IEEE Computational Intelligence Magazine - August 2021 - 79

Bo Peng
North China Electric Power University, CHINA
Application
Notes
Ying Bi, Bing Xue, Mengjie Zhang
Victoria University of Wellington, NEW ZEALAND
Shuting Wan
North China Electric Power University, CHINA
Multi-View Feature Construction Using Genetic Programming
for Rolling Bearing Fault Diagnosis
Abstract
R
olling bearing fault diagnosis
is an important task in
mechanical engineering.
Existing methods have several limitations,
such as requiring domain
knowledge and a large number of
training samples. To address these
limitations, this paper proposes a
new diagnosis approach, i.e., multiview
feature construction based on
genetic programming with the idea
of ensemble learning (MFCGPE),
to automatically construct high-level features
from multiple views and build an
effective ensemble for identifying different
fault types using a small number of training
samples. The MFCGPE approach uses
a new program structure to automatically
construct a flexible number of features
from every single view. A new fitness
function based on accuracy and distance is
developed in MFCGPE to improve the
discriminability of the constructed features.
To further improve the generalization
performance, an ensemble of classifiers
based on k-nearest neighbor is created by
using the constructed features from every
single view. Three bearing datasets and 19
competitive methods are used to validate
the effectiveness of the new approach. The
Digital Object Identifier 10.1109/MCI.2021.3084495
Date of current version: 15 July 2021
failures and the percentage of failures
caused by rolling bearing damage
reaches nearly 30% [3]. Therefore, it
is important to identify the faults of
rolling bearing to monitor bearing
status, ensure machinery safety, reduce
economic losses, and avoid casualties.
In recent years, many methods
©SHUTTERSTOCK.COM/SERGEY TARASOV
results show that MFCGPE achieves
higher diagnostic accuracy than all the
compared methods on the three datasets
with a small number of training samples.
I. Introduction
Rolling bearings are important supporting
equipment and have been used in all
kinds of rotating machinery, such as electric
motors, turbine generators, and highspeed
trains [1]. The performance of
rolling bearings can be affected by the
high temperature, high pressure, and alternating
load during the equipment operation
[2]. The rolling bearings are often
inevitably damaged as the running time
increases [2]. The damage of rolling bearings
will lead to mechanical equipment
Corresponding Author: Ying Bi (e-mail: ying.bi@ecs
.vuw.ac.nz).
have been developed for rolling
bearing fault diagnosis. Because the
vibration signals are easy to collect
and often contain information of
the running states, the studies on
using vibration signals for fault diagnosis
have gained much attention. The collected
vibration signals often have background
noise or information loss. A
variety of signal processing methods, such
as wavelet transform [4], empirical mode
decomposition [5], fast spectral kurtosis
[6], maximum correlated kurtosis deconvolution
[7], and variational mode decomposition
[8], have been used to suppress
the interference of noise and harmonics
and strengthen signal characteristics.
Experts have conducted the spectrum
analysis on the processed signals and recognized
the fault characteristic frequency
for fault diagnosis [9]. However, these
methods do not provide satisfactory results
and require extensive domain expertise,
which is time-consuming and expensive.
To automatically identify faults,
methods based on machine learning
1556-603X/21©2021IEEE
AUGUST 2021 | IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE 79
http://www.SHUTTERSTOCK.COM/SERGEY
IEEE Computational Intelligence Magazine - August 2021

Table of Contents for the Digital Edition of IEEE Computational Intelligence Magazine - August 2021
