Análise Orientada a Dados da Infraestrutura Elétrica: Identificando Padrões de Consumo e Irregularidades em Minas Gerais - Brasil

Abstract

Data science and system identification emerge as fundamental themes for optimizing complex processes across various industry sectors. This study proposes a data integration methodology to analyze and identify consumption patterns in the electrical infrastructure of a region in the state of Minas Gerais, Brazil, using data from irregularity reports and network monitoring. Initially, the data was extracted and processed using Extraction, Transformation, and Loading (ETL) techniques, allowing for a detailed understanding of consumption patterns and commercial losses over time. Additionally, georeferenced data was used to map areas with the highest incidence of reports and energy losses. The analysis revealed a direct association between areas with high commercial losses and occurrences of clandestine connections. Regions with losses above 30% were shown to have a higher incidence of irregularities. Furthermore, the behaviors of feeders with high and low commercial losses over time were examined. In feeders with high losses, a greater disparity between measurement and expected consumption was observed, suggesting irregularities in energy distribution. On the other hand, in feeders with low losses, a proximity between measurement and expected consumption curves was observed, indicating a more efficient operation of the electrical network.