Multi-input CNN based vibro-acoustic fusion for accurate fault diagnosis of induction motor

Induction motor (IM) is a highly efficient prime mover in industrial applications. To maintain an uninterrupted operation, accurate fault diagnosis system of IM is required. It can help to improve operational safety and prevent unexpected economic losses. The traditional diagnosis methods are less capable of dealing with real-time and varying working environments. This paper presents a vibro-acoustic fusion technique for an accurate fault diagnosis under varying working conditions. The suggested method fuses the features of vibration and acoustic signals using Multi Input-Convolutional Neural Network (MI-CNN) technique. At first, raw vibration and acoustic signals are acquired at varying speeds and converted into a time–frequency spectrum using the Constant Q-Non-Stationary Gabor Transform (CQ-NSGT). Thereafter, a MI-CNN-based vibro-acoustic fusion is adopted for the fusion of vibration and acoustic features. Six distinct motor conditions are utilized to compute the effectiveness of the suggested MI-CNN model. Further, two additional datasets, i.e., bearing and the gearbox datasets, are employed to validate the suggested approach. The experimental results demonstrate that the suggested methodology is accurate and reliable for IMs and other components of rotating machine.