Citation: Zheng, J.; Liao, J.; Chen, Z.
End-to-End Continuous/Discontinuous
Feature Fusion Method with
Attention for Rolling Bearing Fault
Diagnosis. Sensors 2022, 22, 6489.
https://doi.org/10.3390/s22176489
Academic Editors: Kim Phuc Tran,
Athanasios Rakitzis and Khanh T.
P. Nguyen
Received: 1 August 2022
Accepted: 21 August 2022
Published: 29 August 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Article
End-to-End Continuous/Discontinuous Feature Fusion Method
with Attention for Rolling Bearing Fault Diagnosis
Jianbo Zheng
1,2
, Jian Liao
1,2,
* and Zongbin Chen
1,2
1
Institute of Vibration and Noise, Naval University of Engineering, Wuhan 430033, China
2
Naval Key Laboratory of Ship Vibration and Noise, Naval University of Engineering, Wuhan 430033, China
* Correspondence: jl_zss@163.com
Abstract:
Mechanical equipment failure may cause massive economic and even life loss. Therefore,
the diagnosis of the failures of machine parts in time is crucial. The rolling bearings are one of
the most valuable parts, which have attracted the focus of fault diagnosis. Many successful rolling
bearing fault diagnoses have been made based on machine learning and deep learning. However,
most diagnosis methods still rely on complex signal processing and artificial features, bringing many
costs to the deployment and migration of diagnostic models. This paper proposes an end-to-end
continuous/discontinuous feature fusion method for rolling bearing fault diagnosis (C/D-FUSA).
This method comprises long short-term memory (LSTM), convolutional neural networks (CNN)
and attention mechanism, which automatically extracts the continuous and discontinuous features
from vibration signals for fault diagnosis. We also propose a contextual-dependent attention module
for the LSTM layers. We compare the method with the other simpler deep learning methods and
state-of-the-art methods in rolling bearing fault data sets with different sample rates. The results
show that our method is more accurate than the other methods with real-time inference. It is also
easy to be deployed and trained in a new environment.
Keywords: fault diagnosis; rolling bearing; deep learning; LSTM; CNN; attention
1. Introduction
The fault diagnosis of mechanical equipment is vital in modern industry. Once the
failure of mechanical equipment occurs, it will cause huge damage to the economy and
property and even bring casualties. Therefore, finding a better fault diagnosis method is
necessary to ensure the normal operation of the machine [
1
]. Rolling bearing is the most
commonly used part in mechanical equipment, known as the joint of the machinery. It has
the advantages of high efficiency, small friction resistance, convenient assembly and easy
lubrication, so it is widely used in rotating machinery. As one of the core components of
rotating machinery, such as gearbox and turbine machinery, the health of rolling bearings
significantly influences the machine’s stability and life [
2
]. In the process of working, rolling
bearings may be damaged by the outer raceway, inner raceway and rolling body due to
lubricant pollution, overload and other reasons. Therefore, an effective fault diagnosis
method is crucial to the stability of rolling bearings [3].
Most fault diagnosis methods used for rolling bearings are based on vibration signals.
By detecting and analyzing the vibration data of rolling bearings, this method can diagnose
various faults in real time [
4
]. Model-based [
5
] and data-driven [
6
] methods can diagnose
rolling bearing faults based on vibration data. Data-driven methods generally use machine
learning to learn bearing vibration data and identify different types of fault modes. This
method can effectively and quickly process mechanical signals, requires less prior expertise,
and can provide an accurate diagnosis [
7
]. It has become a common method for rolling
bearing fault diagnosis. K-nearest neighbor (KNN), support vector machine (SVM), self-
organizing mapping (SOM) networks and other machine learning algorithms have been
Sensors 2022, 22, 6489. https://doi.org/10.3390/s22176489 https://www.mdpi.com/journal/sensors
