Thermal imaging fault diagnosis of three-phase induction motors using neural networks

Abstract

The article presents a technique for diagnosing faults in three-phase induction motors. It uses two thermal imaging cameras and a novel method called Differences of Color Thermal Images (DoCTI). Eight three-phase induction motors (TPIMs) were analyzed: four 550 W motors and four 500 W motors, under the following conditions: healthy, faulty squirrel cage ring, one broken bar, two broken bars, and three broken bars. Thermographic measurements were conducted with thermal camera vibrations ranging from 0 to 1.2 meters per second squared. A novel feature extraction method for color thermal images (DoCTI) was proposed. Three neural networks, NnetV04, NnetV05, and NnetV06, were presented. Convolutional neural networks were used to analyze the thermal images. High accuracy recognition of motor fault conditions was achieved. The computed results confirm the effectiveness of the proposed approach for the recognition of electrical faults of three-phase induction motors.