Measurement: Sensors 25 (2023) 100661

Contents lists available at ScienceDirect

Measurement: Sensors
journal homepage: www.sciencedirect.com/journal/measurement-sensors

Digital twin and IOT technology for secure manufacturing systems

Lisa Gopal a, *, Harbaksh Singh b, Panguluri Mounica c, N. Mohankumar d,
Nagendra Panini Challa e, P. Jayaraman f
a
Computer Science and Engineering, Graphic Era Hill University, Dehradun, M.Tech, Graphic Era Deemed to Be University, Dehradun, 248002, Uttarakhand, India
b
BITS Pilani, India
c
Department of CSE, Koneru Lakshmaiah Education Foundation, Vijayawada, Andhra Pradesh, India
d
Department of Electronics and Communication Engineering,Amrita School of Engineering Coimbatore, Amrita Vishwa Vidyapeetham, India
e
School of Computer Science and Engineering, VIT-AP University, Amaravati, 522237, Andhra Pradesh, India
f
Department of Mechanical Engineering, Prathyusha Engineering College, Aranvoyalkuppam, Tiruvallur, Chennai, Tamil Nadu, 602025, India

A R T I C L E I N F O A B S T R A C T

Keywords: The Digital Twin (DT) method offers new concepts for choosing the context of information technologies and
Industry 4.0 smart production systems. This research focuses on low-price, extremely effective defect diagnosis methods, low-
IoT efficiency, high-cost devices to obtain timely feedback and accurate fault detection results, and secure
Digital twin
manufacturing systems. The data structure, control plane, and output units are the three components of the
Fault diagnosis
manufacturing system that creates a data link between the virtual model using Micro-Electro-Mechanical (MEM)
devices and the Zigbee wireless transmission system in the database layer. This study acquired DT information
from the control plane using the Internet of Things (IoT) through sensors for secure manufacturing information.
It separates and calls the pertinent data by the attribute processor and transfers it to the outcome units. To
produce the classification and outcomes of work build features information, the evaluation method analysis the
output nodes that split the test set and learning group using a dynamic database. The hybrid IoT with DT
technology to examine the consequences of defects detected efficiently predicted and secures the manufacturing
system.

1. Introduction Learning (DL) methods, cloud frameworks, and high-performance sys­

tems [9,10]. The DT makes a highly accurate virtual system that de­
A dynamic virtual network, procedure, or function was referred to as velops with the lifecycle. On the other hand, with the advent of digital
a “digital twin”. In the context of IoT, it has been shown that DT tech­ design tools such as 3-D Computer Aided Design (CAD) building infor­
nology is cost-effective [1]. Integrating DT technology with enterprise mation modeling, and Finite-Element Analysis (FEA), the construction
IoT offers broad growth potential for large-scale, complex, and phase is becoming more and more digitized making it easier to develop
capital-intensive infrastructure [2,3]. High-end production would be a electronic information relating to infrastructure and frameworks [11].
crucial indicator of the basic competitiveness of a country or region as This paper introduces a cloud-based DT approach for structural secure
the foundation of the country’s economy. The industrial internet management that aims to execute continual assessment & predictive
becoming increasingly well-known in the industrial sector as connec­ maintenance regularly transferring data from actual things to digital
tivity and manufacturing technology advance [4]. Businesses have equivalents [12]. Through the evolution of monitoring services from
greater standards for the effectiveness of data communication and regular, generic, or physics-based approaches for real-time, individual­
decision-making. The conventional problem monitoring and assessment ized, data-driven models, maintenance strategies were optimized the
techniques involving individuals were far from present requirements or machine’s dependability security is improved.
complex processes.
A new data-driven framework known as the DT has recently emerged 2. Related works
with increased attention [5–8]. This is due to remarkable advances in
technologies like 5G, Wireless Sensor Networks (WSN), IoT, Deep The combination of these factors has led to the development of new

* Corresponding author.
E-mail address: [email protected] (L. Gopal).

https://doi.org/10.1016/j.measen.2022.100661
Received 15 August 2022; Accepted 29 December 2022
Available online 31 December 2022
2665-9174/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
L. Gopal et al. Measurement: Sensors 25 (2023) 100661

Fig. 1. Machines construction structure (a) Basic structure (b) Smart DT structure.

“DT technologies” that operate in a physical and digital setting poten­ surroundings that has been improved by the addition of digital and
tially enhancing commercial safety and protection among other things. computer-generated data. This method has applications in a wide range
This paper emphasizes industrial security especially based on contem­ of contexts, including manufacturing, security, and learning. Smart
porary communication methods that could detect employment threats glasses, virtual retinal monitors, and head-mounted showcases are all
and their effects. Modeling the DT, Augmented Reality (AR), and the used in AR. Smart glasses have the benefit of providing data needed to
commercial IoT methods employed for this purpose [13–15]. The defi­ manage a particular event or task effectively in actual time [16].
nition of AR would be an actual time, direct or indirect image of physical IoT has an interconnection of computer gadgets into common things

2
L. Gopal et al. Measurement: Sensors 25 (2023) 100661

them using certain methods or classifiers, and then the output of data
characteristics processed into a dynamic dataset. Finally, this system
generates data features that can be used for finding erroneous results.
According to this, the procedure could provide the information by
modifying the structure or predicting the defects completely based on
the information observed.

3.1. Function components

The DT technology has broken down into three layers: the level of
information, the control plan, and the output units depending on how
the information is conveyed. Fig. 2 illustrates the structure of the system.
Fig. 2. The framework for the system components. Four modules are used in the system that employs a modularization
prototype to achieve hierarchical components: the data gathering, the
transmission device, the failure attributes information storage and
which permits the exchange or transmission of information that con­
management module, the defect detection or forecast subsystem, and
nects all portable technologies. Data between physical and virtual spaces
the ROC assessment device.
merged through DT with these innovations. The conventional concept of
a virtual DT should be a representation that captures the many stages of
3.2. Data collection
a manufacturer’s or factory’s lifecycle and reflects into the real world
[17]. Integrating modern technology solves the current problems and
Monitoring systems through sensors used to capture the data char­
promotes secure manufacturing systems. As a result, the purpose of the
acteristic of the DT system [24]. A distributed control method was
work is to create and integrate the software for tracking the pressures
deployed to the exterior framework and incorporated the gathered in­
placed on loaded metal shelves that should enable users to assess the
formation about the system modification, maintenance, and surround­
stress level in actual time by connecting the required sensors to AR de­
ings [25]. Continuously collecting the information from the sensor could
vices [18]. In this application, many hardware elements, software, and
be utilized to anticipate the prospects of the internet using big data,
sensors are used as a prototype system that has been developed. The
dynamic information assessment, and Judgment framework systems to
three devices installed on the shelf that is connected to cloud storage
monitor and secure the manufacturing information. MEM devices are a
would allow the administrator to obtain extra information through
novel method of chip production contains micro/nanoscale mechanical
smart glasses [19]. Information was automatically downloaded onto the
structures that gather the vibration analysis for delicate sections with
internet and the simulation model was associated when the manager
traditional semiconductor circuits. It gathers ecological information
scanned the QR code on the rack. Then it generates the structural
such as temperature, moisture, motion, or working velocity of the ma­
analysis so that the manager can view using the smart glasses.
chines using MEMS detectors. Zigbee network technology allows data
Measuring sensors glued to the frame modify the electrical resistance
connectivity between different sensors, creating the base layer of in­
in a way appropriate to the actual deformation used in the creation of
formation or complementing the actual data collection.
the application. A Wheatstone bridge should be used to monitor changes
in resistance, and the values it produces are then gathered by an Arduino
card & relayed via Wi-Fi to the ThingSpeak framework for data gath­ 3.3. Digital twin and twin data
ering & development [20]. The producers integrated into AR devices
could assess the component’s stress condition in actual time and get a A DT is a technical system designed to create digital data that covers
sense of the pressures that the component has endured in the past [21]. virtually the entire experiment. This is one of the effective approaches to
Data obtained by strain gauges could be compared with traditional implementing dual-purpose engineering. Real-time assessment of con­
values and criteria values readily accessible by contractors [22]. The ditioning the system and the numerical model increase the accuracy of
item has been unambiguously recognized as a QR code that lists key the diagnosis, assessment, and prediction of the DT [26]. Online man­
characteristics and adds it to the digital factory through the idea of the agement system operation maintenance and administration could
IoT. The F4 spectacles that have been employed in this application reduce structural design variability, prevent regular repairs, or ensure
rationale are illustrated in Fig. 1. system safety. Real-time observation data enhance or expand the model
to integrate all network information. For complex processes, the DT
3. Proposed methodology combines all components, providing the basis for high-accuracy
modeling. The central driver of the DT provides precise and compre­
This article presents the DT technique and builds & improves the hensive data resources for merging virtual and physical elements.
existing approach based on features & essential information from the Complexity, compatibility, and the lack of coordination of information
smart equipment for the conventional fault detection procedure. The merging were caused by the diversity of data categorization and the
operation of smart equipment suited to this technology provides less cost inconsistency of data sets. To improve diagnostics, prediction, and
testing and gets real-time feedback. Fig. 1 depicts the machine’s me­ decision-making systems used to discover the hidden link between
chanical structure. The detectors incorporate intelligent instrumenta­ heterogeneous data gathered from various sources of information.
tion and their digital models to provide data connection. The
functioning of intelligence smart equipment and tools on the technical 3.4. Information management system
side generate a significant amount of historical information gathered by
detectors that are saved in the dataset to create a data connection, or To create a consistent characterization & identification of various
transferred using the DT concept [23]. Mapping information is first power types, the framework of data was focused on the specification of
processed through the database administration system, which then re­ the CIM prototype of IEC61970, and the constant computation carried
cords it for future access. out on the relational model. The solution was centered on a big data
The DT model has appropriate performance standards and matching analytics framework and a decentralized big data cloud service. The
variables based on real needs. The assessment structure method sum­ cloud-based platform was designed to store or manage huge amounts of
marizes the processed data and performance-related variables, filters data and relies on technologies like Hadoop, Spark, and others. Struc­
tures can provide good operation and storage management, download

3
L. Gopal et al. Measurement: Sensors 25 (2023) 100661

Fig. 3. The process of fault detection.

data to information management systems, and work with both organized

and unorganized data. The MEM device structure of the commercial Table 1
Confusion matrix of the system.
Internet consists mainly of the dynamic analytical data controller or
decision-making technology that forms the foundation of the digital Positive Judgment Negative Judgment
information management system. The two aspects of DT are continuous Positive Judgment True Positive Records False Positive Records
development and actual time connection. A new simulation application Negative Judgment False Positive Records True Positive Records
mode called Dynamic Data Driven Application Systems could of course
integrate design into the physical process. The updated design could
obtain a lot of data that cannot be produced immediately during its
operation, allowing for better prediction of the system and assisting the
decision-maker. For this purpose, real-time monitoring data is used to
continuously modify the model.

3.5. Fault diagnosis

Time nodes are used the record tags of ecological information such as
temperature and moisture measured by MEMS sensors; vital smart
equipment attributes such as resonant frequencies. The direction of the
link structure’s impact, the operating characteristics or speed of com­
ponents that lead to destabilization errors; or through data like the de­
gree of fit between a portion of the element waveforms selected by the
data management system and the reference waveform.
This paper was focused on the secure method to develop an uncon­
trolled method to produce the persistent vector space of independent Fig. 4. ROC curve diagram.
information because the training dataset of the DT network is quite
huge. The large vector size also responds to the issue of the information
4. Stochastic optimization
collection’s multidimensional dispersion, guaranteeing the precision of
the model. This approach enables the data scale to be condensed while
The Confusion Matrix summarizes the records in the information
preserving the extracted characteristics. The characteristic dataset in­
source as a vector and bases its conclusions on the vector classification
cludes all-time components presented as a D-dimensional matrix in the
algorithm. Table 1 displays the confusion matrix’s organizational
input layer, where D would be the maximum number of personal data in
layout. A confusion matrix method for assessing the categorization of
the data set. This method of one-hot encoding was used to decode each
work and feature method yields the likelihood to choose a threshold.
word. The dimension of the vector that corresponds to each piece of data
Various substitution matrices can be created by adjusting the threshold
was changed to 1, while the other parameters’ values were changed to 0.
several times will change the ROC and the promotional parameter curve.
The entire function is extracted through the Long Short Term
The effectiveness of the classification model could be further assessed
Memory (LSTM) array, as shown in Fig. 3. The Dropout stochastic
and compared using the proposed approach. It is simple to adapt the
optimization element has been introduced to the LSTM network to allow
centralized confusion matrix of the binary categorization problem pre­
the model to better integrate with complex datasets while preventing the
sented to the multi-variable classification problem.
model from overflowing and securing the manufacturing information.
The ROC curve was originally used for signal processing. The false
Increase the model effectively with output value ranges between [0, 1].
positive ratio was represented by the X-axis and the false positive ratio
The following are some examples of this method’s exploratory stages:
by the Y-axis on the ROC curve [27]. The classifier-based problem
the SVC technique, picking the RBF Kernel to create and develop the
predictive model uses different categorization techniques, and the
model, and getting the classification algorithm variables C = 1.28, g =
unpredictability of its general characteristics may have an impact on
3.56. Calibrate training information, such as the testing phase.
how much emphasis was given to various types of information. To ac­
quire classification performance, a general black box type prediction

4
L. Gopal et al. Measurement: Sensors 25 (2023) 100661

5. Conclusions

This paper focuses on low-price, extremely effective defect diagnosis

methods, low-efficiency, high-cost devices to obtain timely feedback
and accurate fault detection results, and secure manufacturing systems.
The hybrid IoT with DT technology to examine the consequences of
defects detected efficiently predicted and secures the manufacturing
system. The primary results of this investigation as follows: (1) The
managing damages in precision instrument components issues are
examined through literature study and field data collecting from IoT. (2)
A module with a defect prediction capability was created using the DT
technique. (3) The system architecture of damage detection feature is
created to verify the model optimization process, and several assessment
techniques of the manufacturing systems. The proposed method
continuously evaluate the pressures of machine systems, resulting in
advantages for employee safety and quick operational response in the
event of warning messages and secure the manufacturing information.

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.

References

[1] W. Shen, T. Hu, C. Zhang, S. Ma, Secure sharing of big digital twin data for smart
Fig. 5. The variable and condition of the equipment. manufacturing based on blockchain, J. Manuf. Syst. 61 (2021) 338–350.
[2] J. Leng, D. Wang, W. Shen, X. Li, Q. Liu, X. Chen, Digital twins-based smart
manufacturing system design in Industry 4.0: a review, J. Manuf. Syst. 60 (2021)
system that can accept many forms of data input is necessary. The last 119–137.
stage consists in regularly preparing the model to correspond to the [3] J. Leng, M. Zhou, Y. Xiao, H. Zhang, Q. Liu, W. Shen, L. Li, Digital twins-based
remote semi-physical commissioning of flow-type smart manufacturing systems,
benchmark, which should be the AUC index associated with the ROC J. Clean. Prod. 306 (2021), 127278.
matrix. [4] J. Leng, M. Zhou, Y. Xiao, H. Zhang, Q. Liu, W. Shen, L. Li, Digital twins-based
The ROC curve graph was illustrated in Fig. 4, and the AUC variable remote semi-physical commissioning of flow-type smart manufacturing systems,
J. Clean. Prod. 306 (2021), 127278.
would be an indication of the ROC curve. The relevance of the AUC
[5] R. Sahal, S.H. Alsamhi, K.N. Brown, D. O’Shea, C. McCarthy, M. Guizani,
variable is located in the region bounded by the ROC curve and the Blockchain-empowered digital twins collaboration: smart transportation use case,
horizontal plane. The bigger the AUC score, the lower the threshold and Machines 9 (9) (2021) 193.
the more successful the model was in resolving a long-term classification [6] G.N.R. Devi, S. Ponnusamy, J. Pandit, B. Buvaneswari, Development of medicinal
industries in building A replica to the damaged human tissue for artificial organs
issue. The AUC was determined by the following Equation (1) with the application of micro-and nano technology (mnt), J. Optoelectron. - Laser
41 (3) (2022) 79–83.
Σ instant Rins − P X (P+1) [7] M. Mashaly, Connecting the twins: a review on Digital Twin technology & its
AUC = 2
(1) networking requirements, Procedia Comput. Sci. 184 (2021) 299–305.
PXQ
[8] X. Liu, Y. Jiang, Z. Wang, R.Y. Zhong, H.H. Cheung, G.Q. Huang, imseStudio:
blockchain-enabled secure digital twin platform for service manufacturing, Int. J.
Where P is the number of positive samples, and Q is the number of Prod. Res. (2021) 1–20.
negative samples. [9] V. Vimal, K.U. Singh, A. Kumar, M. Gupta, R.K. Saket, S. Padmanaban, Clustering
isolated nodes to enhance network’s life time of WSNs for IoT applications, IEEE
Syst. J. 15 (4) (2021) 5654–5663, https://doi.org/10.1109/JSYST.2021.3103696.
4.1. Observation [10] V. Vimal, S. Maheshwari, Energy management routing in wireless sensor networks,
Adv. Intell. Syst. Comput. 258 (2014) 161–168, https://doi.org/10.1007/978-81-
A set of defect detection and prediction systems for smart equipment 322-1771-8_14. ISBN: 978-813221770-1.
[11] T. Mortlock, D. Muthirayan, S.Y. Yu, P.P. Khargonekar, M.A. Al Faruque, Graph
were developed in this work [26,27] to achieve high-efficiency, data-­ learning for cognitive digital twins in manufacturing systems, IEEE Transactions on
based, and smart defect detection and prediction of high-cost, low-yield Emerging Topics in Computing 10 (1) (2021) 34–45.
technology. The system could self-analyze, has high learning capabil­ [12] C. Qian, X. Liu, C. Ripley, M. Qian, F. Liang, W. Yu, Digital twin—cyber replica of
physical things: architecture, applications, and future research directions, Future
ities, increases efficiency, accelerates the transfer of defect data, and can Internet 14 (2) (2022) 64.
function in the manufacturing unit intelligently compared to the con­ [13] Q. Qi, F. Tao, T. Hu, N. Anwer, A. Liu, Y. Wei, A.Y.C. Nee, Enabling technologies
ventional detection procedure. Fig. 5 illustrates how the predictive and tools for the digital twin, J. Manuf. Syst. 58 (2021) 3–21.
[14] S. Almeaibed, S. Al-Rubaye, A. Tsourdos, N.P. Avdelidis, Digital twin analysis to
failure model categorizes current flow conditions using real data cor­
promote safety and security in autonomous vehicles, IEEE Communications
relation to arrive at the expected result. The defect prediction perfor­ Standards Magazine 5 (1) (2021) 40–46.
mance of the transverse stress loading element and the composite sector [15] G. Cathey, J. Benson, M. Gupta, R. Sandhu, Edge centric secure data sharing with
loading component of the impression device, respectively, achieves 91% digital twins in smart ecosystems, in: 2021 Third IEEE International Conference on
Trust, Privacy and Security in Intelligent Systems and Applications (TPS-ISA), IEEE,
or 90%, according to the forecasting model given in this study. 2021, December, pp. 70–79.
However, due to its shortcomings, the technology still needs to be [16] B. Karnan, A. Kuppusamy, Graph theory and matrix approach for machinability
improved. The accurate usage of information still has to be enhanced, enhancement of cryogenic treated cobalt bonded tungsten carbide inserts, Journal
homepage 39 (4) (2021) 1372–1382. http://iieta.org/journals/ijht.
the method’s operational performance and repetition rate can still be [17] D.P. Möller, H. Vakilzadian, W. Hou, Intelligent manufacturing with digital twin,
enhanced, and there are data black boxes in the data classifier extraction in: 2021 IEEE International Conference on Electro Information Technology (EIT),
and criteria choice of the method. IEEE, 2021, May, pp. 413–418.

5
L. Gopal et al. Measurement: Sensors 25 (2023) 100661

[18] S.Y. Teng, M. Touš, W.D. Leong, B.S. How, H.L. Lam, V. Máša, Recent advances on [24] M. Monica, P. Sivakumar, S.J. Isac, K. Ranjitha, PMSG based WECS: control
industrial data-driven energy savings: digital twins and infrastructures, Renew. techniques, MPPT methods and control strategies for standalone battery integrated
Sustain. Energy Rev. 135 (2021), 110208. system, in: AIP Conference Proceedings, vol. 2405, AIP Publishing LLC, 2022,
[19] M.A. Al Faruque, D. Muthirayan, S.Y. Yu, P.P. Khargonekar, Cognitive digital twin April, 040013. No. 1.
for manufacturing systems, in: 2021 Design, Automation & Test in Europe [25] S.S. Kamble, A. Gunasekaran, H. Parekh, V. Mani, A. Belhadi, R. Sharma, Digital
Conference & Exhibition (DATE), IEEE, 2021, February, pp. 440–445. twin for sustainable manufacturing supply chains: current trends, future
[20] T.P. Latchoumi, L. Parthiban, Quasi oppositional dragonfly algorithm for load perspectives, and an implementation framework, Technol. Forecast. Soc. Change
balancing in cloud computing environment, Wireless Pers. Commun. 122 (3) 176 (2022), 121448.
(2022) 2639–2656. [26] V.K.S. Maddala, K. Jayarajan, M. Braveen, R. Walia, P. Krishna, S. Ponnusamy,
[21] M.A. Al Faruque, D. Muthirayan, S.Y. Yu, P.P. Khargonekar, Cognitive digital twin K. Kaliyaperumal, Multisensor data and cross-validation technique for merging
for manufacturing systems, in: 2021 Design, Automation & Test in Europe temporal images for the agricultural performance monitoring system, J. Food Qual.
Conference & Exhibition (DATE), IEEE, 2021, February, pp. 440–445. (2022) 2022.
[22] Y. Wu, K. Zhang, Y. Zhang, Digital twin networks: a survey, IEEE Internet Things J. [27] R. Sahal, S.H. Alsamhi, K.N. Brown, D. O’Shea, B. Alouffi, Blockchain-based digital
8 (18) (2021) 13789–13804. twins collaboration for smart pandemic alerting: decentralized COVID-19
[23] Y. Wang, X. Kang, Z. Chen, A Survey of Digital Twin Techniques in Smart pandemic alerting use case, Comput. Intell. Neurosci. 2022 (2022).
Manufacturing and Management of Energy Applications, Green Energy and
Intelligent Transportation, 2022, 100014.