Spatiotemporal prediction of the number of electric power outages

Registro completo de metadados
MetadadosDescriçãoIdioma
Autor(es): dc.contributorFerreira, Vitor Hugo-
Autor(es): dc.contributorColombini, Angelo Cesar-
Autor(es): dc.contributorPinho, Andre da Costa-
Autor(es): dc.creatorSilva, Ramon Barino-
Data de aceite: dc.date.accessioned2025-01-03T11:42:38Z-
Data de disponibilização: dc.date.available2025-01-03T11:42:38Z-
Data de envio: dc.date.issued2024-07-25-
Data de envio: dc.date.issued2024-07-25-
Fonte completa do material: dc.identifierhttps://app.uff.br/riuff/handle/1/33663-
Fonte: dc.identifier.urihttp://educapes.capes.gov.br/handle/capes/920267-
Descrição: dc.descriptionPower outages are a significant issue that cause financial losses and negatively impact the quality of life for consumers. To better manage these occurrences, energy utilities require effective tools and indicators for decision-making and efficient resource allocation. This work proposes the application of Machine Learning techniques to forecast the monthly volume of power outages based on regional climatic, structural, and electrical factors. The study utilizes publicly available and cost-free data from Brazilian governmental regulatory agencies, such as the National Electric Energy Agency (ANEEL), the National Institute of Meteorology (INMET), and the Brazilian Institute of Geography and Statistics (IBGE). A framework was developed using open-source technologies such as Python and MySQL, inspired by the Data Lakehouse Medallion architecture, to facilitate data extraction, storage, and transformation, resulting in a robust database prepared for analysis and machine learning model training. Multiple Regression and Random Forest models were trained and evaluated using the Mean Absolute Percentage Error (MAPE) metric, and their results were compared with a heuristic strategy for outage prediction. Considering the heterogeneity of the Brazilian national territory and its potential influence on outage occurrences, consumer units were clustered into five distinct groups based on regional characteristics using the K-Means algorithm, enabling tailored model training within each cluster. The results demonstrated consistent model performance, underscoring the framework's effectiveness across diverse Brazilian regions-
Descrição: dc.description106 f.-
Formato: dc.formatapplication/pdf-
Idioma: dc.languageen-
Direitos: dc.rightsOpen Access-
Direitos: dc.rightsCC-BY-SA-
Palavras-chave: dc.subjectMachine Learning-
Palavras-chave: dc.subjectPower Outages-
Palavras-chave: dc.subjectPower System-
Palavras-chave: dc.subjectAprendizado de máquina-
Palavras-chave: dc.subjectSistema de transmissão de energia-
Palavras-chave: dc.subjectGerenciamento de energia-
Título: dc.titleSpatiotemporal prediction of the number of electric power outages-
Tipo de arquivo: dc.typeTrabalho de conclusão de curso-
Aparece nas coleções:Repositório Institucional da Universidade Federal Fluminense - RiUFF

Não existem arquivos associados a este item.