Robust Feature Extraction on Vibration Data under Deep-Learning Framework: An Application for Fault Identification in Rotary Machines

Abstract

Mechanical failures in rotating machinery (e.g. wind turbines, generators, motor-derives etc.) may result in catastrophic failures. Different mechanical faults induce characteristic vibrations in the equipment structure. Online vibration monitoring helps mitigate catastrophic failures through early detection and identification of underlying mechanical faults. However, extracting characteristic vibration features that improve fault classification performance and are robust to various noises in the vibration signals is a challenging task. Various statistical and signal processing-based vibrationfeatures have been proposed in the literature. These vibrationfeatures were devised on the basis of prior knowledge about characteristics of vibration signals from different fault types. Recently, automatic feature extraction through unsupervised learning in deep neural architectures has resulted in state of art performance on image and speech recognition tasks. So, Instead of feature-engineering, we, here, hypothesized that feature learning on raw vibration signal possibly will extract vibration-features that can improve fault identification performance of subsequent classifier. To the purpose, we explored Convolutional Neural Network for unsupervised feature learning on vibration signals and Denoising AutoEncoder for extracting vibration features that are robust and invariant to the noises in vibration signals. We proposed a Hybrid deep-model consisting of a Multi-channel Convolutional Neural Network followed by a stack of Denoising Auto-Encoders (MCNN-SDAE) with a single classification layer at the top. We compared the fault identification and classification performance of the proposed model with other models employing tradition statistical and signal processing based vibration-features. We validated the performance of all models on a benchmark vibration data collected from an experimental test-rig specifically designed to study vibration characteristics of bearing related faults.

Fault Diagnosis of Rotary Machines Using Deep Convolutional Neural Network with Wide Three Axis Vibration Signal Input

“…They also defined the process of converting a signal to a 2D grayscale image as a way of preprocessing. Further on, Shaheryar et al [ 28 ] explored CNN in fault identification with spectrograms of vibration images previously converted using short time Fourier transform. Specifically, they used a convolutional neural network in combination with an autoencoder network and a fully coupled classification layer to identify the stages of bearing damage on an available dataset.…”

Aircraft Gearbox Fault Diagnosis System: An Approach based on Deep Learning Techniques

“…An adaptive multi-sensor data fusion method based on Deep Convolutional Neural Networks (DCNN) was proposed for fault diagnosis in gearbox systems [13]. Hybrid-deep model was proposed to diagnose faults in rotary machines [14]. This model consists of multi-channel CNN followed by stack of denoising encoders.…”

A comprehensive review of artificial intelligence-based approaches for rolling element bearing PHM: shallow and deep learning

“…In [220], a locality preserving projection (LPP) was adopted to fuse the deep features, and thus to build a new deep AE method constructed with a denoising autoencoder (DAE) and a contractive autoencoder (CAE) for the enhancement of featurelearning ability with the goal of diagnosing REB faults. A hybrid deep model consisting of a multi-channel CNN followed by a stack of denoising autoencoders (MCNN-SDAE) was developed by A. Shaheryar et al, [221] for fault identification in rotary machines. In the study, these researchers explored the MCNN for unsupervised feature learning on vibration signals and SDAE for extracting vibration features that are robust and invariant to the noises in vibration signals.…”

“…The possibility of the sparse representation FDD [158], [207], [208][209][210][211][212][213][214][215][216][217][218][219], [221][222][223][224][225] Fault prognosis [220] pros and cons of these SL-based REB PHM, and DL-based REB PHM methods, and their advancements and applications were reviewed and summarized. From this survey study, several key points can be concluded, including:…”