Aplicação do método de árvore de decisão para identificação de alterações no padrão de consumo de energia elétrica em sistema de iluminação

Abstract

Electricity consumption was one of the reduction parameters in the industry with potential for cost reduction. As a contribution to this demand, this work presents a generic method for the efficiency of energy systems and identification of changes in the pattern of electric energy consumption of electric systems. A visa proposal to identify the daily periods where there is extra consumption of electricity in a textile industry lighting system with no necessary justification of use. As the identification system, the concept of Artificial Intelligence was used through the Decision Tree method. The platform selected for the implementation of this tool was the Sipina Survey, which allowed an analysis of the collected data, indicating satisfactory responses in the developed system. To validate the proposed method, it was applied to a lighting system of a textile industry, where a data set for system training and another data set of the same dimension were used for learning validation. The data selected for validation of the method were analyzed using the concept of confusion matrix. As a result of this study, use a generic method for power system efficiency and consumption control, coupled with a 32% reduction in power consumption in the system used to validate the tool used, showing the effectiveness of the method.