Estudo de um algoritmo de mineração de dados aplicado à avaliação de curvas de consumo de energia elétrica

Abstract

Nowadays, the quantity of inherent information to our environment is huge. For treat these information appeared an area called Data Mining. Inside of this area, there are some algorithms to carry out the knowledge discovery in Data Bases. Some examples of algorithms can be mentioned, such as: Neural Networks, Fuzzy Logic, Genetic Algorithm, Bayesian Algorithm, and Induction Tree. In this work topics related with Data Mining were studied, with the purpose of apply in problems related with Electrical Engineering. The Naive Bayes algorithm was chosen with objective of recognize electrical energy consumption patterns, as an alternative to Artificial Neural Networks that are remembered in pattern recognition problems. Initially, some concepts of statistics were analyzed, and subsequently the concepts of the algorithm Naive Bayes are presented. The original algorithm was modified with the purpose of recognize load curves of industrial, commercial, residential consumers in Electrical Power Systems. A computational package, called Weka, has been used. This package implemented was implemented in Java and has several data mining algorithms.