologies with fully connected systems without buses discon-nected from system, while the solutions obtained by the solver presented disconnected topolo-gies.
The main objective of this work is the implementation of a methodology that, using the scatter search meta-heuristic (SS) solves the problem of optimal power flow (OPF) considering trans-mission switching (TS) to reduce the operation costs. In order to evaluate the potential of the meta-heuristic, the SS algorithm was implemented to optimize constrained multimodal func-tions, a methodology called BD-FMR, and to solve the OPF problem, a methodology called BD-FPO. Eleven constrained multimodal problems available in the specialized literature were solved using the BD-FMR method, and the results obtained are comparable with the best results available in the literature. The OPF problem was solved by the BD-FPO methodology using three test systems with 6, 14, and 57 buses, and the results were not satisfactory when compared to the solutions of the exact formulation of the problem obtained by the KNITRO solver. How-ever, the BD-FPO algorithm served as the basis for the implementation of the main method of this work. Finally, the BD-OTS method was implemented in the C/C ++ programming lan-guage, using parallel programming resources through the OpenMP library. In this work, the formulation used to represent the operation of the grid considers the alternating current (AC) model, which consists of a mixed-integer nonlinear programming (MINLP) problem due to the presence of discrete variables related to the operation state a line, transformer tap position and the operating state of the shunt compensators. In the BD-OTS method, the meta-heuristic man-ages the integer and binary variables, providing these variables to the mathematical model in the objective function calculation step, transforming the model into a nonlinear programming (NLP) problem. The NLP problem is then solved by the KNITRO solver using the interior point meth-od, which returns to the meta-heuristic the values of all the continuous variables of the problem. Eleven systems, two 14-bus, three 30-bus, two 57-bus, two 118-bus, and two 162-bus systems were used in the tests. The solutions of the TS problem by the BD-OTS method were compared with the results of the solutions of these same systems obtained in the literature and the results calculated by the KNITRO solver. The BD-OTS method proved to be a robust and promising tool as it obtained feasible solutions for all systems and better results for nine systems. Two re-sults obtained by the BD-OTS algorithm presented objective function values practically equal to those calculated by the solver, and better than those in the literature. In addition, the solutions obtained by BD-OTS provided topologies with fully connected systems without buses discon-nected from system, while the solutions obtained by the solver presented disconnected topolo-gies.
Descrição
Palavras-chave
Busca dispersa, Meta-heurística, Otimização, Fluxo de potência ótimo, Desligamento de linhas de transmissão, Meta-heuristic, Optimal power flow, Optimization, Scatter search, Transmission switching