Intelligent fault detection using raw vibration signals via dilated convolutional neural networks

Mohammad Azam Khan, Yong Hwa Kim, Jaegul Choo

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Fault detection and diagnosis is critical to improve the reliability and availability in induction motors (IMs). Machine learning and deep learning techniques have been widely used in induction motor fault detection and diagnosis. In this paper, we propose a new deep learning model based on a dilated convolutional neural network (D-CNN) for detecting bearing faults in IMs. The proposed model works directly on raw vibration signals without any hand-crafted feature extraction process. Our model can incorporate global context without losing important local information by stacking dilated convolutions with an increasing width. Numerical results show that the proposed D-CNN is not only capable of classifying normal signals perfectly but also can achieve higher accuracy than conventional techniques under noisy environments.

Original languageEnglish
JournalJournal of Supercomputing
DOIs
Publication statusAccepted/In press - 2018 Jan 1

Fingerprint

Vibration Signal
Induction Motor
Fault Detection
Fault detection
Induction motors
Fault Detection and Diagnosis
Neural Networks
Neural networks
Failure analysis
Bearings (structural)
Stacking
Convolution
Feature Extraction
Learning systems
Feature extraction
Machine Learning
High Accuracy
Fault
Availability
Model-based

Keywords

  • Convolutional neural networks
  • Deep neural networks
  • Dilated convolution
  • Intelligent fault detection
  • Vibration signals

ASJC Scopus subject areas

  • Software
  • Theoretical Computer Science
  • Information Systems
  • Hardware and Architecture

Cite this

Intelligent fault detection using raw vibration signals via dilated convolutional neural networks. / Khan, Mohammad Azam; Kim, Yong Hwa; Choo, Jaegul.

In: Journal of Supercomputing, 01.01.2018.

Research output: Contribution to journalArticle

@article{85cdd4358c644e4e9879de75f56747db,
title = "Intelligent fault detection using raw vibration signals via dilated convolutional neural networks",
abstract = "Fault detection and diagnosis is critical to improve the reliability and availability in induction motors (IMs). Machine learning and deep learning techniques have been widely used in induction motor fault detection and diagnosis. In this paper, we propose a new deep learning model based on a dilated convolutional neural network (D-CNN) for detecting bearing faults in IMs. The proposed model works directly on raw vibration signals without any hand-crafted feature extraction process. Our model can incorporate global context without losing important local information by stacking dilated convolutions with an increasing width. Numerical results show that the proposed D-CNN is not only capable of classifying normal signals perfectly but also can achieve higher accuracy than conventional techniques under noisy environments.",
keywords = "Convolutional neural networks, Deep neural networks, Dilated convolution, Intelligent fault detection, Vibration signals",
author = "Khan, {Mohammad Azam} and Kim, {Yong Hwa} and Jaegul Choo",
year = "2018",
month = "1",
day = "1",
doi = "10.1007/s11227-018-2711-0",
language = "English",
journal = "The Journal of Supercomputing",
issn = "0920-8542",
publisher = "Springer Netherlands",

}

TY - JOUR

T1 - Intelligent fault detection using raw vibration signals via dilated convolutional neural networks

AU - Khan, Mohammad Azam

AU - Kim, Yong Hwa

AU - Choo, Jaegul

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Fault detection and diagnosis is critical to improve the reliability and availability in induction motors (IMs). Machine learning and deep learning techniques have been widely used in induction motor fault detection and diagnosis. In this paper, we propose a new deep learning model based on a dilated convolutional neural network (D-CNN) for detecting bearing faults in IMs. The proposed model works directly on raw vibration signals without any hand-crafted feature extraction process. Our model can incorporate global context without losing important local information by stacking dilated convolutions with an increasing width. Numerical results show that the proposed D-CNN is not only capable of classifying normal signals perfectly but also can achieve higher accuracy than conventional techniques under noisy environments.

AB - Fault detection and diagnosis is critical to improve the reliability and availability in induction motors (IMs). Machine learning and deep learning techniques have been widely used in induction motor fault detection and diagnosis. In this paper, we propose a new deep learning model based on a dilated convolutional neural network (D-CNN) for detecting bearing faults in IMs. The proposed model works directly on raw vibration signals without any hand-crafted feature extraction process. Our model can incorporate global context without losing important local information by stacking dilated convolutions with an increasing width. Numerical results show that the proposed D-CNN is not only capable of classifying normal signals perfectly but also can achieve higher accuracy than conventional techniques under noisy environments.

KW - Convolutional neural networks

KW - Deep neural networks

KW - Dilated convolution

KW - Intelligent fault detection

KW - Vibration signals

UR - http://www.scopus.com/inward/record.url?scp=85058101941&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85058101941&partnerID=8YFLogxK

U2 - 10.1007/s11227-018-2711-0

DO - 10.1007/s11227-018-2711-0

M3 - Article

AN - SCOPUS:85058101941

JO - The Journal of Supercomputing

JF - The Journal of Supercomputing

SN - 0920-8542

ER -