Análise estatística para seleção dos atributos que caracterizam os indicadores de continuidade do fornecimento de energia elétrica

Abstract

The quality of electricity (QEE) supplied by electricity distributors is regulated by the National Electric Energy Agency (ANEEL) in 3 main aspects: service quality, product quality and service quality. With regard to service quality, service continuity indicators and limits for these indicators are established. The current methodology for defining the limits is described in Technical Note no102/2014-SRD/ANEEL. This methodology is constantly updated since Brazil has an extensive territory and, therefore, determining the variables that are related to interruptions in energy supply becomes a non-trivial task. In this context, the present work aims to analyze the change in the statistical analysis criteria for selecting the attributes that characterize the continuity indicators. The proposed analysis is based on the same procedures used by ANEEL to define the current model for characterizing continuity indicators, however, statistical techniques different from those employed by ANEEL are used. To validate the proposed methodology, eight case studies were developed using the software SAS University Edition. In these case studies, the change in the correlation limit and the variance inflation factor (VIF) were evaluated, as well as the change in the regression technique. It was found that, using Stepwise regression, the increase in the correlation threshold tends to decrease the explanatory power and the number of variables in the models, however, it increases the partial explanatory power of each variable present in the models.