Classificação de defeitos de barras quebradas de rotor em motores de indução trifásicos acionados por inversores de frequência utilizando transformada wavelet

Abstract

Due to the important role played by the induction motors in the industry it is necessary to develop predictive tools that offer support for the correct diagnosis and classification of failures still in the initial phase, allowing an increase in productivity and reducing of the economic losses generated by unwanted stops. This work presents a study related the failures of broken rotor bars in three-phase induction motors driven by different frequency inverters. For the pre-processing of the current signals using the Discrete Wavelet Transform, where they are applied directly to the signs of the stator currents. For the diagnosis and the classification of the severity level of faults are used and compared two different techniques of machine learning: (i) Decision Trees and (ii) Random Forest.