Identificação de barras quebradas em motores de indução através de parâmetros característicos e SVM

Abstract

Triphasic induction machines (MIT) are the backbone in modern electromechanical conversion of energy, however, not even those simple and robust machines are completely failsafe, making a non-invasive damage identification method required, to make possible preemptive maintenance and avoid financial losses. The goal of this research is to create a database of electrical current signals extracted from an MIT, with and without rotoric damage to analyse the viability of using characteristic parameters and support vector machine (SVM) classifier to identify broken bars in the squirrel cage rotor.