Induction Motor Failure Analysis using Machine Learning and Infrared Thermography

Abstract

Induction motor are electrical machines used in a wide variety of industrial applications. However, due to their applications, are subjected to undesirable operating conditions. A complementary technique that aids in fault diagnosis in induction motors is infrared thermography. This paper proposes a methodology based on automatic learning and unsegmented infrared imaging for classifies and diagnosis failures on induction motor and their kinematic chain. The proposed methodology is analyzing the unsegmented infrared thermography, taking directly from the thermogram significant statistical features that describe the thermal behavior of the electromechanical system, to later reduce the set of characteristics and, through a machine learning algorithm, classify the fault condition. To demonstrate the efficiency of the proposed methodology, this paper presents the health condition analysis and three fault conditions in an induction motor: a broken rotor bar, bearing damage, and misalignment.