Programação genética aplicada à identificação de acidentes de uma usina nuclear PWR

Detalhes bibliográficos
Ano de defesa: 2018
Autor(a) principal: Pinheiro, Victor Henrique Cabral
Orientador(a): Não Informado pela instituição
Banca de defesa: Não Informado pela instituição
Tipo de documento: Dissertação
Tipo de acesso: Acesso aberto
Idioma: por
Instituição de defesa: Universidade Federal do Rio de Janeiro
Brasil
Instituto Alberto Luiz Coimbra de Pós-Graduação e Pesquisa de Engenharia
Programa de Pós-Graduação em Engenharia Nuclear
UFRJ
Programa de Pós-Graduação: Não Informado pela instituição
Departamento: Não Informado pela instituição
País: Não Informado pela instituição
Palavras-chave em Português:
Link de acesso: http://hdl.handle.net/11422/13165
Resumo: This work presentes the results of the study that evaluated the efficiency of the evolutionary computation algorithm genetic programming as a technique for the optimization and feature generation at a pattern recognition system for the diagnostic of accidents in a pressurized water reactor nuclear power plant. The foundations of a typical pattern recognition system, the state of the art of genetic programming and of similar accident/transient diagnosis systems at nuclear power plants are also presented. Considering the set of the time evolution of seventeen operational variables for the three accident scenarios approached, plus normal condition, the task of genetic programming was to evolve non-linear regressors with combination of those variables that would provide the most discriminatory information for each of the events. After exhaustive tests with plenty of variable associations, genetic programming was proven to be a methodology capable of attaining success rates of, or very close to, 100%, with quite simple parametrization of the algorithm and at very reasonable time, putting itself in levels of performance similar or even superior as other similar systems available in the scientific literature, while also having the additional advantage of requiring very little pretreatment (sometimes none at all) of the data