2026-01-092026-01-092021-10-06LIMA, Fabrício Silva. Modelo de predição de tensão crítica em isoladores de distribuição. Orientador: Reinaldo Corrêa Leite. 2021. Trabalho de Curso (Bacharelado em Engenharia Elétrica e Biomédica) – Faculdade de Engenharia Elétrica e Biomédica, Instituto de Tecnologia, Universidade Federal do Pará, Belém, 2021. Disponível em:https://bdm.ufpa.br/handle/prefix/9048 . Acesso em:.https://bdm.ufpa.br/handle/prefix/9048This paper aims to develop a model in machine learning (ML) aiming to improve predictive detection techniques for failures of electrical insulators in the distribution network, seeking to find the best performance of these devices in the Brazilian electrical sector. Initially, a theoretical study is carried out on the types of electrical insulators found in the distribution network, their characteristics, physical format and voltage class. Then, the different types of pollution that these insulators can suffer during their useful life are discussed, and the consequences, such as the appearance of partial discharges that can develop into the flashover phenomenon, causing problems such as networks shutdown and high maintenance costs. Subsequently, the work deals with artificial intelligence algorithms, explaining their concepts and the way they act on the datasets they receive. The work is limited to supervised machine learning algorithms. Thus, it is detailed how they can be applied to solve the work in question. Finally, the studied algorithms are applied in the dataset referring to weather conditions, leakage current and critical voltage to establish test results that can predict future problems caused by the flashover phenomenon in distribution insulators.Acesso AbertoIsoladores elétricosrede de distribuiçãopoluiçãotensão crítica,algoritmos de inteligência artificial.Electrical insulatorsdistribution networkpollutioncritical tensionmachine learningartificial intelligence algorithmsCNPQ::ENGENHARIAS::ENGENHARIA ELETRICATrabalho de Curso - Graduação - MonografiaAttribution-NonCommercial-NoDerivs 3.0 Brazil