www.sciencedirect.com Open in urlscan Pro
162.159.137.70  Public Scan

Form analysis
1 forms found in the DOM

GET /search#submit

<form class="QuickSearch" action="/search#submit" method="get" aria-label="form">
  <div class="search-input">
    <div class="search-input-container search-input-container-no-label"><label class="search-input-label u-hide-visually" for="article-quick-search">Search ScienceDirect</label><input type="search" id="article-quick-search" name="qs" value=""
        class="search-input-field" aria-label="Search ScienceDirect" aria-describedby="article-quick-search-description-message" placeholder="Search ScienceDirect"></div>
    <div class="search-input-message-container">
      <div aria-live="polite" class="search-input-validation-error"></div>
      <div id="article-quick-search-description-message"></div>
    </div>
  </div><button type="submit" class="button small u-margin-xs-left button-primary button-icon-only" aria-disabled="false" aria-label="Submit search"><svg focusable="false" viewBox="0 0 100 128" height="20" width="18.75" class="icon icon-search">
      <path
        d="m19.22 76.91c-5.84-5.84-9.05-13.6-9.05-21.85s3.21-16.01 9.05-21.85c5.84-5.83 13.59-9.05 21.85-9.05 8.25 0 16.01 3.22 21.84 9.05 5.84 5.84 9.05 13.6 9.05 21.85s-3.21 16.01-9.05 21.85c-5.83 5.83-13.59 9.05-21.84 9.05-8.26 0-16.01-3.22-21.85-9.05zm80.33 29.6l-26.32-26.32c5.61-7.15 8.68-15.9 8.68-25.13 0-10.91-4.25-21.17-11.96-28.88-7.72-7.71-17.97-11.96-28.88-11.96s-21.17 4.25-28.88 11.96c-7.72 7.71-11.97 17.97-11.97 28.88s4.25 21.17 11.97 28.88c7.71 7.71 17.97 11.96 28.88 11.96 9.23 0 17.98-3.07 25.13-8.68l26.32 26.32 7.03-7.03">
      </path>
    </svg></button><input type="hidden" name="origin" value="article"><input type="hidden" name="zone" value="qSearch">
</form>

Text Content

JavaScript is disabled on your browser. Please enable JavaScript to use all the
features on this page. Skip to main contentSkip to article
ScienceDirect
 * Journals & Books

 * 
 * Search

RegisterSign in

 * Access through your institution
 * Purchase PDF

Search ScienceDirect



ARTICLE PREVIEW

 * Abstract
 * Introduction
 * Section snippets
 * References (44)


JOURNAL OF MANUFACTURING SYSTEMS

Volume 68, June 2023, Pages 670-679




PHYSICS VERIFICATION AND VALIDATION FOR TRANSFERRING DATA BETWEEN BEARINGS

Author links open overlay panelEthan Wescoat, Mihir Bangale, Vinita Jansari,
Laine Mears
Show more
Add to Mendeley
Share
Cite
https://doi.org/10.1016/j.jmsy.2023.05.017Get rights and content




ABSTRACT

Bearings are critical components for transferring load and motion between
subsystems with reduced friction for rotational equipment. Manufacturers
implement condition monitoring technology to prevent failures of bearings by
using different data acquisition methods, such as vibration, acoustics,
temperature, motor current, and ultrasonic sensing, to monitor and predict
changes that could indicate early equipment degradation. However, the data
quality and availability varies depending on the application, and low data
availability from physical environments can lead to poorly trained models. This
paper explores how to transfer data and information about failure propensity
between bearings of different sizes using a combined physics and data-driven
approach. Though the approach is exemplified with roller bearings, the method is
extensible to other types of failures in different-sized equipment. Data are
generated using an experimental test stand measuring failure of one component,
with the intent to scale findings to represent a real-world system; findings are
verified with a physics-based model. Data from three different bearing sizes,
i.e. 6205, 6206, and 6207, are used to train the algorithms to identify
similarity among the sets. Classifiers trained with the raw data provide over
90% accuracy, leading to the conclusion that the data classes are separable
based on bearing size. 6205 and 6207 data were scaled to simulate 6206 and
tested to see if a classifier could differentiate between the true 6206 bearing
data and the simulated bearing data derived from the 6205 and 6207 data. In the
simulated data, the classifier accuracy for each algorithm dropped below 90% to
as low as 50% (Naive Bayes case). The lower accuracy implied a greater overlap
in the vibration features, increasing the data similarity between the different
bearing sizes. Future work will further investigate defining dimensionless
numbers as scaling parameters for bearing data.


INTRODUCTION

Rolling bearings are an integral part of rotational equipment, which operates
under extreme environmental factors like high temperature, high pressure, and
high load [1]. Prognostics and Health Management (PHM) [2] and Condition-based
Maintenance (CbM) utilize past and present equipment information to detect
lifetime degradation, diagnose the failure symptoms, and predict and manage
failures [3]. CbM has replaced the standard strategy for most industries, which
was formerly “fix it when it breaks” [4]. The biggest problem associated with
this corrective strategy was the aspect of unexpected downtime, which negatively
impacted manufacturing competitiveness. Unexpected downtime reporting varies
based on the industry and failure type; however, it is a substantial cost in
terms of time and money. As a matter of perspective, Rao et al. [5] found that
approximately 14 million hours of downtime occurred across the United States
economy in industrial and commercial facilities due to unexpected failures in
industrial motors alone. Thomas et al. [6] found that approximately $119.1
billion was attributable to losses from unexpected, preventable downtime. Hence,
there is the potential for significant cost and time savings by reducing or
eliminating unexpected downtime.

Fatigue damage in rolling bearings is one of the most commonly observed failure
modes, which tends to develop gradually and progressively until it leads to a
structural or system failure [7]. Fatigue monitoring and early detection helps
create robust predictive maintenance strategies and are some of the main foci
for PHM. There are several different methods and models established to identify
fatigue defects in rotodynamic equipment, including motors [8], gearboxes [9],
turbines [10], and bearings [11]. In recent years, these methods have
transitioned from physics-based applications towards adopting more data-driven
measures to approximate the physical nature of the failure. Using data-driven
methods helps to reduce the high computational cost associated with physics
modeling through approximation. However, accurate predictions require high data
quality and quantity which is challenging to acquire in engineering
environments. Physics-based methods can improve this accuracy; however, the
proper way to blend physics-based and pure data methods remains an open
question.

Data scarcity is a constant concern for researchers in condition monitoring and
prognostic and health management. Several researchers have proposed approaches
using simulated or artificially computationally generated data to overcome this
challenge. Several synthetic data sets have been generated and made publicly
available for model training [12]. Akrim et al. [3] proposed a framework for
generating training data sets using Paris–Erdogan’s crack growth model.
Additionally, others have used sampling techniques to increase the amount of
data in their minority classes [13] or through simplified classifiers in
one-class classification models [14]. However, some of the underlying physics
knowledge and information that helps identify defects are lost during the
adoption of these data-driven models. The paradigm has been noted outside
rotational equipment with the adoption of physics and data-driven models for
battery safety [15]. This concern is further compounded by the lack of available
failure data to correctly identify failure modes for certain equipment as noted
in condition-based monitoring reviews [16]. A low amount of failure data could
stem from incorrectly labeled data or low failure occurrences. Data augmentation
[17] and transfer learning [18] help reduce the data gaps that occur with low
amounts of failure data. However, data augmentation generally relies on an
existing data source for augmentation. Transfer learning relies on system
similarity between the source and target data for model transfer. The success
chances for data augmentation and transfer learning efforts are reduced if there
is not an existing data source or if there is little system similarity.

This paper investigates a novel approach by incorporating physics-based methods
for transferring data between bearings to increase training data for condition
monitoring systems. The physics incorporation verifies the collected data with a
physics simulation prior to transfer. The case study presented investigates
three different bearing sizes for a specific operating condition. The generated
data are collected offline and validated with a bearing physics model. Using the
physics model as a filter removes potential noise and outliers from the data.
Unsupervised and supervised learning techniques are then applied to the bearing
data to determine if there is an underlying similarity in the bearing data.
Bearing data similarity refers to how much the data from one bearing class
resembles a different bearing class. A set of scaling factors and parameters for
translating one data set for representing a different equipment configuration is
derived based on the analysis of the learning approaches. A discussion is
provided to describe the results, followed by conclusions and future work.


SECTION SNIPPETS


EXPERIMENTAL APPARATUS

Fig. 1 depicts the designed bearing test stand for generating data sets. The
test block houses the bearing under test and connects to the support block. The
support block holds the drive shaft connected to a 1hp motor. The loading arm
applies the bearing force by adding weights to the loading arm and contacting
the bearing block. Tests were conducted to collect bearing data under different
stages of damage with different speed and load combinations to study their
effects on the different


RESULTS & DISCUSSION

Fig. 4 shows the MATLAB© 2022 Simulink response from each tested bearing after
reaching steady-state conditions. A difference between Mishra et al. [44] and
the model response in Fig. 4 is the removal of the system response noise. The
removal of the system noise accounts for why the bearing vibration appears flat
with the listed parameters. The ODE 45 solver was used to solve and generate the
bearing data as conducted by Mishra et al. [44]. Based on the system response
from the parameters in


CONCLUSION

The primary focus of the paper was to explore and validate the similarity trends
in vibration data among three different bearing sizes: 6205, 6206, and 6207.
These bearings were initially chosen based on their common use in research
literature and similarity in the ratio of width, inner, and outer race. Bearing
data were collected from 11 bearings across the three sizes under different
operating conditions. For the initial scope of experimentation, the operating
conditions of 1687 rpm and 1049 N


FUTURE WORK

Further model tuning for the physical analysis will take place to identify a
series of scaling factors that allow for bearing data scaling between similar
systems. In addition, the other collected operating conditions will be examined
to see if the established trends hold true under different operating conditions.
Additionally, bearing failure data will be added to the analysis. With the
failure analysis, frequency features can be considered for the similarity
comparison, provided it is


DECLARATION OF COMPETING INTEREST

The authors declare that they have no known competing financial interests or
personal relationships that could have appeared to influence the work reported
in this paper.


ACKNOWLEDGMENTS

The authors would like to thank Gary Mathis for fabricating some of the test
stand parts, and Ranveer Patil and Josh Bradford for collecting data. The
authors would also like to thank the Department of Defense for their support of
this work through OUSD/R&E (The Under Secretary of Defense-Research and
Engineering), National Defense Education Program (NDEP) / BA-1, Basic Research.

Recommended articles



REFERENCES (44)

 * VasudevanA.K. et al.
   
   
   FATIGUE DAMAGE ANALYSIS: ISSUES AND CHALLENGES
   
   
   INT J FATIGUE
   
   (2016)
 * WangTianyang et al.
   
   
   VIBRATION BASED CONDITION MONITORING AND FAULT DIAGNOSIS OF WIND TURBINE
   PLANETARY GEARBOX: A REVIEW
   
   
   MECH SYST SIGNAL PROCESS
   
   (2019)
 * StetcoAdrian et al.
   
   
   MACHINE LEARNING METHODS FOR WIND TURBINE CONDITION MONITORING: A REVIEW
   
   
   RENEW ENERGY
   
   (2019)
 * CerradaMariela et al.
   
   
   A REVIEW ON DATA-DRIVEN FAULT SEVERITY ASSESSMENT IN ROLLING BEARINGS
   
   
   MECH SYST SIGNAL PROCESS
   
   (2018)
 * ZhangYuyan et al.
   
   
   IMBALANCED DATA FAULT DIAGNOSIS OF ROTATING MACHINERY USING SYNTHETIC
   OVERSAMPLING AND FEATURE LEARNING
   
   
   J MANUF SYST
   
   (2018)
 * LeeJeongsu et al.
   
   
   FAULT DETECTION BASED ON ONE-CLASS DEEP LEARNING FOR MANUFACTURING
   APPLICATIONS LIMITED TO AN IMBALANCED DATABASE
   
   
   J MANUF SYST
   
   (2020)
 * FineganDonal P. et al.
   
   
   THE APPLICATION OF DATA-DRIVEN METHODS AND PHYSICS-BASED LEARNING FOR
   IMPROVING BATTERY SAFETY
   
   
   JOULE
   
   (2021)
 * LeukelJoerg et al.
   
   
   ADOPTION OF MACHINE LEARNING TECHNOLOGY FOR FAILURE PREDICTION IN INDUSTRIAL
   MAINTENANCE: A SYSTEMATIC REVIEW
   
   
   J MANUF SYST
   
   (2021)
 * LiWeihua et al.
   
   
   A PERSPECTIVE SURVEY ON DEEP TRANSFER LEARNING FOR FAULT DIAGNOSIS IN
   INDUSTRIAL SCENARIOS: THEORIES, APPLICATIONS AND CHALLENGES
   
   
   MECH SYST SIGNAL PROCESS
   
   (2022)
 * McFaddenP.D. et al.
   
   
   VIBRATION MONITORING OF ROLLING ELEMENT BEARINGS BY THE HIGH-FREQUENCY
   RESONANCE TECHNIQUE - A REVIEW
   
   
   TRIBOL INT
   
   (1984)

SawalhiN. et al.


THE ENHANCEMENT OF FAULT DETECTION AND DIAGNOSIS IN ROLLING ELEMENT BEARINGS
USING MINIMUM ENTROPY DECONVOLUTION COMBINED WITH SPECTRAL KURTOSIS


MECH SYST SIGNAL PROCESS

(2007)
SinghSarabjeet et al.


ANALYSES OF CONTACT FORCES AND VIBRATION RESPONSE FOR A DEFECTIVE ROLLING
ELEMENT BEARING USING AN EXPLICIT DYNAMICS FINITE ELEMENT MODEL


J SOUND VIB

(2014)
SawalhiN. et al.


SIMULATING GEAR AND BEARING INTERACTIONS IN THE PRESENCE OF FAULTS. PART I. THE
COMBINED GEAR BEARING DYNAMIC MODEL AND THE SIMULATION OF LOCALISED BEARING
FAULTS


MECH SYST SIGNAL PROCESS

(2008)
ZhiyiHe et al.


TRANSFER FAULT DIAGNOSIS OF BEARING INSTALLED IN DIFFERENT MACHINES USING
ENHANCED DEEP AUTO-ENCODER


MEASUREMENT: J INT MEAS CONFED

(2020)
YuJianbo


BEARING PERFORMANCE DEGRADATION ASSESSMENT USING LOCALITY PRESERVING PROJECTIONS
AND GAUSSIAN MIXTURE MODELS


MECH SYST SIGNAL PROCESS

(2011)
LiHai et al.


FAULT DIAGNOSIS OF ROLLING BEARING USING SYMMETRIZED DOT PATTERN AND
DENSITY-BASED CLUSTERING


MEASUREMENT: J INT MEAS CONFED

(2020)
SaariJuhamatti et al.


DETECTION AND IDENTIFICATION OF WINDMILL BEARING FAULTS USING A ONE-CLASS
SUPPORT VECTOR MACHINE (SVM)


MEASUREMENT: J INT MEAS CONFED

(2019)
ZhangYuyan et al.


IMBALANCED DATA FAULT DIAGNOSIS OF ROTATING MACHINERY USING SYNTHETIC
OVERSAMPLING AND FEATURE LEARNING


J MANUF SYST

(2018)
JamadarI.M. et al.


A NOVEL APPROACH INTEGRATING DIMENSIONAL ANALYSIS AND NEURAL NETWORKS FOR THE
DETECTION OF LOCALIZED FAULTS IN ROLLER BEARINGS


MEASUREMENT: J INT MEAS CONFED

(2016)
MishraC. et al.


BALL BEARING DEFECT MODELS: A STUDY OF SIMULATED AND EXPERIMENTAL FAULT
SIGNATURES


J SOUND VIB

(2017)
LiXin et al.


A FUSION CWSMM-BASED FRAMEWORK FOR ROTATING MACHINERY FAULT DIAGNOSIS UNDER
STRONG INTERFERENCE AND IMBALANCED CASE


IEEE TRANS IND INF

(2021)
ZioEnrico


PROGNOSTICS AND HEALTH MANAGEMENT OF INDUSTRIAL EQUIPMENT


DIAGNOSTICS PROGNOSTICS ENG SYST: METHODS TECH

(2013)
View more references


CITED BY (0)



View full text
© 2023 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All
rights reserved.


RECOMMENDED ARTICLES


 * CUTTING TOOL WEAR STATE RECOGNITION BASED ON A CHANNEL-SPACE ATTENTION
   MECHANISM
   
   Journal of Manufacturing Systems, Volume 69, 2023, pp. 135-149
   Rongyi Li, …, Kailin Hou
   


 * CNN-BASED IN SITU TOOL WEAR DETECTION: A STUDY ON MODEL TRAINING AND DATA
   AUGMENTATION IN TURNING INSERTS
   
   Journal of Manufacturing Systems, Volume 68, 2023, pp. 85-98
   Alberto García-Pérez, …, Matthias Weigold
   


 * A NOVEL KNOWLEDGE TRANSFER NETWORK WITH FLUCTUATING OPERATIONAL CONDITION
   ADAPTATION FOR BEARING FAULT PATTERN RECOGNITION
   
   Measurement, Volume 158, 2020, Article 107739
   Peng Chen, …, Ming J. Zuo
   


 * LIFE CYCLE ASSESSMENT OF ALUMINUM ALLOYS CHIPS RECYCLING THROUGH SINGLE AND
   MULTI-STEP FRICTION STIR CONSOLIDATION PROCESSES
   
   Journal of Manufacturing Systems, Volume 68, 2023, pp. 651-659
   Giuseppe Ingarao, …, Livan Fratini
   


 * MULTILAYER GRAD-CAM: AN EFFECTIVE TOOL TOWARDS EXPLAINABLE DEEP NEURAL
   NETWORKS FOR INTELLIGENT FAULT DIAGNOSIS
   
   Journal of Manufacturing Systems, Volume 69, 2023, pp. 20-30
   Sinan Li, …, Xuefeng Chen
   


 * ENERGY-EFFICIENT AND QUALITY-AWARE PART PLACEMENT IN ROBOTIC ADDITIVE
   MANUFACTURING
   
   Journal of Manufacturing Systems, Volume 68, 2023, pp. 644-650
   Suyog Ghungrad, …, Azadeh Haghighi
   

Show 3 more articles


ARTICLE METRICS

Captures
 * Readers: 1


View details
 * About ScienceDirect
 * Remote access
 * Shopping cart
 * Advertise
 * Contact and support
 * Terms and conditions
 * Privacy policy

We use cookies to help provide and enhance our service and tailor content and
ads. By continuing you agree to the use of cookies.

All content on this site: Copyright © 2024 Elsevier B.V., its licensors, and
contributors. All rights are reserved, including those for text and data mining,
AI training, and similar technologies. For all open access content, the Creative
Commons licensing terms apply.