Fault Detection in Electric Motors using Acoustic Signals and Image Classification

The classification of sound signals can be applied to the fault diagnosis of electromechanical devices and systems, such as electrical machines, drives, production machines, etc. Traditional methods identify acoustic signals by analyzing their characteristics in both the time and frequency domains. This paper focuses on one of the latest trends for visualizing sound signals which can then be used by image classification models for fault detection in electrical machines and drives. The experiment described uses an electric motors sound recording dataset together with Domain Generalization approach based on Symmetrized Dot Pattern. Image classification is utilized to distinguish the different conditions of the electric motors. The results suggest that the presented approach could be a fast and cost-effective method for electrical machine diagnostics. Despite the promising outcomes, further work can be done based on more versatile data and different image classification models relying on autoencoders or convolutional neural networks together with large language models.