processes

Article
Conditional Generative Adversarial Networks with Optimized
Machine Learning for Fault Detection of Triplex Pump in
Industrial Digital Twin
Amged Sayed 1,2, *, Samah Alshathri 3 and Ezz El-Din Hemdan 4,5

1 Department of Electrical Energy Engineering, College of Engineering & Technology, Arab Academy for
Science Technology & Maritime Transport, Smart Village Campus, Giza 12577, Egypt
2 Industrial Electronics and Control Engineering Department, Faculty of Electronic Engineering, Menoufia
University, Menoufia 32952, Egypt
3 Department of Information Technology, College of Computer and Information Sciences, Princess Nourah bint
Abdulrahman University, P.O. Box 84428, Riyadh 11671, Saudi Arabia; [email protected]
4 Structure and Materials Research Lab, Prince Sultan University, P.O. Box 66833, Riyadh 11586, Saudi Arabia;
[email protected]
5 Department of Computer Science and Engineering, Faculty of Electronic Engineering, Menoufia University,
Menoufia 32952, Egypt
* Correspondence: [email protected]

Abstract: In recent years, digital twin (DT) technology has garnered significant interest from both
academia and industry. However, the development of effective fault detection and diagnosis models
remains challenging due to the lack of comprehensive datasets. To address this issue, we propose
the use of Generative Adversarial Networks (GANs) to generate synthetic data that replicate real-
world data, capturing essential features indicative of health-related information without directly
referencing actual industrial DT systems. This paper introduces an intelligent fault detection and
diagnosis framework for industrial triplex pumps, enhancing fault recognition capabilities and
offering a robust solution for real-time industrial applications within the DT paradigm. The proposed
Citation: Sayed, A.; Alshathri, S.; framework leverages Conditional GANs (CGANs) alongside the Harris Hawk Optimization (HHO)
Hemdan, E.E.-D. Conditional as a metaheuristic method to optimize feature selection from input data to enhance the performance
Generative Adversarial Networks
of machine learning (ML) models such as Bagged Ensemble (BE), AdaBoost (AD), Support Vector
with Optimized Machine Learning for
Machine (SVM), K-Nearest Neighbors (KNNs), Decision Tree (DT), and Naive Bayes (NB). The
Fault Detection of Triplex Pump in
efficacy of the approach is evaluated using key performance metrics such as accuracy, precision,
Industrial Digital Twin. Processes 2024,
12, 2357. https://doi.org/10.3390/
recall, and F-measure on a triplex pump dataset. Experimental results indicate that hybrid-optimized
pr12112357 ML algorithms (denoted by “ML-HHO”) generally outperform or match their classical counterparts
across these metrics. BE-HHO achieves the highest accuracy at 95.24%, while other optimized models
Academic Editors: Yuhe Wang,
also demonstrate marginal improvements, highlighting the framework’s effectiveness for real-time
Shaoke Wan, Naipeng Li and
fault detection in DT systems, where SVM-HHO attains 94.86% accuracy, marginally higher than
Zijian Qiao
SVM’s 94.48%. KNN-HHO outperforms KNNs with 94.73% accuracy compared to 93.14%. Both DT-
Received: 15 September 2024 HHO and DT achieve 94.73% accuracy, with DT-HHO exhibiting slightly better precision and recall.
Revised: 10 October 2024 NB-HHO and NB show near-equivalent performance, with NB-HHO at 94.73% accuracy versus NB’s
Accepted: 25 October 2024 94.6%. Overall, the optimized algorithms demonstrate consistent, albeit marginal, improvements
Published: 27 October 2024
over their classical versions.

Keywords: machine learning; fault diagnosis; digital twins; conditional GANs (CGANs); Harris
Copyright: © 2024 by the authors. Hawk Optimizer (HHO); industrial control systems; Internet of Things (IoT)
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons 1. Introduction
Attribution (CC BY) license (https:// Recently, predictive analytics has been considered a key field within data science
creativecommons.org/licenses/by/
through the use of statistical models and machine learning algorithms to forecast future oc-
4.0/).

Processes 2024, 12, 2357. https://doi.org/10.3390/pr12112357 https://www.mdpi.com/journal/processes

 Recently, predictive analytics has been considered a key field within data science
through the use of statistical models and machine learning algorithms to forecast future
occurrences or behaviors in many applications such as fault diagnosis, customer segmen-
tation, demand estimation, risk management, and healthcare improvement, among oth-
Processes 2024, 12, 2357 ers. 2 of 20
The advancements in recent technologies such as the Internet of Things (IoT), cloud
computing (CC), machine learning (ML), and Cyber–Physical Systems (CPSs), coupled
with developments
currences or behaviorsinintelecommunication,
many applications such haveasrevolutionized
fault diagnosis,information transmission.
customer segmentation,
This revolution, attributed to digitalization, has permeated
demand estimation, risk management, and healthcare improvement, among others. all aspects of life and given
riseThe
to the concept of digital twins (DTs) in the context of Industry
advancements in recent technologies such as the Internet of Things (IoT), cloud 4.0. The DT represents
a virtual replica
computing of a physical
(CC), machine learning product
(ML), within the framework
and Cyber–Physical of Cyber–Physical
Systems (CPSs), coupled Systems,
with
mimicking the
developments inbehavior of the real system
telecommunication, throughout itsinformation
have revolutionized lifecycle. Bytransmission.
integrating digital
This
and physical
revolution, twins, efficient
attributed management,
to digitalization, control, andalldecision-making
has permeated aspects of life and processes
given riseare to
en-
abled during the operation of the real system. The DT captures
the concept of digital twins (DTs) in the context of Industry 4.0. The DT represents a virtualdata from physical sensors
to monitor
replica the system’s
of a physical product response
within the andframework
predicts and diagnoses its behavior
of Cyber–Physical Systems,to anticipate
mimicking
faults, enabling proactive maintenance actions. However, the
the behavior of the real system throughout its lifecycle. By integrating digital and physicalperformance and reliability
of predictive
twins, efficientmaintenance
management, models
control, mayand bedecision-making
affected greatly by the unavailability
processes are enabledofduring enough
data
the or by the
operation of existence of imbalanced
the real system. datasets. data
The DT captures This from
lack of data may
physical ultimately
sensors cause
to monitor
performance
the system’s responsebias, misclassification,
and predicts and diagnosesor poor analysis, thustocausing
its behavior anticipate thefaults,
breakdown
enablingof
maintenance
proactive systems. actions. However, the performance and reliability of predictive
maintenance
Across various
maintenance models may industries,
be affected including
greatlytheby manufacturing,
the unavailability healthcare,
of enoughand dataautomotive
or by the
fields, the
existence of DT has become
imbalanced an invaluable
datasets. This lackasset
of data[1–3].
mayDT technology
ultimately supports
cause data fusion,
performance bias,
modeling, and technology
misclassification, or poor analysis,integration by simulating
thus causing physical of
the breakdown systems, enabling
maintenance effective
systems.
problem-solving
Across various forindustries,
complex interdisciplinary challenges. Central
including the manufacturing, to the and
healthcare, DT concept
automotive is the
exchange
fields, the DT of has
databecome
streamsanfrom learning
invaluable components
asset [1–3]. DT and remote supports
technology sensors indata simulation,
fusion,
which areand
modeling, crucial for developing
technology integration intricate processes
by simulating and exploring
physical systems,“what-if” scenarios.
enabling effective
Notably, the field of environmental sciences has also recognized the importance of the DT,
problem-solving for complex interdisciplinary challenges. Central to the DT concept is
encompassing areas such as hydrology, agriculture, smart farming, animal farming, re-
the exchange of data streams from learning components and remote sensors in simulation,
mote are
which sensing,
crucial and forearth sciences intricate
developing [4–11]. processes and exploring “what-if” scenarios.
Notably,Digital twins
the field of are increasinglysciences
environmental being integrated with ML algorithms
has also recognized the importancefor fault diagno-
of the DT,
sis schemes based
encompassing areas on suchmachine learning
as hydrology, and digitalsmart
agriculture, twinsfarming,
for fault-tolerant systems.
animal farming, How-
remote
ever, there
sensing, and areearth a lack of available
sciences [4–11]. industrial data, so there is a serious need to provide a
promising approach for solving the
Digital twins are increasingly scarcity
being of fault
integrated dataML
with byalgorithms
generating for synthetic data with
fault diagnosis
better adaptability like Conditional Generative Adversarial Networks (CGANs).
schemes based on machine learning and digital twins for fault-tolerant systems. CGANs
However,
there are a lack ofan
are considered available
advanced industrial data, somodel
deep learning there is a serious
capable need to provide
of generating a promising
realistic data sam-
approach
ples based foron solving
provided the criteria.
scarcity CGANsof fault generate
data by generating synthetic
highly convincing data with
outputs better
by combin-
adaptability
ing generative models and adversarial training. Figure 1 shows the general steps ofare
like Conditional Generative Adversarial Networks (CGANs). CGANs the
considered an advanced
Fault Diagnosis Model deep learning control
for industrial model capable
systems.of generating realistic data samples
based on provided criteria. CGANs generate highly convincing outputs by combining
generative models and adversarial training. Figure 1 shows the general steps of the Fault
Diagnosis Model for industrial control systems.

Figure 1. General steps of industrial Fault Diagnosis Model.

Figure 1. General steps of industrial Fault Diagnosis Model.
Nevertheless, the development of effective fault detection and diagnosis models re-
mains challenging due to the lack of comprehensive datasets. To address this issue, we
propose the use of Generative Adversarial Networks (GANs) to generate synthetic data
that replicate real-world data, capturing essential features indicative of health-related infor-
mation without directly referencing actual industrial DT systems. This paper introduces
an intelligent fault detection and diagnosis framework for industrial triplex pumps, en-
hancing fault recognition capabilities and offering a robust solution for real-time industrial
applications within the DT paradigm.
Processes 2024, 12, 2357 3 of 20

Therefore, this paper introduces an efficient fault diagnosis framework for industrial
digital twin systems. The framework aims to achieve accurate and efficient fault detection
and diagnosis by incorporating Conditional Generative Adversarial Networks (CGANs) for
generating synthesis industrial pump data and hybrid-optimized machine learning meth-
ods with Harris Hawk Optimization (HHO). The novelty of the work is in the combination
of CGAN with the HHO for the feature selection process in machine learning models. This
approach renders a significant improvement in the performance of models used in fault
detection systems. This resolves a very typical challenge in industrial fault detection where
such datasets are often unbalanced, enhancing the generalization and performance of the
model. Therefore, the key contributions of this paper can be summarized as follows:
• Develop a smart and vigorous fault detection and diagnosis framework for industrial
triplex pumps, enhancing fault recognition capabilities and offering a robust solution
for real-time industrial applications within the DT paradigm.
• The proposed framework leverages Conditional GANs (CGANs) alongside the Harris
Hawk Optimization (HHO) metaheuristic method to optimize feature selection from
input data effectively for machine learning (ML) models such as Bagged Ensemble
(BE), AdaBoost (AD), Support Vector Machine (SVM), K-Nearest Neighbors (KNNs),
Decision Tree (DT), and Naive Bayes (NB). The efficacy of the approach is evaluated
using key performance metrics such as accuracy, precision, recall, and F-measure on a
triplex pump dataset.
• From the experimental results, the suggested hybrid-optimized ML algorithms outper-
form or match their classical counterparts across various metrics. BE-HHO achieves
the highest accuracy at 95.24%, slightly surpassing BE’s 95.17%. SVM-HHO attains
94.86% accuracy, marginally higher than SVM’s 94.48%. KNN-HHO outperforms
KNNs with an accuracy of 94.73% compared to 93.14%. Both DT-HHO and DT achieve
94.73% accuracy, with DT-HHO displaying slightly better precision and recall. NB-
HHO and NB show nearly equivalent performance, with NB-HHO at 94.73% accuracy
versus NB’s 94.6%. Although AD-HHO and AD have lower accuracies at 92.57% and
92.06%, respectively, AD achieves higher recall.
• Hybrid-optimized machine learning models using the HHO will outperform classical
models in terms of accuracy, precision, and recall in diagnosing faults in industrial
pump systems.
Thus, this research proposes an innovative approach that successfully exploits both the
features optimization and the synthetic data production to fit the needs of typical dynamic
industrial environments based on the digital twin.
The structure of the paper is as follows: In Section 2, relevant work on the paper’s
theme is explored, while a brief overview of Conditional GAN (CGAN) and Harris Hawk
Optimization (HHO) as they pertain to the proposed system is provided in Section 3. The
suggested framework is explained and clarified in Section 4, while a high-level proposed
framework for remote fault monitoring and detection in smart industrial IoT systems
is delivered in Section 5. Section 6 shows the experimental findings and comparative
effectiveness of the suggested approaches in comparison with classical ML models, while
the paper’s conclusion and future scope of this innovative topic are explored in Section 8.

2. Previous Studies
To identify and diagnose defective equipment, this study intends to develop and
use a digital twin system for the triplex pump. For several industrial processes, fault
detection and diagnosis have been carried out to boost effectiveness, safety, and continuous
production. In recent decades, numerous Artificial Intelligence techniques for failure
diagnosis have been introduced to increase the reliability and security of sophisticated
equipment. One of the effective machine learning techniques used is Generative Adversarial
Networks (GANs). GANs have shown promising results in various fields, including fault
diagnosis [12–15]. These studies show how GANs can be used to diagnose faults in a range
of sectors, including manufacturing, energy, and transportation. However, there are still
Processes 2024, 12, 2357 4 of 20

several issues and open research paths in this field, including how to deal with imbalanced
data, how to deal with the absence of labeled data, and how to use other GAN techniques
in fault diagnosis.
One of the well-known and effective methods for defect diagnosis is the model-based
approach, in which an accurate complicated apparatus model is built by different analytic
terms [16–18]. Physically and mathematically informed approaches have been effectively
employed to tackle the intricacies of sophisticated industrial machinery, leveraging es-
tablished model-based methodologies. Nevertheless, the profound complexity of certain
equipment often necessitates a thorough comprehension of the underlying physical princi-
ples to create an accurate model. Due to its potential to revolutionize several industries,
including gaming, education, healthcare, and manufacturing, the ideas of industrial digital
twins have attracted significance recently.
On the other hand, digital twins are virtual representations of actual things, systems,
or settings that can be used for testing, simulation, and monitoring. Numerous industries,
including engineering, architecture, urban planning, and healthcare, use them. The pub-
lication by Grieves et al. [19], which presents the idea of digital twins and explores their
potential advantages and disadvantages, is one of the foundational works on this topic. The
authors contend that digital twins can enhance consumer experience, lower expenses, and
improve product development. Although the present DT schemes and executions are still
in their initial phases and require significant effort, they have been successfully integrated
into various applications such as healthcare systems, various industries including aviation
and farming, smart cities, and climate prediction [20,21].
Designing a competent digital twin system for any physical system requires the exper-
tise of specialized engineers and computer scientists. Their duties comprise constructing
and proposing the necessary product model and creating a comprehensive description of
the virtual system. The authors demonstrate the effectiveness of their approach in reducing
costs and improving transparency. Also, Wang et al. [22] create a digital twin platform for
smart cities that combines data from several sources to offer in-the-moment monitoring
and optimization of municipal infrastructure. However, they also note the difficulties in
managing data, scaling, and cybersecurity that come with building and sustaining digital
twins. A framework for building digital twins of the triplex pump and using hybrid ma-
chine learning for fault diagnosis in the industrial system is given in [23]. Fault diagnosis
becomes faster, more accurate, and more cost-effective, leading to improved operational
efficiency and reduced downtime in various industries, such as manufacturing, energy, and
transportation [24,25].
The current state of the art in industrial IoT applications is characterized by a scarcity of
contributions focused on the integration of digital twins and machine learning algorithms.
In response to this gap, this study proposes a novel fault prediction framework comprising
four phases, namely (1). a Data Acquisition Step (DAS), (2). a Data Synthesizing Step (DSS),
(3). ML-based Model Training and Testing (MLMT2), and finally (4). a Failure Diagnosis
Step (FDS). The proposed framework seeks to develop an advanced powerful forecast
digital twin-assisted AI framework that leverages Generative Adversarial Networks (GANs)
in combination with diverse HHO-based optimized machine learning techniques, such
as Bagged Ensemble (BE), AdaBoost (AD), Support Vector Machine (SVM), K-Nearest
Neighbors (KNNs), Decision Tree (DT), and Naive Bayes (NB) to identify and classify
faults effectively.
In conclusion, this section presents a critical review of the literature on fault detection
and diagnosis (Table 1) using digital twins and Artificial Intelligence techniques, specifically
Generative Adversarial Networks (GANs) and machine learning algorithms. Various
studies are summarized, highlighting their objectives, methodologies, key findings, and
their pros and cons.
Processes 2024, 12, 2357 5 of 20

Table 1. A critical review of the literature on fault detection and diagnosis.

Reference Objective Methods Findings Pros Cons

[12–15] Use GANs for fault Generative GANs can enhance High accuracy in Issues with
diagnosis across Adversarial fault detection but fault detection imbalanced data,
various sectors (e.g., Networks (GANs) challenges remain with absence of labeled
manufacturing, energy, imbalanced and data
transportation) unlabeled data
[16–18] Model-based approach Physically and Effective for In-depth Requires deep
for defect diagnosis mathematically sophisticated understanding of understanding of
using physical and informed machinery, but building machinery physical principles
mathematical models model-based accurate models is mechanics
methodologies complex
[19] Exploring the concept Digital twin Digital twins improve Enhanced Initial
of digital twins and framework design user experience, reduce transparency and implementation
their applications and costs, and enhance reduced costs phases, requires
across various implementation product development further
industries development
[22] Develop a digital twin Digital twin Improved real-time Real-time Challenges with
platform for smart platform infrastructure monitoring and data management,
cities, focusing on combining data monitoring but faces optimization of scaling, and
real-time monitoring from multiple challenges with scaling smart city cybersecurity
and optimization sources and cybersecurity infrastructure
[23–25] Hybrid machine Hybrid ML-based Faster and more Cost-effective and Complexity in
learning for fault model with GANs accurate fault diagnosis, precise fault integrating hybrid
diagnosis in the and HHO reducing downtime diagnosis machine learning
industrial system, algorithms (BE, and improving models
focusing on triplex AD, SVM, KNNs, operational efficiency
pumps DT, NB)

3. Background
This section discusses the topics of Conditional GAN (CGAN) and Harris Hawk
Optimization (HHO) as they pertain to the proposed system.

3.1. Conditional GAN (CGAN)

A new kind of deep learning network model called a Generative Adversarial Network
directly generates similar distributions from real data. Due to GAN’s powerful data-
generating capabilities, data imbalance is a common problem that it is utilized to solve. The
Generative Adversarial Network (GAN), which is used as a machine learning framework
for training generative models, is extended to create the conditional generative adversarial
network (CGAN). Therefore, to simulate real data input to the networks, synthetic data can
be produced using conditional Generative Adversarial Networks (CGANs).
The CGAN uses a conditional setting, which contains two networks: generator and
discriminator, as shown in Figure 2. New data generated by the generator network have
the same structure as the real data and correspond to the same label. By the discriminator
network, on the other hand, observations are categorized as “real” or “generated.” When
presented with batches containing both actual and created labeled data, the discriminator’s
goal is to avoid being “fooled” by the generator. The discriminator and generator both rely
on additional data, such as class labels or details from different modalities [26].
 the same structure as the real data and correspond to the same label. By the discriminator
network, on the other hand, observations are categorized as “real” or “generated.” When
presented with batches containing both actual and created labeled data, the discrimina-
tor’s goal is to avoid being “fooled” by the generator. The discriminator and generator
Processes 2024, 12, 2357 both rely on additional data, such as class labels or details from different modalities [26]. 6 of 20

Figure 2. The system architecture of Conditional GAN method for synthetic data.
Figure 2. The system architecture of Conditional GAN method for synthetic data.

Consequently, the detailed explanation of how a CGAN operates is summarized as

Consequently, the detailed explanation of how a CGAN operates is summarized as
follows: First, it is necessary to have a sizable collection of genuine data samples that fit the
follows: First, it is necessary to have a sizable collection of genuine data samples that fit
desired criteria. This dataset will act as our CGAN’s “teacher”. The next step is to construct
the desired criteria. This dataset will act as our CGAN’s “teacher”. The next step is to
a generator
construct network
a generator that outputs
network that outputs a synthetic
a syntheticdata sample
data sample from
froman aninput
input random
random noise
vector. The instructor dataset’s real data samples are used to
noise vector. The instructor dataset’s real data samples are used to train the generator train the generator network
to reduce the difference between its output and those samples.
network to reduce the difference between its output and those samples. Then, a discrimi- Then, a discriminator
network
nator networkis constructed
is constructedto produce
to produce a probability
a probability score
scorethat
thatindicates
indicates whether
whethereach eachsample
sample of real and synthetic data is real or fraudulent. The discriminator networkskilled
of real and synthetic data is real or fraudulent. The discriminator network is is at
skilled at telling the difference between true samples and false ones. Two loss functions—
telling the difference between true samples and false ones. Two loss functions—one for the
one for the generator
generator and one for andthe
one for the discriminator—are
discriminator—are specifiedspecified during training.
during training. The
The discriminator
discriminator
loss function loss functionthe
promotes promotes the discriminator
discriminator to accurately to accurately
distinguish distinguish
betweenbetween real and false
real and false
samples, samples,
whereas thewhereas
generator theloss
generator
function loss function encourages
encourages the generator the generator to
to create synthetic
create
samplessynthetic
that samples that are comparable
are comparable to the real data to thesamples.
real dataThesamples. The generator
generator and
and discriminator
discriminator
networks are networks
optimizedare optimized
with each with eachcycle.
training training cycle.
Fixing theFixing the discriminator
discriminator network will
network
allow the willmodel
allow the model the
to assess to assess the generator
generator network’s network’s
loss. The loss. The generator
generator network net-is then
work is thentoupdated
updated reducetothis
reduce
loss.this loss.the
After After the generator
generator networknetwork has been
has been fixed, fixed, it is
it is used to
used to calculate
calculate the discriminator
the discriminator network’s
network’s loss.loss.
TheThe discriminator
discriminator network
network is isthen
thenupdated
up- to
dated to maximize this loss. Subsequently, we carry out the generator
maximize this loss. Subsequently, we carry out the generator and discriminator networks’ and discriminator
networks’ optimizations
optimizations once more onceuntil
more until
they theyareach
reach stablea stable conclusion.
conclusion.

3.2. Harris Hawk Optimization (HHO)

A metaheuristic algorithm called the Harris Hawks Optimization (HHO) algorithm
is a bio-inspired optimization technique that mimics the behavior of hawks to optimize a
problem [27]. Its purpose was to solve issues with non-linear objectives and many local
optima. Utilizing a set of prey items that serve as potential solutions to the problem, the
algorithm searches iteratively for better alternatives. Each prey item has a fitness rating
assigned to it that describes how it stacks up against other prey items. The algorithm
searches for fresh prey items and gradually increases their fitness values by combining
exploration and exploitation tactics. The HHO has been successfully applied in various
domains such as power systems, control engineering, biomedical applications, and com-
munication systems. In the context of feature selection, the HHO can be used to identify
the most relevant features in a dataset. The following is a high-level overview of how the
HHO works for feature selection:
1. Initialize the population: Start by randomly selecting a subset of features from the
original dataset. This initial population represents the first generation of hawks.
2. Evaluate the fitness: Assess the fitness of each individual in the population based
on its ability to predict the target variable. In this case, the fitness function would
evaluate the accuracy of the model built using the selected features.
Processes 2024, 12, 2357 7 of 20

3. Mating Pool: Select the fittest individuals from the current population to form the
mating pool. The size of the mating pool determines the number of offspring produced
in the next generation.
4. Crossover and Mutation: Apply crossover and mutation operators to the members
of the mating pool to generate new offspring. Crossover involves combining two
parent individuals to produce a single offspring, while mutation involves introducing
random changes to an individual.
5. Replacement: Replace the least fit individuals in the current population with the
newly generated offspring. This maintains the diversity of the population and pre-
vents the algorithm from getting stuck in the local optimum.
6. Repeat: Go back to step 2 and repeat the process until a stopping criterion is met,
such as reaching a maximum number of generations or achieving a desired level
of accuracy.
7. Selection of final features: Once the algorithm converges, select the top-ranked
features from the final population as the optimal set of features for the given dataset.
The key advantage of the HHO is its ability to handle complex, non-linear problems
and its robustness against noise and outliers in the data. Additionally, the HHO can be
easily parallelized, making it suitable for large datasets. However, the algorithm requires
careful parameter tuning for optimal performance.

4. Proposed Framework
In this work, we present a powerful scheme for automatically recognizing faults in
industrial DT systems. The proposed framework is demonstrated, which consists of a
hybrid-optimized ML model through the HHO method with different machine learning
methods such as Bagged Ensemble (BE), AdaBoost (AD), Support Vector Machine (SVM),
K-Nearest Neighbors (KNNs), Decision Tree (DT), and Naive Bayes (NB) on the CGANS-
based generated dataset. Furthermore, the detailed suggested framework comprises five
significant steps, as follows, to achieve the diagnostic procedure explained in Figure 3:
Stage 1: (Gaining and Gathering of Data): in this phase, the data will be gathered from
digital twin sources to handle the subsequent steps in the proposed system.
Stage 2: (Generating Synthetic Data): In this stage, the CGAN is utilized to create synthetic
data that resemble actual networks’ input. By utilizing labeled data, Conditional GANs
(CGANs) can generate synthetic samples that belong to specific, predefined categories.
Stage 3: (Generated Data Validation): In this phase, the principal component analysis
(PCA) is applied to assess the properties of both the created and actual signals. PCA enables
a complex dataset to be transformed into a set of uncorrelated variables, which are referred
to as the principal components. The goal of PCA is to use the numerical structures of the
actual data and assign the features of the created data to the same PCA subspace.
Stage 4: (Training/Testing the proposed model): At this point, the data generated in phase
one are used to train an effective ML model that diagnoses the fault of the optimized ML
methods to organize all the input data generated from the digital twin model. Then, the
HHO technique is applied to enhance the performance of the machine learning algorithm.
Finally, the data collected in real-time digital twins of machinery data are used to test
different machine learning algorithms. Therefore, the performing assessment indices are
applied to assess the suggested framework.
Stage 5: (Cloud-based Monitoring System for Fault Classification and Prediction): To
monitor machine data from their industrial systems, the supervisor operators in this step
use a cloud-based industrial monitoring tool of the proposed framework. To optimize
operations, operators can use this system to identify patterns and trends indicative of
impending failures, enabling proactive remediation before issues arise.
Processes 2024, 12, 2357 8 of 20
Processes 2024, 12, x FOR PEER REVIEW 9 of 22

Figure 3. Proposed
Figure hybrid-optimized
3. Proposed fault detection
hybrid-optimized framework.
fault detection framework.

5. High-Level Proposed Industrial IoT-Based DT Framework

The proposed framework aims to revolutionize the field of machinery fault diagnosis
by leveraging cutting-edge technologies such as digital twins, Cyber—Physical Systems,
cloud computing, and Artificial Intelligence. By harnessing these innovations, the sug-
gested framework offers real-time fault diagnosis and monitoring of industrial systems,
thereby significantly improving their overall efficiency. The proposed framework comprises
three distinct phases that collaboratively work towards achieving the desired outcomes.
Each phase has a defined set of tasks and operations that coordinate with the others to
deliver a seamless and effective solution. Figure 4 explains the suggested system with the
three phases as follows:
• Phase 1: industrial IoT can be used for real-time digital twin generated data gathering,
as shown in Figure 5.
• Phase 2: For storing and processing user data, the cloud infrastructure will serve
as a centralized repository and data will be transmitted remotely via the Internet.
Once received, the data will be sorted and organized, making it readily accessible for
comprehensive analysis and thorough assessment.
 prises three distinct phases that collaboratively work towards achieving the desired out-
comes. Each phase has a defined set of tasks and operations that coordinate with the oth-
ers to deliver a seamless and effective solution. Figure 4 explains the suggested system
with the three phases as follows:
 Phase 1: industrial IoT can be used for real-time digital twin generated data gather-
Processes 2024, 12, 2357 ing, as shown in Figure 5. 9 of 20
 Phase 2: For storing and processing user data, the cloud infrastructure will serve as
a centralized repository and data will be transmitted remotely via the Internet. Once
• received,
Phase 3: Thetheindustrial
data will be sorted and organized,
supervisor employs amaking it readilymonitoring
cloud-based accessible forsystem
com- to track
prehensive analysis and thorough assessment.
vital signs of their machinery in real-time. This intuitive dashboard provides super-
 Phase 3: The industrial supervisor employs a cloud-based monitoring system to track
visors
vitalwith
signssimple
of their access
machineryto critical data,This
in real-time. allowing
intuitivethem to examine
dashboard providesdetailed
super- reports
generated
visors with simple access to critical data, allowing them to examine detailed reportsthis infor-
by the system’s advanced analytics capabilities. Armed with
generated
mation, by the system’s
the supervisor canadvanced analytics decisions
make informed capabilities.toArmed with their
optimize this infor-
equipment’s
mation, the and
performance supervisor
maintaincan make
maximuminformed decisions to optimize their equipment’s
uptime.
performance and maintain maximum uptime.

Generated data Digital Twins Fault diagnosis

•Raw data •Digital Twins
• Real-time data gathering •Cloud based Metahueristic-AI
•CGAN model
Analytics Scheme

Processes 2024, 12, x FOR PEER REVIEW 11 of 22

Figure 4. Key phases of fault diagnosis in industrial digital twins.

Figure 4. Key phases of fault diagnosis in industrial digital twins.

Figure 5. A typical high-level

Figure 5. A suggested real-time
typical high-level IoT-based
suggested fault
real-time diagnosis
IoT-based faultframework.
diagnosis framework.

6. Experimental Study
The generated pump dataset and the experimental setup are provided in this section.
In conclusion, this section provides an analysis of the findings and a discussion of the
suggested framework.
Processes 2024, 12, 2357 10 of 20

6. Experimental Study
The generated pump dataset and the experimental setup are provided in this section.
In conclusion, this section provides an analysis of the findings and a discussion of the
suggested framework.

6.1. Experiment Setup

The machine learning classifiers were developed in MATLAB on an AMD Ryzen-5000
series (7) CPU with 8 GB RAM running Windows 11; tests were conducted to assess the
suggested framework for problem diagnosis based on a triplex pump dataset.

6.2. Triplex Pump Dataset

To develop an automated fault diagnosis algorithm for the triplex pump, a simulated
model of the pump was introduced [28]. This allowed for the generation of 1575 pump
output flow measurements, with 760 healthy signals and 815 faulty signals. The statistical
characteristics of these real signals were analyzed using principal component analysis
(PCA) to compare them to the created signals from the CGAN. Next, different machine
learning models were trained based on the generated signals from the CGAN and tested
on the real signals to determine their ability to accurately classify healthy and faulty
signals. The trained models were then used to obtain predicted labels for the actual signals.
The following steps were taken to train different machine learning models based on the
generated signals from CGAN and then predict whether a real signal was healthy or faulty:
• Create a training dataset using the generated signals.
• Create a test dataset using the real signals.
• Train the model using the training dataset.
• Obtain the predicted labels for the actual signals using the trained model under the test.

6.3. Assessment Criteria

The proposed framework blends ML with a Conditional GAN to enhance the classi-
fication accuracy of fault detection in industrial control systems. We evaluate proposed
ML different classifiers using precision, recall, and F-measure as our primary performance
metrics. On each classifier, these metrics are computed for both positive (‘P’) and negative
(‘N’) categorized documents, as depicted in the confusion matrix presented in Table 2. This
table provides a comprehensive overview of four crucial parameters—true positive (TrueP),
true negative (TrueN), false positive (FalseP), and false negative (FalseN)—that are essential
in evaluating the performance of a classification model. TP represents the accurate identifi-
cation of anomalies, whereas TN refers to the incorrect estimation of regular instances. On
the other hand, FP signifies the misclassification of regular instances as anomalies, while FN
denotes the failure to identify actual anomalies. By carefully considering these parameters,
we can gain effective perceptions of the robustness and weaknesses of each classifier and
optimize our framework for enhanced reliability and accuracy in fault detection.

Table 2. Confusion matrix for fault detection.

Predicted Heathy Predicted Faulty

Actual Healthy TrueP FalseN
Actual Faulty FalseP TrueN

Assessment metrics like accuracy, precision, recall, and F1-Score can be processed after
finding the parameters in the confusion matrix as follows [29–32]:
Accuracy: To ensure the reliability and accuracy of our system, it is imperative to determine
certain key parameters that influence the quality of the model. Specifically, we must evalu-
ate the symmetry of the datasets and the balance between false positive and false negative
rates, as expressed in Equation (1). A well-balanced dataset with minimal disparity between
Processes 2024, 12, x FOR PEER REVIEW 13 of 22

Processes 2024, 12, 2357 11 of 20

Recall: Recall, as quantified by the ratio of anticipated true positive values to submitted
expected true positive values minus predicted false negative values (Equation (3)), serves
as a vital
these metric
rates will for evaluating
enable the performance
us to achieve the highestofpossible
our proposed approach.
F1-Score, therebyAvalidating
higher recall
the
value indicates
efficacy a greater likelihood
of our proposed approach. of detecting actual anomalies within the data, thereby
underscoring the effectiveness of our methodology.
TrueP + TrueN
Accuracy = 𝑇𝑟𝑢𝑒𝑃 (1)
TrueP
𝑅𝑒𝑐𝑎𝑙𝑙 = + FalseP + FalseN + TrueN (3)
𝑇𝑟𝑢𝑒𝑃 + 𝐹𝑎𝑙𝑠𝑒𝑁
Precision: the correlation between actual positive predicted values and full positive pre-
F1-Score: It is aisgeneral
dicted values indicator
depicted of a model’s
in Equation (2). accuracy that blends recall and precision in
the strange way that both multiplication and addition combine two components to create
a whole new item. Equation (4) illustrates that the F1-Score is equal to twice the product
TrueP
Precision = (2)
of the precision and recall submission multiplied TrueP +byFalseP
two.
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 ∗ true
Recall: Recall, as quantified by the ratio of anticipated 𝑅𝑒𝑐𝑎𝑙𝑙
positive values to submitted
F1-score = 2 ∗ (4)
expected true positive values minus predicted false negative
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 + 𝑅𝑒𝑐𝑎𝑙𝑙values (Equation (3)), serves
as a vital metric for evaluating the performance of our proposed approach. A higher recall
value indicates a greater likelihood of detecting actual anomalies within the data, thereby
6.4. Results Analysis
underscoring the effectiveness of our methodology.
To thoroughly assess the effectiveness of the proposed classification framework, it is
crucial to investigate the individual performance TruePof each classifier in distinguishing be-
tween normal and problematic states Recallin=machinery pump data. The CGAN generates the (3)
TrueP + FalseN
synthetic data for the triplex pump to be utilized as the training set for our model to judge
validityItand
F1-Score:
the is a accuracy
general indicator of a model’s
of the proposed accuracy
framework. that
The blendsmodel
CGAN recallused
and precision in
in the pro-
posed model is depicted in Figure 6. Then, we test different ML models including Baggeda
the strange way that both multiplication and addition combine two components to create
whole newSupport
Ensemble, item. Equation (4) illustrates
Vector Machine (SVM),that the F1-Score
K-Nearest is equal(KNNs),
Neighbors to twiceAdaBoost,
the product
De-of
the precision and recall submission multiplied by two.
cision Tree (DT), and Naïve Bias (NB) models using the real data gained from the actual
system.
Therefore, the data generated from Precision ∗ Recall
F1-score = 2CGAN
∗ are used to train the different algorithms,
(4)
Precision
and then the real data are used to examine the performance + Recall of the proposed algorithm.
The
6.4. advantage of this technique is that there is no need for splitting the actual data and
Results Analysis
one can test the proposed algorithm on whole actual data. We can grasp its capabilities
To thoroughly assess the effectiveness of the proposed classification framework, it is
due to this approach. From Figure 7, the distribution of the generated signals is like the
crucial to investigate the individual performance of each classifier in distinguishing between
distribution of the real signals. Both faulty and healthy signals, whether generated or real,
normal and problematic states in machinery pump data. The CGAN generates the synthetic
lie in the same region of the PCA subspace, indicating that the properties of the generated
data for the triplex pump to be utilized as the training set for our model to judge the validity
signals are equivalent to those of the real signals. Table 3 shows a comparison of the results
and accuracy of the proposed framework. The CGAN model used in the proposed model
and these are also depicted in Figure 8. Figure 9 shows a confusion matrix for different
is depicted in Figure 6. Then, we test different ML models including Bagged Ensemble,
ML methods and Figure 10 displays the confusion ROC curve for all algorithms in the ML
Support Vector Machine (SVM), K-Nearest Neighbors (KNNs), AdaBoost, Decision Tree
framework.
(DT), and Naïve Bias (NB) models using the real data gained from the actual system.

Figure
Figure6.6.The
TheCGAN
CGANmodel
modelfor
forthe
thetriplex
triplexpump
pumpsystem.
system.
 Processes 2024, 12, 2357 12 of 20

Therefore, the data generated from CGAN are used to train the different algorithms,
and then the real data are used to examine the performance of the proposed algorithm. The
advantage of this technique is that there is no need for splitting the actual data and one can
test the proposed algorithm on whole actual data. We can grasp its capabilities due to this
approach. From Figure 7, the distribution of the generated signals is like the distribution
Processes 2024, 12, x FOR PEER REVIEW of the real signals. Both faulty and healthy signals, whether generated or real, 14 oflie
22 in the
same region of the PCA subspace, indicating that the properties of the generated signals are
equivalent to those of the real signals. Table 3 shows a comparison of the results and these
are also depicted in Figure 8. Figure 9 shows a confusion matrix for different ML methods
and Figure 10 displays the confusion ROC curve for all algorithms in the ML framework.

Figure Figure
7. Signal
7. feature visualization
Signal feature using the
visualization firstthe
using three
firstPCAs.
three PCAs.

For the3.mentioned
Table ComparisonML models,
results the resulting
of proposed systemconfusion matrices are indicated in Fig-
for fault detection.
ures 8 and 9 for the triplex model. Likewise, the ROC curves of the proposed system are
illustratedAlgorithm
in Figure 10. TheAccuracy
Harris Hawk Optimization
Recall (HHO) is applied to machine
Precision F1-Score
learning algorithms
BE to enhance the performance96.5789
95.1746 and boost the accuracy
93.6224of the system. The
95.0777
HHO is performedAD in a highly competitive
92.0635 manner in
96.8421terms of the caliber
87.9331 of its exploration
92.1728
and exploitation. The optimization algorithm is used for feature selection. The results in
SVM 94.6032 99.6053 90.2265 94.6842
Figure 11 show that the HHO reserves and enhances the performance of all the models.
KNNs 93.1429 99.8684 87.6443 93.3579
Table 3. Comparison
DT results of proposed
94.4762 system for 95.6579
fault detection. 93.0858 94.3543

AlgorithmNB Accuracy 94.6032 Recall Precision 91.9255 F1-Score 94.5687

97.3684
BE 95.1746 96.5789 93.6224 95.0777
AD For the mentioned
92.0635 ML models,
96.8421 the resulting
87.9331 92.1728
confusion matrices are indicated in
SVM
Figures 8 and 994.6032
for the triplex99.6053
model. Likewise,90.2265 94.6842
the ROC curves of the proposed system
are illustrated in
KNNs Figure 10. The
93.1429 Harris Hawk Optimization
99.8684 87.6443 (HHO) is93.3579
applied to machine
learning algorithms to enhance the performance and boost the accuracy of the system. The
DT 94.4762 95.6579 93.0858 94.3543
HHO is performed in a highly competitive manner in terms of the caliber of its exploration
NB 94.6032 97.3684 91.9255 94.5687
and exploitation. The optimization algorithm is used for feature selection. The results in
Figure 11 show that the HHO reserves and enhances the performance of all the models.
Processes 2024,2024,
Processes 12, x12,
FOR PEER REVIEW
2357 15 of 22
13 of 20

Accuracy Precision Recall F1-score

100

98 12, x FOR PEER REVIEW

Processes 2024, 15 of 22

96

94
Accuracy Precision Recall F1-score
Measuremnts %

92 100

90 98

88 96

86 94
Measuremnts %

84 92

90
82
88
80
86 BE AD SVM KNN DT NB
84 ML Models

82
Figure 8. 8.Results
Figure Resultsfor
forML
ML models forfault
models for faultdetection
detection without
without optimization.
optimization.
80
BE AD SVM KNN DT NB
SVM Figure 11 and Table 4 show KNNs
the performance of various
ML Models
machine learning (ML) al-
gorithms in their optimized and classical forms across four metrics: accuracy, precision,
recall, and F1-Score. The optimized versions are indicated with the suffix “-HHO” and
Figure 8. Results
consistently for ML models
outperform for fault
or closely detection
match their without
classicaloptimization.
counterparts.

SVM KNNs

Figure 9. Cont.
Processes 2024, 12, x FOR PEER REVIEW 16 of 22
Processes2024,
Processes 2024,12,
12,2357
x FOR PEER REVIEW 16 of 22
14 of 20

BEBE NBNB

DTDT ADAD

Figure 9. Confusion
Figure matrix
9. Confusion of the
matrix proposed
of the framework.
proposed framework.
Figure 9. Confusion matrix of the proposed framework.

SVM
SVM KNNs
KNNs
True Positive Rate

True Positive Rate

True Positive Rate

True Positive Rate

Figure 10. Cont.

Processes 2024,
Processes 12,12,
2024, x FOR
2357PEER REVIEW 17 of 22
15 of 20

BE NB
True Positive Rate

DT AD

Figure 10. ROC curves for the proposed system.

Figure 10. ROC curves for the proposed system.
The BE-HHO shows slightly higher performance than BE, with an accuracy of 95.24%
Figure 11
compared to and Table
95.17%, 4 show
and the performance
similarly ofrecall,
high precision, variousandmachine learning
F1-Scores. AD-HHO(ML)and
algo-
AD
rithms
have in their optimized
a notable differenceand classical forms
in accuracy, across four
with AD-HHO at metrics: accuracy,
92.57% and precision,
AD at 92.06%, re-
though
call,
ADand F1-Score.
achieves Therecall.
higher optimized versions arevariant
The SVM-HHO indicated with an
exhibits theaccuracy
suffix “-HHO” andslightly
of 94.86%, con-
sistently
higher outperform or closely
than the classical SVMmatch theirwith
at 94.48%, classical
both counterparts.
showing high recall values. KNN-HHO
surpasses KNNs in accuracy (94.73% vs. 93.14%) and demonstrates improved precision
and recall. DT-HHO and DT both have an accuracy of 94.73%, but DT-HHO has marginally
better precision and recall. NB-HHO and NB are almost equivalent, with NB-HHO having
a slightly higher accuracy of 94.73% compared to 94.6%. Overall, the optimized algorithms
demonstrate marginal but consistent improvements over their classical versions in most
metrics, as clarified in Figure 12.
Processes 2024,
Processes 2024, 12,
12, 2357
x FOR PEER REVIEW 18
16of 22
of 20

100

98

96
Measuremnts %

94

92

90

88

86

84
BE-HHO AD-HHO SVM-HHO KNN-HHO DT-HHO NB-HHO
Optmized ML Models

Accuracy Precision Recall F1-score

Figure 11. Results for optimized ML algorithms.

Figure 11. Results for optimized ML algorithms.

Table 4. Comparative study between optimized ML and classical ML.

Table 4. Comparative study between optimized ML and classical ML.
Algorithm Accuracy Precision Recall F1-Score
Algorithm Accuracy Precision Recall F1-Score
BE-HHO 95.24 93.854 96.447 95.133
BE BE-HHO 95.17 95.24 93.6293.854 96.58
96.447 95.08
95.133
AD-HHO BE 92.57 95.17 89.44893.62 95.921
96.58 92.571
95.08
AD AD-HHO 92.06 92.57 87.9389.448 95.84
95.921 92.17
92.571
SVM-HHO AD 94.86 92.06 91.75987.93 98.158
95.84 94.851
92.17
SVM SVM-HHO 94.48 94.86 90.2491.759 99.61
98.158 94.68
94.851
KNN-HHO 94.73 92.472 96.974 94.669
SVM 94.48 90.24 99.61 94.68
KNNs 93.14 87.64 99.87 93.36
KNN-HHO 94.73 92.472 96.974 94.669
DT-HHO 94.73 93.231 96.053 94.621
KNNs 93.14 87.64 99.87 93.36
DT 94.48 93.09 95.66 94.35
NB-HHO DT-HHO 94.73 94.73 92.47293.231 96.974
96.053 94.669
94.621
NB DT 94.6 94.48 91.9393.09 97.37
95.66 94.57
94.35
NB-HHO 94.73 92.472 96.974 94.669
TheNBBE-HHO shows94.6 slightly higher performance
91.93 than BE, with an accuracy
97.37 of 95.24%
94.57
compared to 95.17%, and similarly high precision, recall, and F1-Scores. AD-HHO and
AD have a notable difference in accuracy, with AD-HHO at 92.57% and AD at 92.06%,
From
though ADthe resultshigher
achieves above,recall.
it is evident that different
The SVM-HHO models
variant have
exhibits antheir own of
accuracy strengths,
94.86%,
with the best-performing ones being evaluated based on key metrics such
slightly higher than the classical SVM at 94.48%, with both showing high recall as accuracy,
values.
precision,
KNN-HHOrecall, and KNNs
surpasses F1-Score. These metrics
in accuracy (94.73%provide a more
vs. 93.14%) general view improved
and demonstrates of model
performance, beyond just accuracy. The BE-HHO model leads with the highest accuracy at
precision and recall. DT-HHO and DT both have an accuracy of 94.73%, but DT-HHO has
95.24%, followed closely by the BE model at 95.17%. Next, the SVM-HHO model has an
marginally better precision and recall. NB-HHO and NB are almost equivalent, with NB-
accuracy of 94.86%, slightly outperforming the classical SVM at 94.48%. KNN-HHO and
HHO having a slightly higher accuracy of 94.73% compared to 94.6%. Overall, the opti-
DT-HHO both have an accuracy of 94.73%, with KNN-HHO marginally outperforming
mized algorithms demonstrate marginal but consistent improvements over their classical
classical KNNs (93.14%) and DT-HHO closely matching classical DT (94.48%). The NB-
versions in most metrics, as clarified in Figure 12.
HHO model also has an accuracy of 94.73%, slightly higher than the classical NB at 94.6%.
AD-HHO and AD have the lowest accuracies at 92.57% and 92.06%, respectively, but AD
achieves the highest recall among all models. Therefore, the ranking of models based
on their performance is as follows: BE-HHO, BE, SVM-HHO, DT-HHO, KNN-HHO, NB-
Processes 2024, 12, 2357 17 of 20

HHO, SVM, DT, NB, KNN, AD-HHO, and AD. Small improvements in recall, accuracy, or
precision can have a significant impact on predictive maintenance, fault detection quality,
and system reliability in industrial applications.. Any improvement, no matter how little,
helps to increase operational efficiency, lower risk, and improve decision-making in real-
Processes 2024, 12, x FOR PEER REVIEW 19 of 22
world systems where faulty or inaccurate diagnostics can cause expensive disruptions. So,
the total cost of maintenance will be reduced.

100
Measurements %

95

90

85
Recall

80 Accuracy

Models

Accuracy Precision Recall F1-score

Figure 12. A comparison between ML and optimized ML models.

Figure 12. A comparison between ML and optimized ML models.

From 5the
Table results above,
provides it is evident
a comparative that of
analysis different models
different have their
optimization own strengths,
strategies that can
with the best-performing ones being evaluated based on key
be used to improve the accuracy of the Support Vector Machine (SVM) algorithm metrics such as accuracy,
based
precision,
on accuracy, recall, and F1-Score.
precision, recall, These metrics provide
and F1-Score. Amonga more general view
the different methodsof model per-
evaluated,
formance, beyond just accuracy. The BE-HHO model leads with
the Harris Hawk Optimization (HHO) method emerges as the leader with a remarkable the highest accuracy at
95.24%, followed closely by the BE model at 95.17%. Next, the SVM-HHO
accuracy of 94.86%, a precision of 91.76%, and an F1-Score of 94.85 for this task, clearly model has an
accuracy of 94.86%,
demonstrating slightly
its ability outperforming
in the detection ofthe classical
faults. SVM at 94.48%.
Considering KNN-HHO
the results and
of all covered
DT-HHO both have an accuracy of 94.73%, with KNN-HHO marginally
methods, the HHO appears to be the most successful optimization approach for enhancing outperforming
classical KNNs
ML-based (93.14%) and DT-HHO closely matching classical DT (94.48%). The NB-
fault detection.
HHO model also has an accuracy of 94.73%, slightly higher than the classical NB at 94.6%.
AD-HHO
Table and AD have
5. Comparative studythe lowestdifferent
between accuracies at 92.57%methods
optimization and 92.06%, respectively, but AD
for SVM.
achieves the highest recall among all models. Therefore, the ranking of models based on
their performance
Optimization is as follows: BE-HHO,
Accuracy BE, SVM-HHO,
Precision DT-HHO,
Recall KNN-HHO, F1 NB-
HHO,
ParticalSVM,
SwarmDT,Optimization
NB, KNN, AD-HHO, and AD. Small
94.35 improvements
90.9646 98.0263in recall, 94.3635
accuracy,
orWhale
precision can have a significant
Optimization Algorith impact
92.8889on predictive
88.57maintenance, fault
97.8947 detection93qual-
ity, and system reliability in industrial applications.. Any improvement, no matter how
Slime Mould Algorithm 90.6032 85.4988 96.9737 90.8755
little, helps to increase operational efficiency, lower risk, and improve decision-making in
real-world systems where faulty or inaccurate diagnostics can cause expensive
Sine Cosine Algorithm 93.1429 89.2771 97.5000 disrup-
93.2075
tions. Generalized
So, the totalNormal
cost of maintenance will be reduced.
94.1587 90.6326 98.0263 94.1846
Table 5 provides
Distribution a comparative analysis of different optimization strategies that can
Optimization
be used to improve
Genatic the accuracy of93.6508
Algorithm the Support Vector
89.7590Machine98.0263
(SVM) algorithm based
93.7107
on accuracy, precision, recall, and F1-Score. Among the different methods evaluated, the
HHO 94.8571 91.7589 98.1579 94.8506
Harris Hawk Optimization (HHO) method emerges as the leader with a remarkable ac-
curacy of 94.86%, a precision of 91.76%, and an F1-Score of 94.85 for this task, clearly
demonstrating its ability in the detection of faults. Considering the results of all covered
methods, the HHO appears to be the most successful optimization approach for enhanc-
ing ML-based fault detection.
Processes 2024, 12, 2357 18 of 20

7. Limitation
While the research shows good prospects with using the HHO with CGANs in digital
twins for fault detection, some limitations have to be addressed to improve the robustness
and scalability of the proposed approach:
• Computation complexity and cost: The use of the HHO along with CGANs would
increase the computational complexity and cost. Both methods would involve heavy
computational loads—both the optimization for the HHO and CGANs, like any other
generative model, tend to require power-intensive calculations because of the need
to fit an auxiliary conditional variable in between, creating generator and discrimi-
nator networks. When these two are put together, the computational burden could
increase even more, thus inhibiting real-time applications of digital twins in fast fault
detection processes.
• Overfitting and biases in fault: As in the case of every other machine learning model,
there might be tendency of overfitting with the CGAN model, especially if it is trained
on small or unbalanced datasets. When the training data do not cover enough fault
types or some operating regimes, the learning algorithm may ‘overfit’ and learn to
always favor certain types of failures/diminish the importance of failures that are
less treated or are deemed rare. Such risk is more pronounced in large-scale complex
industrial systems where there are tendencies of faults happening frequently and
the amount of training data is small. Although the HHO is useful for the purpose of
finding the optimal parameter values, it is not a solution to the problem of limited data.
• Parameter sensitivity of HHO: In the absence of the self tuning nature of the HHO and
other metaheuristic algorithms, initialization parameter values in those cases tend to
impact the success of the HHO. With a small change in some parameters such as the
population size or even the number of iterations, the performance experienced may
be totally different. Such sensitivity probably causes the results obtained in different
scenarios of fault detection tasks to be quite different, most probably when real-time
data processing is involved, in particular in manufacturing systems.
• Data quality and representation: The effectiveness of Generative Artificial Intelligence
such as the CGAN heavily relies on the quality of the data used for training. Inaccurate,
incomplete, or biased datasets can lead to poor fault detection performance, particu-
larly when digital twins are expected to simulate real-world operational conditions.
These constraints can also be addressed and further researched to enhance the integra-
tion of the HHO with Generative Artificial Intelligence in digital twins for enhanced fault
detection, making it more effective, efficient, and applicable to more industries.

8. Conclusions and Future Scope

In recent years, fault detection has emerged as a pivotal component within industrial
DT systems, crucial for ensuring operational reliability and efficiency. This process involves
identifying deviations from expected behavior or performance in physical assets mirrored
by their digital representations. Furthermore, fault detection facilitates the implementation
of predictive maintenance strategies, enabling targeted repairs and the optimization of
asset performance in the industrial DT. This study proposes an efficient fault detection
framework leveraging conditional Generative Adversarial Networks (GANs) and meta-
heuristic machine learning for fault detection. The experimental results demonstrate that
hybrid-optimized ML algorithms, denoted as “ML-HHO”, consistently outperform or
closely match their classical counterparts across various performance metrics. Particularly,
BE-HHO achieves the highest accuracy at 95.24%, slightly surpassing BE’s 95.17%. Sim-
ilarly, SVM-HHO achieves an accuracy of 94.86%, marginally exceeding SVM’s 94.48%.
KNN-HHO exhibits superior performance with 94.73% accuracy compared to KNN’s
93.14%. DT-HHO and DT both achieve 94.73% accuracy, with DT-HHO showing slightly
better precision and recall. NB-HHO and NB demonstrate comparable performance, with
NB-HHO achieving 94.73% accuracy against NB’s 94.6%. Despite lower accuracies, AD-
HHO and AD achieve higher recall rates at 92.57% and 92.06%, respectively. Overall, the
Processes 2024, 12, 2357 19 of 20

optimized algorithms consistently deliver marginal enhancements over their classical coun-
terparts. With respect to synthetic data generation, the framework employs CGANs and
feature selection is carried out using the HHO, which makes the framework a powerful and
adaptable architecture for continuous monitoring and diagnosis. As a result, operational
efficiency is enhanced, downtime is reduced, and enhanced system uptime is observed. In
addition, the ability of this approach to fit various industrial applications indicates greater
prospects for enhancing predictive maintenance and improvement strategies in several
applications over time.
Future research could explore the integration of other machine learning methods
for anomaly detection techniques such as Autoencoders, Deep Belief Networks, deep
reinforcement learning, or Recurrent Neural Networks and incorporate them into real-
time data streams for continuous monitoring and adaptive fault detection in dynamic
industrial DT systems. Expanding the framework beyond pumps opens up exciting
avenues for research in a range of industries that rely on dynamic equipment such as
turbines, compressors, and other industrial processes. Furthermore, efforts could be
directed toward improving CGANs’ performances by fine-tuning the hyperparameters,
such as learning rates, batch sizes, and the architecture of the generator and discriminator.

Author Contributions: Conceptualization and methodology, A.S. and E.E.-D.H.; software, validation
and formal analysis A.S.; investigation, E.E.-D.H.; writing—original draft preparation A.S., writing—
review and editing, S.A., E.E.-D.H. and A.S.; project administration and funding acquisition, S.A. All
authors have read and agreed to the published version of the manuscript.
Funding: this project is funded by Princess Nourah bint Abdulrahman University Researchers
Supporting Project number (PNURSP2024R197), Princess Nourah bint Abdulrahman University,
Riyadh, Saudi Arabia.
Data Availability Statement: The original contributions presented in the study are included in the
article; further inquiries can be directed to the corresponding author.
Acknowledgments: Princess Nourah bint Abdulrahman University Researchers Supporting Project
number (PNURSP2024R197), Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia.
Conflicts of Interest: The authors declare that they have no conflicts of interest to report regarding
the present study.

References
1. Bin, B.; Jian, K. Digital twin-based sustainable intelligent manufacturing: A review. Adv. Manuf. 2021, 9, 1–21.
2. Liu, Y.; Zhang, L.; Yuan, Y.; Zhou, L.; Ren, L.; Wang, F.; Liu, R.; Pang, Z.; Jamal Deen, M. A novel cloud- based framework for
elderly healthcare Services using a digital twin. IEEE Access 2019, 7, 49088–49101. [CrossRef]
3. Caputo, F.; Greco, A.; Fera, M.; Macchiaroli, R. Digital twins to enhance the integration of ergonomics in workplace design. Int. J.
Ind. Ergon. 2019, 71, 20–31. [CrossRef]
4. Zayed, S.M.; Attiya, G.; El-Sayed, A.; Sayed, A.; Hemdan, E.E.-D. An Efficient Fault Diagnosis Framework for Digital Twins Using
Optimized Machine Learning Models in Smart Industrial Control Systems. Int. J. Comput. Intell. Syst. 2023, 16, 69. [CrossRef]
5. Hemdan, E.E.-D.; El-Shafai, W.; Sayed, A. Integrating Digital Twins with IoT-Based Blockchain: Concept, Architecture, Challenges,
and Future Scope. Wirel. Pers. Commun. 2023, 131, 2193–2216. [CrossRef] [PubMed]
6. Pedersen, A.N.; Borup, M.; Brink-Kjær, A.; Christiansen, L.E.; Mikkelsen, P.S. Living and prototyping digital twins for urban
water systems: Towards multi-purpose value creation using models and sensors. Water 2021, 13, 592. [CrossRef]
7. Pylianidis, C.; Osinga, S.; Athanasiadis, I.N. Introducing digital twins to agriculture. Comput. Electron. Agric. 2021, 184, 105942.
[CrossRef]
8. Verdouw, C.; Tekinerdogan, B.; Beulens, A.; Wolfert, S. Digital twins in smart farming. Agric. Syst. 2021, 189, 103046. [CrossRef]
9. Neethirajan, S.; Kemp, B. Digital Twins in Livestock Farming. Animals 2021, 4, 1008. [CrossRef]
10. Nativi, S.; Mazzetti, P.; Craglia, M. Digital ecosystems for developing digital twins of the earth: The destination earth case. Remote
Sens. 2021, 13, 2119. [CrossRef]
11. Guo, H.; Nativi, S.; Liang, D.; Craglia, M.; Wang, L.; Schade, S.; Corban, C.; He, G.; Pesaresi, M.; Li, J.; et al. Big Earth Data science:
An information framework for a sustainable planet. Int. J. Digit. Earth 2020, 13, 743–767. [CrossRef]
12. Liu, H.; Zhou, J.; Xu, Y.; Zheng, Y.; Peng, X.; Jiang, W. Unsupervised fault diagnosis of rolling bearings using a deep neural
network based on generative adversarial networks. Neurocomputing 2018, 315, 412–424. [CrossRef]
Processes 2024, 12, 2357 20 of 20

13. Sabuhi, M.; Zhou, M.; Bezemer, C.-P.; Musilek, P. Applications of generative adversarial networks in anomaly detection: A
systematic literature review. IEEE Access 2021, 9, 161003–161029. [CrossRef]
14. Lian, Y.; Geng, Y.; Tian, T. Anomaly Detection Method for Multivariate Time Series Data of Oil and Gas Stations Based on Digital
Twin and MTAD-GAN. Appl. Sci. 2023, 13, 1891. [CrossRef]
15. Liu, H.; Zhao, H.; Wang, J.; Yuan, S.; Feng, W. LSTM-GAN-AE: A promising approach for fault diagnosis in machine health
monitoring. IEEE Trans. Instrum. Meas. 2021, 71, 1–13. [CrossRef]
16. Li, W.; Li, H.; Gu, S.; Chen, T. Process fault diagnosis with model-and knowledge-based approaches: Advances and opportunities.
Control Eng. Pract. 2020, 105, 104637. [CrossRef]
17. Syed, M.M.; Lemma, T.A.; Vandrangi, S.K.; Ofei, T.N. Recent developments in model-based fault detection and diagnostics of gas
pipelines under transient conditions. J. Nat. Gas Sci. Eng. 2020, 83, 103550. [CrossRef]
18. Costamagna, P.; De Giorgi, A.; Magistri, L.; Moser, G.; Pellaco, L.; Trucco, A. A classification approach for model-based fault
diagnosis in power generation systems based on solid oxide fuel cells. IEEE Trans. Energy Convers. 2015, 31, 676–687. [CrossRef]
19. Grieves, M. Digital Twin: Manufacturing Excellence Through Virtual Factory Replication. White Pap. 2014, 1, 1–7.
20. Rasheed, A.; San, O.; Kvamsdal, T. Digital Twin: Values, Challenges and Enablers from a Modeling Perspective. IEEE Access 2020,
8, 21980–22012. [CrossRef]
21. Jones, D.; Snider, C.; Nassehi, A.; Yon, J.; Hicks, B. Characterising the Digital Twin: A Systematic Literature Review. CIRP J. Manuf.
Sci. Technol. 2020, 29, 36–52. [CrossRef]
22. Wang, H.; Chen, X.; Jia, F.; Cheng, X. Digital twin-supported smart city: Status, challenges and future research directions. Expert
Syst. Appl. 2023, 217, 119531. [CrossRef]
23. Alshathri, S.; Hemdan, E.E.-D.; El-Shafai, W.; Sayed, A. Digital twin-based automated fault diagnosis in industrial IoT applications.
Comput. Mater. Contin. 2023, 75, 183–196. [CrossRef]
24. Rachmawati, S.M.; Putra, M.A.P.; Lee, J.M.; Kim, D.S. Digital twin-enabled 3D printer fault detection for smart additive
manufacturing. Eng. Appl. Artif. Intell. 2023, 124, 106430. [CrossRef]
25. Kuru, K. MetaOmniCity: Towards immersive urban metaverse cyberspaces using smart city digital twins. IEEE Access 2023, 11,
43844–43868. [CrossRef]
26. Mirza, M.; Simon, O. Conditional generative adversarial nets. arXiv 2014, arXiv:1411.1784.
27. Heidari, A.A.; Mirjalili, S.; Faris, H.; Aljarah, I.; Mafarja, M.; Chen, H. Harris hawks optimization: Algorithm and applications.
Future Gener. Comput. Syst. 2019, 97, 849–872. [CrossRef]
28. Multi-Class Fault Detection Using Simulated Data. Available online: https://ssd.mathworks.com/supportfiles/SPT/data/
PumpSignalGAN.zip (accessed on 14 August 2024).
29. El-Naby, A.A.; Hemdan, E.E.-D.; El-Sayed, A. An efficient fraud detection framework with credit card imbalanced data in financial
services. Multimed. Tools Appl. 2023, 82, 4139–4160. [CrossRef]
30. Rezk, N.G.; Attia, A.-F.; El-Rashidy, M.A.; El-Sayed, A.; Hemdan, E.E.-D. An Efficient Plant Disease Recognition System
Using Hybrid Convolutional Neural Networks (CNNs) and Conditional Random Fields (CRFs) for Smart IoT Applications in
Agriculture. Int. J. Comput. Intell. Syst. 2022, 15, 65. [CrossRef]
31. Sharaf, M.; Hemdan, E.E.; El-Sayed, A.; El-Bahnasawy, N.A. An efficient hybrid stock trend prediction system during COVID-19
pandemic based on stacked-LSTM and news sentiment analysis. Multimed. Tools Appl. 2023, 82, 23945–23977. [CrossRef]
32. Abd El Naby, A.; Hemdan, E.E.D.; El-Sayed, A. Deep learning approach for credit card fraud detection. In Proceedings of the
2021 International Conference on Electronic Engineering (ICEEM), Menouf, Egypt, 3–4 July 2021; IEEE: New York, NY, USA, 2021;
pp. 1–5.

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.