Enabling technologies and tools for digital twin

Abstract

Digital twin is revolutionizing industry. Fired by sensor updates and history data, the sophisticated models can mirror almost every facet of a product, process or service. In the future, everything in the physical world would be replicated in the digital space through digital twin technology. As a cutting-edge technology, digital twin has received a lot of attention. However, digital twin is far from realizing their potential, which is a complex system and long-drawn process. Researchers must model all the different parts of the objects or systems. Varied types of data needed to be collected and merged. Many researchers and participators in engineering are not clear which technologies and tools should be used. 5-dimension digital twin model provides reference guidance for understanding and implementing digital twin. From the perspective of 5-dimension digital twin model, this paper tries to investigate and summarize the frequently-used enabling technologies and tools for digital twin to provide technologies and tools references for the applications of digital twin in the future.

Full text

Contents lists available at ScienceDirect
Journal of Manufacturing Systems
journal homepage: www.elsevier.com/locate/jmansys
Enabling technologies and tools for digital twin
Qinglin Qi
a
, Fei Tao
a,
*, Tianliang Hu
b
, Nabil Anwer
c
, Ang Liu
d
, Yongli Wei
b
, Lihui Wang
e
,
A.Y.C. Nee
f
a
School of Automation Science and Electrical Engineering, Beihang University, Beijing, 100083, China
b
School of Mechanical Engineering, Shandong University, Jinan, 250061, China
c
LURPA, ENS Paris-Saclay, Universite Paris-Sud, Universite Paris-Saclay, 94235, Cachan, France
d
School of Mechanical and Manufacturing Engineering, University of New South Wales, Sydney, 2053, Australia
e
Department of Production Engineering, KTH Royal Institute of Technology, Stockholm, Sweden
f
Department of Mechanical Engineering, National University of Singapore, Singapore, Singapore
ARTICLE INFO
Keywords:
Digital twin
Five-dimension model
Enabling technologies
Enabling tools
ABSTRACT
Digital twin is revolutionizing industry. Fired by sensor updates and history data, the sophisticated models can
mirror almost every facet of a product, process or service. In the future, everything in the physical world would
be replicated in the digital space through digital twin technology. As a cutting-edge technology, digital twin has
received a lot of attention. However, digital twin is far from realizing their potential, which is a complex system
and long-drawn process. Researchers must model all the different parts of the objects or systems. Varied types of
data needed to be collected and merged. Many researchers and participators in engineering are not clear which
technologies and tools should be used. 5-dimension digital twin model provides reference guidance for under-
standing and implementing digital twin. From the perspective of 5-dimension digital twin model, this paper tries
to investigate and summarize the frequently-used enabling technologies and tools for digital twin to provide
technologies and tools references for the applications of digital twin in the future.
1. Introduction
Currently, digitalization has become a consensus, especially digital
twin, precise virtual copies of machines or systems, is revolutionizing
industry [1]. Many companies and fields already use digital twin to spot
problems and increase efficiency [2]. With the advancement of in-
formation technologies, especially the emergence of new generation of
information technologies (New ITs) such as Internet of things (IoT),
cloud computing, big data analytics, and artificial intelligence (AI), the
digitalization process is greatly accelerating. Through the convergence
of the physical and virtual worlds, digitalization is becoming one of the
main drivers of innovation in all sectors [3]. As shown in Fig. 1, the
evolution of digitalization has gone through four progressive stages:
digital enablement, digitalization assistance, digital control and link,
and cyber-physical integration. Digital enablement refers to the process
of converting paper document into digital forms without making any
different-in-kind changes to the process itself [4,5]. Due to the limita-
tion of digitalization tools, initially, only the most essential information
is digitalized for storage, processing, and transfer. With the extensive
applications of CAX technologies (e.g., CAD, CAE, and CAM) in the
1980s, the paradigm of digitalization shifted toward assisting engineers
to work with computers effectively. With the development of Internet
and advanced control technologies in the 1990s, business digitalization
provided new revenue and value-producing opportunities for en-
terprises [5]. After entering the 21 st century, the New IT (e.g., IoT,
cloud computing, big data, and AI) makes it possible to progressively
converge the physical and virtual worlds (i.e., the cyber-physical in-
tegration [6]) toward the digitalization of industrial ecology.
Digital twin (DT) provides a unique means to achieve the cyber-
physical integration, which is a notion embraced by more and more
enterprises [7]. DT means an organic whole of physical asset (or phy-
sical entity) as well as its digitized representation, which mutually
communicate, promote, and co-evolve with each other through bidir-
ectional interactions [8]. Through various digitization technologies, the
entities, behaviors, and relations in the physical world are digitized
holistically to create high-fidelity virtual models [9,10]. Such virtual
models depend on real-world data from the physical world to formulate
their real-time parameters, boundary conditions, and dynamics, leading
to a more representative reflection of the corresponding physical enti-
ties [10,11]. DT is attracting attention from both academia and in-
dustry. Gartner classified DT as one of the top 10 technological trends
with strategic values for 3 years from 2017 to 2019 [12]. In 2018,
https://doi.org/10.1016/j.jmsy.2019.10.001
Received 29 August 2019; Received in revised form 5 October 2019; Accepted 7 October 2019
⁎
Corresponding author.
E-mail address: ftao@buaa.edu.cn (F. Tao).
Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
0278-6125/ © 2019 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.
Please cite this article as: Qinglin Qi, et al., Journal of Manufacturing Systems, https://doi.org/10.1016/j.jmsy.2019.10.001

Lockheed Martin listed DT as one of the six game-changing technologies
for the defense industry [12].
DT has many strategic benefits. In particular, DT provides a unique
way to reflect a physical entity in the digital world with respect to its
shape, position, gesture, status, and motion [13]. Together with the
sensory data acquisition, big data analytics, as well as AI and machine
learning, DT can be used for monitoring, diagnostics, prognostics and
optimization [14,15]. Through the assessment of ongoing states, the
diagnosis of historical problems, and the prediction of future trends, DT
can provide more comprehensive supports for the decision-making of a
wide spectrum of operations. Once integrated with the digital re-
presentation of facilities, environments, and people, DT can be used for
the training of users, operators, maintainers, and service providers [16].
Through DT, it is also possible to digitize expert experience, which can
be documented, transferred, and modified throughout an enterprise to
reduce the knowledge gap. Through simulation tools and virtual reality
tools, DT can deepen the operator's understandings of complex physical
entities and processes.
DT is an effective means to improve enterprises productivity and
efficiency, as well as to reduce cost and time [17]. The current studies
mainly focus on the macro level in terms of framework, process, and
know-what as opposed to the micro level in terms of specific technol-
ogies, tools, and knowhow. Despite a strong desire from small and
medium-size enterprises (SMEs) to incorporate DT into their daily
businesses, most of the SMEs are unfamiliar with the key technologies
and tools of DT. Moreover, DT is a highly complex system that requires
a long-term process to orientate, operate, and optimize. To facilitate
researchers and practitioners to implement DTs, this paper presents a
summary of the key enabling technologies and tools for DT.
The rest of this paper is organized as follows. Section 2presents a
brief overview of DT. Section 3analyzes the ideal potential functions
and practice emphasis. Sections 4and 5present the key enabling
technologies and tools for DT. Finally, Section 6draws conclusions and
outlines future work.
2. A brief overview of digital twin
2.1. History of digital twin
Strictly speaking, DT is not a completely new concept. It is rooted in
some existing technologies [18], such as 3D modeling, system simula-
tion, digital prototyping (including geometric, functional, and beha-
vioral prototyping), etc. The increasing popularity of DT reflects the
inevitable trend that the virtual world and the physical world are be-
coming increasingly linked to each other and integrated as a whole.
From the conception of “a virtual, digital equivalent to a physical
product”by Grieves [11] to the debut of DT thanks to national aero-
nautics and space administration (NASA) and air force research la-
boratory (AFRL) [10], DT represents the breakthrough of numerous
limitations (e.g., data acquisition, digital description, and computer
performance and algorithms, etc.). Then, DT was applied in Industry
4.0 by Siemens in 2016. As more and more researchers devoted to the
research of DT, the number of relevant publications begun to grow
exponentially [19]. Tao et al. [20] proposed the concept of DT shop-
floor in January 2017 and discussed the characteristics, composition
and operation mechanism and key technologies of DT shop-floor, which
provided theoretical support for the application of DT in the manu-
facturing. Later then, to promote the further applications of DT in more
fields, Tao et al. [21] extended the existing 3-dimension DT model and
added two dimensions (DT data and services) to propose a five-di-
mension DT model. Some milestones of DT development are shown in
Fig. 2.
Various definitions of DT appeared, which are reviewed by Tao
et al. [19] and Negri et al. [22]. At present, the two most widely ac-
cepted definitions were given by Grieves and NASA. NASA defined DT
for a space vehicle as “A Digital Twin is an integrated multi-physics,
multiscale, probabilistic simulation of an as-built vehicle or system that uses
the best available physical models, sensor updates, fleet history, etc., to
mirror the life of its corresponding flying twin”[23]. In 2014, Grieves
published the white paper about DT, according to which, the basic DT
model consists of three main parts: (a) physical products in Real Space,
Fig. 1. Evolvement of digitalization paradigm.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
2

(b) virtual products in Virtual Space, as well as (c) the connections of data
and information that tie the virtual and real products together [11]. In es-
sence, DT involves creating a virtual model for a physical entity in the
digital form in order to simulate entity behaviors, monitor the ongoing
status, recognize internal and external complexities, detect abnormal
patterns, reflect system performance, and predict future trend [24].
Currently, the three-dimension DT model originally defined by Grieves
is most applied. However, with the continuous expansion and up-
grading of application requirements, the development and applications
of DT present new trends and demands. For example, the applications
of DT have gradually expanded into the civilian fields in recent years
from the military and aerospace fields in the initial stage. With the
expansion of the application fields, DT is faced with more service de-
mands from different fields, different levels of users, and different
businesses. Meanwhile, the Internet of everything provides conditions
for realizing the cyber-physical interaction and data integration of DT.
Therefore, on the basis of the DT model proposed by Grieves, Tao et al.
[21] proposed the five-dimension DT model to promote the further
applications of DT in more fields.
2.2. Five-dimension digital twin model
The five-dimension digital twin model can be formulated as formula
(1) [25].
=
M
PE VM Ss DD CN(, ,, , )
DT
(1)
where PE are physical entities, VM are virtual models, Ss are services,
DD is DT data, and CN are connections. According to formula (1), the 5-
dimension DT model is shown in Fig. 3.
2.2.1. Physical entities in digital twin
DT is to create the virtual models for physical entities in the digital
way to simulate their behaviors [8]. The physical world is the foun-
dation of DT. The physical world may consist of device or product,
physical system, activities process, even an organization. They imple-
ment activities according to physical laws and deals with uncertain
environments. The physical entities can be divided into three levels
according to function and structure, which are unit level, system level,
and system of system (SoS) level [6].
2.2.2. Virtual models in digital twin
Virtual models ought to be faithful replicas of physical entities,
which reproduce the physical geometries, properties, behaviors, and
rules [26]. The 3-dimension geometric models describe a physical en-
tity in terms of its shape, size, tolerance, and structural relation. Based
on physical properties (e.g. speed, wear and force), physics model re-
flects the physical phenomena of the entities, such as the deformation,
delamination, fracture and corrosion. Behavior model describes the
behaviors (e.g., state transition, performance degradation and co-
ordination) and responding mechanisms of the entities against changes
in the external environment. The rule models equip DT with logical
abilities such as reasoning, judgement, evaluation, and autonomous
decision-making, by following the rules extracted from historical data
or come from domain experts.
2.2.3. Digital twin data
Twin data is a key driver of DT [12]. DT deals with multi-temporal
scale, multi-dimension, multi-source, and heterogeneous data. Some
data is obtained from physical entities, including static attribute data
and dynamic condition data. Some data is generated by virtual models,
which reflects the simulation result. Some data is obtained from ser-
vices, which describes the service invocation and execution. Some data
is knowledge, which is provided by domain experts or extracted from
existing data. Some data is fusion data, which is generated as a result of
fusion of all the aforementioned data.
2.2.4. Services in digital twin
Against the background of product-service integration in all aspects
of modern society, more and more enterprises begin to realize the im-
portance of service [27]. Service is an essential component of DT in
light of the paradigm of Everything-as-a-Service (XaaS). Firstly, DT
provides users with application services concerning simulation, ver-
ification, monitoring, optimization, diagnosis and prognosis, prognostic
and health management (PHM), etc. Secondly, a number of third-party
services are needed in the process of building a functioning DT, such as
data services, knowledge services, algorithms services, etc. Lastly, the
operation of DT requires the continuous support of various platform
services, which can accommodate customized software development,
model building, and service delivery.
2.2.5. Connections in digital twin
Digital representations are connected dynamically with their real
counterpart to enable advanced simulation, operation, and analysis.
Connections between physical entities, virtual models, services, and
data enable information and data exchange. There are 6 connections for
DT, which are connection between physical entities and virtual models
(CN_PV), connection between physical entities and data (CN_PD),
connection between physical entities and services (CN_PS), connection
between virtual models and data (CN_VD), connection between virtual
models and services (CN_VS), connection between services and data
(CN_SD) [25]. These connections enable the four parts to collaborate.
2.3. Application fields of digital twin
Through the integration with mobile Internet, cloud computing, big
data analytics and other technologies, DT is potentially applicable for
many fields where it involves the mapping, fusion, and co-evolution
between the physical and virtual spaces. As shown in Fig. 4, the DT
applications can be found in smart city, construction, healthcare,
agriculture, cargo shipping, drilling platform, automobile, aerospace,
manufacturing, electricity, etc. [2]. As a relatively new technology, the
application of DT was pioneered by the leading enterprises (e.g., GE,
PTC, Siemens, ANSYS, Dassault, etc.) [28,29]. For civil engineering,
Dassault used its 3D Experience Platform to build a “Digital Twin Sin-
gapore”to support urban planning, construction, and service [30]. In-
tellectsoft is exploring the DT applications on construction site to detect
Fig. 2. The milestones of DT development.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
3

potential problems and prevent dangerous operations. In the healthcare
field, Sim&Cure developed patient-based digital twin for treating an-
eurysms [31], and Dassault conducted a “Living Heart Project (LHP)”
toward a human heart DT [32]. According to the whitepaper about DT
by Microsoft, DT has the power to accelerate agricultural business and
support agricultural sustainability [33]. DNV GL established a “virtual
sister ship”(i.e., a vessel DT) to increase reliability, reduce operational
cost, and improve safety throughout the vessel’s lifecycle [34]. A dril-
ling platform DT for the Blue Whale #1 in China enabled the visuali-
zation display, operational monitoring, and design training. Tesla at-
tempted to develop a DT for each electric car to enable the
simultaneous data transfer between car and plant [28]. In the aviation
industry, Airbus, Boeing, AFRL, and NASA used DT to mirror actual
conditions, identify defects, predict potential faults, and solve the
problem of airframe maintenance [29]. LlamaZOO used DT to enable
mine supervisors to monitor their operators’vehicles [35]. Based on the
Predix platform, GE built a digital wind farm, by creating a DT for every
wind turbine, to optimize maintenance strategy, improve reliability,
and increase energy production [36]. Finally, many DT applications can
be found in the manufacturing field. For example, SAP and Dassault
relied on DT to reduce the deviation between functional requirement
and actual performance. Siemens and PTC relied on DT to improve
manufacturing efficiency and quality control. GE, ANSYS, TESLA, and
Microsoft focused on the real-time monitoring, prognostics and health
management, and manufacturing services [12].
3. Functions and practice emphasis of digital twin
3.1. Category and product lifecycle application of digital twin
DT reflects the virtual-reality integration and mapping relations
between the physical and virtual worlds [6]. By recording, simulating,
and predicting the running trajectory of entities and processes in the
physical and virtual worlds, it can achieve the efficient exchange of
information, optimal allocation of resources, analytical reduction of
cost, and prevention of fatal failures [26]. Physical entities exist in
specific scenarios to achieve their own functions and provide targeted
services. Therefore, DT can be divided into entity DT and scenario DT,
as shown in Fig. 5. Based on the 3D geometric models, entity DT
functions to integrate different information such as monitoring in-
formation, sensing information, service information, and behavior in-
formation regarding a physical entity toward ubiquitous tracking of the
entity throughout the whole lifecycle [9,37]. Physical entity would
have a virtual twin that is exactly the same as its status, running tra-
jectory, and behavioral characteristics. As for the DT scenarios, the
physical scenarios are represented in the virtual space with static and
dynamic information. Static information includes spatial layout,
equipment, and geographic location [29,38]. Dynamic information in-
volves environment, energy consumption, equipment operation, dy-
namic process, etc. [29,38]. The activities in the physical scenario can
be simulated by DT.
Some DT applications focused on the entity in terms of functional
modeling, concept verification, behavior simulation, performance op-
timization, status monitoring, diagnosis and prediction, etc. Such ap-
plications can be found in areas such as healthcare, cargo shipping,
drilling platform, automobile, aviation, aerospace, and Internet of
Things. Some other DT applications focused on the scenario (i.e., the
ideal conditions at its best to achieve specific functions). For example,
production scenario is to produce the target product in optimal way.
Construction is the production process of constructing the buildings,
and manufacturing is the production process of turning raw materials or
parts into products. Usage scenario refers to where, how, and when a
product is utilized by end users, which could impact product status and
lifetime. For a shop-floor or factory, it could be a production scenario
for a product or a usage scenario for machine tools. The full potentials
of DT can only be enabled through integration between entity DT and
scenario DT.
According to the product lifecycle, the applications of DT can be
attached to the design, production and use phases, as shown in Fig. 5.
Above all, at the product design stage, DT enables designers to
Fig. 3. Five-dimension digital twin model.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
4

digitalize, visualize, and materialize the intangible concepts of complex
systems (e.g., ship, aircraft, and factory) that have numerous compo-
nents and implicit couplings [38,39]. Besides, the quality of design
schemes can be evaluated, compared, and validated through DT
without building expensive physical prototypes [39]. By virtually run-
ning the design scheme in production and usage scenarios, the manu-
facturability and all the expected functions of the target entity can be
simulated to verify whether the design meets all the requirements. In
this way, the design and production departments can collaboratively
identify design flaws, quality defects, and improve solutions [24]. Guo
et al. [38] and Zhang et al. [40] demonstrated that DT could enable the
designers to simulate the whole factory design process with respect to
factory layout, equipment configuration, material handling, buff;er
capacity, etc. Zhang et al. focused on the development of simulation-
based approach for plant design and production planning [41]. Their
modeling and simulation approaches can be applied to develop digital
twin models of plant. Next, at the production stage, from the perspec-
tive of production management, through the simulation, verification,
and confirmation of process planning and production scheduling, DT
could enable the optimal (re)configuration of on-site resources, equip-
ment, work-in-progress, and workers [42]. From the perspective of
control and execution, DT functions to keep track of everything oc-
curring in the physical world, based on which, to perform operational
forecasting, to optimize control strategy [43,44], and to align actual
process with planning [45]. For example, the DT of construction site
can detect and predict potential issues in the virtual space before they
actually occur in the physical space. The shop-floor DT can optimize
process planning, resources allocation, manufacturing process, and
process control, etc. [26,46]. Besides, Zhang et al. proposed an archi-
tecture of using cloud-based ubiquitous robotic systems for smart
manufacturing of customized product. They also developed an im-
plementation procedure for the development of a cloud-based ubiqui-
tous robotic system [47]. Wang et al. used holon, which possesses a
logical part and a physical part, to mimic the cyber and physical entities
of CPS [48]. Their study is the implementation of digital twin tech-
nology. Lastly, at the service stage, since the same physical entity be-
have differently in various usage scenarios for different purposes, DT is
used to simulate the usage scenarios. DT can lead to new insights for
diagnosis and prognosis of wear [49], remaining life [9], damage lo-
cation [50], etc., so that most of problems can be eliminated in the bud,
reducing costs and downtime [51]. When problems occur, iterative
experiments can be conducted in the digital environment to generate
the best maintenance solution [8]. For example, DT is used to monitor
and simulate the performance of aircraft engines in terms of wear
coefficient and pressure tolerance [14], and DT is also used to drive
PHM for wind turbines [25].
3.2. Practice emphasis for digital twin
In the practical applications of DT, the following key points need
more attention. Firstly, the core of any DT is a high-fidelity virtual
model. To this end, it is critical to fully understand the physical world.
Otherwise, the virtual model cannot correspond effectively with the
physical world. Secondly, while the virtual model is a key part of DT,
Fig. 4. Different application fields of digital twin.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
5

DT modeling is a complex and iterative process. A good virtual model is
characterized by high standardization, modularization, lightweight,
and robustness [2]. The standardization of encoding, interface, and
communication protocol is intended to facilitate information sharing
and integration. Modularization increases the flexibility, scalability,
and reusability through the separation and recombination of individual
models. Lightweight reduces information transfer time and cost. Be-
sides, the robustness of models is indispensable to deal with various
uncertainties. Thirdly, the operations of models and services are all
driven by data. From raw data to knowledge, the data must go through
a series of steps (i.e., data lifecycle) [52]. Each step needs to be re-
structured based on the characteristics of DT. Moreover, DT is unique in
a way that not only it processes data from the physical world, but also it
fuses data generated by the virtual models to make the results more
reliable. Fourthly, the ultimate goal of DT is to provide users with
value-adding services, such as monitoring, simulation, verification,
virtual experiment, optimization, digital education, etc. [15,19]. DT
services are delivered through various mobile apps. Besides, DT can
accommodate some third-party services such as resources service, al-
gorithm service, knowledge service, etc. Therefore, service encapsula-
tion and management are both important parts of DT. Lastly, the phy-
sical world, virtual models, data, and services are not isolated. They
constantly interact with each other through connections among them
towards collective evolution.
4. Enabling technologies for digital twin
According to the 5-dimension model, as shown in Fig. 6, a variety of
enabling technologies are required to support different modules of DT
(i.e., physical entity, virtual model, DT data, smart service, and con-
nection). For the physical entity, the full understanding for the physical
world is a prerequisite for DT. DT involves multidisciplinary knowl-
edge, including dynamics, structural mechanics, acoustics, thermals,
electromagnetism, materials science, hydromechatronics, control
theory, and more. Combined with the knowledge, sensing, and mea-
surement technologies, the physical entities and processes are mapped
to the virtual space to make the models more accurate and closer to the
reality. For the virtual model, various modeling technologies are es-
sential. Visualization technologies are of the essence for real-time
monitoring of physical assets and processes. The accuracy of virtual
models directly affects the effectiveness of DT. Therefore, the models
must be validated by verification, validation & accreditation (VV&A)
technologies and optimized by optimization algorithms. Besides, si-
mulation and retrospective technologies can enable rapid diagnosis of
quality defects and feasibility verification. Since the virtual models
must co-evolve with constant changes in the physical world, model
evolution technologies are needed to drive the model update. During
the operation of DT, a huge volume of data is generated. To extract
useful information from raw data, advanced data analytics and fusion
technologies are necessary. The process involves data collection,
transmission, storage, processing, fusion, and visualization. DT-related
services include application service, resource service, knowledge ser-
vice, and platform service. To deliver these services, it requires appli-
cation software, platform architecture technology, service oriented ar-
chitecture (SoA) technologies, and knowledge technologies. Finally, the
physical entity, virtual model, data, and service of DT are inter-
connected to enable interactions and exchange information. The con-
nection involves Internet technologies, interaction technologies, cyber-
security technologies, interface technologies, communication protocols,
etc.
Fig. 5. Composition and application of digital twin.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
6

4.1. Enabling technologies for cognizing and controlling physical world
The physical world of 5-dimension digital twin model is often
complex. There are intricate attributes and connections (including ex-
plicit and invisible ones) between the various entities in the physical
world. The creation of virtual models is based on the entities in the
physical world, as well as their key internal interaction logic and ex-
ternal relationships. It is very difficult to virtually reproduce such a
complex system. Therefore, the establishment and improvement of DT
is a long process. On the one hand, the virtual model corresponding to
the physical entity is not perfect. As shown in Fig. 7, virtual models
need to evolve to gradually improve the correspondence with physical
entities [26], which requires a full understanding and perception for the
physical world. On the other hand, after the physical entities are digi-
tized, many implicit associations can be discovered, which can be used
to promote the evolution of physical entities to control the physical
world [53].
To create high-fidelity models, it is imperative to cognize the phy-
sical world and perceive data. As shown in Fig. 7, the first step to reflect
the physical world is to measure the parameters, such as size, shape,
structure, tolerance, surface roughness, density, hardness, etc. The ex-
isting measurement technologies include laser measurement, image
recognition measurement, conversion measurement, and micro/nano-
level precision measurement. To synchronize a virtual model with its
real-world counterpart, real-time data (e.g., torque, pressure, dis-
placement speed, acceleration, vibration, voltage, current, temperature,
humidity, etc.) must be collected. To this end, sophisticated digital
twins continuously pull real-time sensor and system data to represent a
near real-time as-is state of physical entities [54]. Structural analysis
models, evolution models, and fault prediction models may vary sig-
nificantly for different industries, which require the expertise. For ex-
ample, smart manufacturing involves knowledge and technologies
about mechanical engineering, material engineering, control and in-
formation processing, etc. Especially, how to automatically control
manufacturing equipment in an adaptive and effective manner is one
major issue [55]. Since methods, technologies, and tools for DT are
forward-looking, it requires effective collaboration between different
industries.
Furthermore, DT serves to improve the performance of physical
entities in the physical world. When the entities in physical world carry
out the intended functions, the energy is controlled by control system to
drive their actuators to accurately complete the specified actions. This
process involves power technologies (e.g., hydraulic power, electric
power, and fuel power), drive systems (e.g., shaftless transmission,
bearing transmission, gear transmission, belt transmission, chain
transmission, and servo drive technologies), process technologies (e.g.,
Fig. 6. Framework of enabling technologies for digital twin.
Fig. 7. Enabling technologies for cognizing and controlling physical world.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
7

process planning, design, management, optimization, and control), and
control technologies (e.g., electrical control, programmable control,
hydraulic control, network control, and interdisciplinary technologies
such as hydromechatronics.
DT applications call for new technologies to better perceive the
physical world. Big data refers to large amounts of multi-source, het-
erogeneous data, which is characterized by 5 Vs, i.e., high volume,
variety, velocity, veracity, and value. Big data analytics provides a new
approach to understand the physical world. Valuable information can
be found from complex phenomena through data analysis, which is
suitable for various industries. As an interdisciplinary technology that
integrates neurobiology, image processing, and pattern recognition,
machine vision can extract information from images for the purposes of
detection, measurement, and control. In addition, the cutting-edge
technologies of various disciplines and industries are all worthy of
further study to make the model more accurate, and the simulation and
prediction results more in line with the actual situation. For example,
for the manufacturing industry, new special processing technologies,
manufacturing processes and equipment technologies, and smart ro-
botic technologies can all help DT to control the smart manufacturing
process. For the construction industry, emerging technologies (e.g., new
materials, construction machinery, and shock absorption technologies)
are transforming the construction industry. For now, it is recommended
to use image recognition and laser measurement technologies to mea-
sure the parameters of the physical world, and use electrical control,
programmable control, embedded control and network control tech-
nologies to control the physical world, as well as use big data analytics
technologies to mine the implicit laws and knowledge.
4.2. Enabling technologies for digital twin modeling
Modeling refers to the process of representing a physical entity in
digital forms that can be processed, analyzed, and managed by com-
puters. Modeling is arguably the cornerstone of DT, which provides
information representation method for product design, analysis, com-
puter numerical control (CNC) machining, quality inspection, produc-
tion management, etc. As shown in Fig. 8, DT-related modeling involves
geometric modeling, physical modeling, behavioral modeling, and rule
modeling.
Geometric model describes a physical entity in terms of its geo-
metric shape, embodiment, and appearance with appropriate data
structures, which are suitable for computer information conversion and
processing. The geometric model includes geometric information (e.g.,
points, lines, surface, and bodies) as well as topological information
(element relations such as intersection, adjacent, tangent, vertical, and
parallel). Geometric modeling includes wireframe modeling, surface
modeling, and solid modeling. Wireframe modeling uses basic lines to
define the ridgeline portion of the target to form a stereoscopic frame.
Surface modeling describes each surface of an entity and then splice all
surfaces to form a holistic model. Solid modeling describes the internal
structure of a three-dimension entity, which includes information such
as vertices, edges, surfaces, and bodys, etc. Besides, to increase the
sense of reality, developers create appearance texture effects (such as
wear, cracks, fingerprints, and stains, etc.) with bitmaps that represent
the surface details of the entity. Texture techniques mainly are texture
blending (with or without transparency) and lightmaps.
Geometric model describes the geometric information of an entity,
but do not describe entity features and constraints. The physical model
adds information such as accuracy information (e.g., dimensional tol-
erance, shape tolerances, position tolerance, and surface roughness),
material information (e.g., material type, performance, heat treatment
requirement, hardness, etc.), and assembly information (e.g., mating
relationship and assembly order). Feature modeling includes interactive
feature definition, automatic feature recognition, and feature-based
design.
Behavioral model describes various behaviors of a physical entity to
fulfill functions, respond to changes, interact with others, adjust in-
ternal operations, maintain health, etc. The simulation of physical be-
haviors is a complex process that involves multiple models, such as
problem model, state model, dynamics model, evaluation model, etc.
These models can be developed based on finite state machines, markov
chains, and ontology-based modeling methods, etc. State modeling in-
cludes state diagram and activity diagram. The former describes the
dynamic behaviors of an entity over its lifecycle (i.e., the representation
of a sequence of states), whereas the latter describes activities required
to complete an operation (i.e., the representation of a sequence of ac-
tivities). Dynamics modeling deals with rigid body motion, elastic
system motion, high-speed rotating body motion, and fluid motion.
Rule model describes the rules extracted from historical data, expert
knowledge, and predefined logic. The rules equip the virtual model
with an ability to reason, judge, evaluate, optimize, and predict. Rule
modeling involves rule extraction, rule description, rule association,
and rule evolution. Rule extraction involves both symbolic methods
(e.g., decision tree and rough set theory) and connectionist methods
(e.g., neural network). Rule description involves methods such as lo-
gical notation, production representation, frame representation, object-
oriented representation, semantic web notation, XML-based re-
presentation, ontology representation, etc. Rule association involves
methods such as category association, diagnostic/inferential associa-
tion, cluster association, behavior association, attribute association, etc.
Rule evolution includes application evolution and periodic evolution.
Application evolution means the process of adjusting and updating the
rules based on feedback obtained from the application process, and
periodic evolution means the process of regularly evaluating the ef-
fectiveness of current rules over a certain period of time (the time varies
depending on the application). The recommendations of key modeling
technologies are solid modeling technologies for the geometric model,
texture technologies for increasing the sense of reality, finite element
analysis technologies for the physical model, finite state machines for
the behavioral model, XML-based representation and ontology re-
presentation for the rule model.
Model VV&A can improve model accuracy and simulation con-
fidence [56]. Model VV&A is intended to analyze whether and to what
extent the correctness, tolerance, availability, and running result meet
the requirement. Model VV&A involves both static methods and dy-
namic methods. Static methods are used to evaluate the static aspect of
modeling and simulation, including grammatical analysis, semantic
analysis, structural analysis, causal maps, control analysis, etc. Dynamic
methods are used to validate the dynamic aspects of modeling and si-
mulation, including black box test, white box test, execution tracking,
regression testing, statistical technique, and graphical comparison.
The current modeling technologies focus on the construction of
geometric and physical models. There is a lack of “multi-spatial scale
models”that can represent the behaviors, features, and rules from
different granularities of different spatial scales. There is a lack of
"multi-time scale models" that can characterize the dynamic process of
physical entities from different time scales. From the system perspec-
tive, it remains a challenge to integrate various models with different
dimensions, different spatial scales, and different time scales. As a re-
sult, the existing virtual models cannot describe physical entities in a
realistic and objective manner. The future modeling technologies are
characterized by multidisciplinary and multifunctional synthesis. The
DT modeling process is an interdisciplinary synthesis process, which
involves mechanical science, hydraulics, aerodynamics, structural me-
chanics, fluid mechanics, acoustics, thermals, electromagnetism, and
control theory. Modeling should be optimized for multi-objective and
full-performance, to reach high accuracy, reliability, and reproduce
both dynamic and static characteristics. Moreover, combined with the
historical usage, maintenance, and upgrade data, various DT models
(e.g., structural analysis model, thermodynamic model, product failure
and life prediction and analysis models, etc.) can be progressively op-
timized through bayesian, machine learning, as well as other data
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
8

mining methods and optimization algorithms.
4.3. Enabling technologies for digital twin data management
Data-driven digital twin can perceive, respond, and adapt to the
changing environment and operational conditions. As illustrated in
Fig. 9, the whole data lifecycle includes data collection, transmission,
storage, processing, fusion and visualization [52].
The data sources include hardware, software, and network [24].
Hardware data includes the static attribute data and dynamic status
data. Barcodes, QR codes, radio frequency identification devices
(RFID), cameras, sensors, and other IoT technologies are widely used
for information identification and real-time perception. Software data
can be collected through software application programming interfaces
(APIs), and open database interfaces. Network data can be collected
from the Internet through web crawlers, search engine, and public APIs.
Data transmission technologies include wire and wireless trans-
missions. Wire transmission technologies include twisted-pair cable
transmission, symmetric cable transmission, coaxial cable transmission,
fiber optic transmission, etc. Wireless transmission includes short-range
and long-distance technologies. The widely used short-range wireless
technologies include Zig-Bee, Bluetooth, Wi-Fi, Ultra-Wideband (UWB),
Fig. 8. Enabling technologies for modeling.
Fig. 9. Enabling technologies for digital twin data management.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
9

and Near Field Communication (NFC) [57]. Long-distance wireless
technologies include GPRS/CDMA, digital radio, spread spectrum mi-
crowave, wireless bridge, satellite communication, etc. Both wire and
wireless transmissions depend on transmission protocols, access
methods, multi-access schemes, channel multiplex modulation and
coding, and multi-user detection technologies.
Data storage is to store the collected data for further processing,
analysis, and management. Data storage is inseparable from database
technologies. However, due to the increasing volume and heterogeneity
of multisource DT data, traditional database technologies are no longer
unfeasible. Big data storage technologies, such as distributed file sto-
rage (DFS), NoSQL database, NewSQL database, and cloud storage, are
drawing growing attention. DFS enables many hosts to access shared
files and directories simultaneously over the network. NoSQL is char-
acterized by the ability to scale horizontally to cope with massive data.
NewSQL denotes new scalable and high-performance databases, which
not only has storage and management capability for massive data, but
also supports ACID and SQL of traditional database. NewSQL imple-
ments replication and failback by using redundant machines.
Data processing means extracting useful information from a large
volume of incomplete, unstructured, noisy, fuzzy, and random raw
data. Firstly, data is carefully preprocessed to remove redundant, irre-
levant, misleading, duplicate, and inconsistent data. The relevant
technologies include data cleaning, data compression, data smoothing,
data reduction, data transformation, etc. Next, the pre-processed data is
analyzed through statistical methods, neural network methods, etc.
Relevant statistical methods include descriptive statistics (e.g., fre-
quency, central tendency, discrete tendency, and distribution analysis),
hypothesis testing (e.g., u-test, t-test, χ2 test, and F-test), correlation
analysis (e.g., linear correlation, partial correlation, and distance ana-
lysis), regression analysis (e.g., linear regression, curve regression,
binary regression, and multiple regression), clustering analysis (e.g.,
partition clustering, hierarchical clustering, density-based clustering,
and grid-based clustering), discriminant analysis (e.g., maximum like-
lihood, distance discriminant, bayesian discriminant, and fisher dis-
criminant), dimension reduction (e.g., principal component analysis,
and factor analysis), time series analysis, etc. Neural network methods
include forward neural network (i.e., neural network based on gradient
algorithm such as BP network, optimal regularization method such as
SVM, radial basis neural network, and extreme learning machine neural
network), feedback network (e.g., Hopfield neural network, Hamming
network, wavelet neural network, bidirectional contact storage net-
work, and Boltzmann machine), and self-organizing neural network
(e.g., self-organizing feature mapping and competitive learning).
Besides, deep learning provides advanced analytics technology for
processing and analysing massive data [58]. Database methods include
multidimensional data analysis and OLAP methods.
Data fusion copes with multisource data through synthesis, filtering,
correlation, and integration. Data fusion includes raw-data-level fusion,
feature-level fusion, and decision-level fusion. Data fusion methods
include random methods and artificial intelligence. Random methods
(e.g., classical reasoning, weighted average method, Kalman filtering,
Bayesian estimation, and Dempster-Shafer evidence reasoning,) are
applicable for all three levels of data fusion. Artificial intelligence
methods (e.g., fuzzy set theory, rough set theory, neural network, wa-
velet theory, and support vector machine) are applicable for the fea-
ture-level and decision-level data fusions.
Data visualization serves to present data analysis results in a
straightforward, intuitive, and interactive manner [52]. Generally
speaking, any method intended to make explicit the underlying prin-
ciples, laws, and logics contained in data by means of graphics is called
data visualization. Data visualization is manifested in various ways such
as histogram, pie chart, line chart, map, bubble chart, tree chart,
dashboards, etc. According to the principle of its visualization, these
methods can be divided into geometry-based technologies, pixel-or-
iented technologies, icon-based technologies, layer-based technologies,
image-based technologies, etc.
As the volume of data continues to increase, the existing data
technologies are bound to advance. For data collection, the future data
acquisition technologies should focus on real-time status data collec-
tion. Therefore, it is necessary to explore smart identification technol-
ogies, advanced sensor technologies, machine vision technologies,
adaptation and access technologies, etc. For data transmission, it is
necessary to explore the applicability of the high-speed, low-latency,
high-performance, and high-secure data transmission protocols (e.g.,
fibre channel protocol and 5 G), and their corresponding devices.
Besides, quantum transmission technology is potentially applicable for
DT as well, including quantum key distribution (QKD), quantum tele-
portation, quantum secure direct communication (QSDC), quantum
secret sharing (QSS). Data storage can be improved by adopting new
storage media (e.g., inductive thin film and magnetic random access
memory) and restructuring storage architecture (e.g., time series, dis-
tributed, and MPP architecture). As algorithms become increasingly
complex, new data processing architectures (e.g., edge computing and
fog computing [59]) can address the issue of massive data processing.
Moreover, new data processing technologies such as graph processing
and domain-oriented data processing technologies should be devel-
oped. The future directions of data fusion include real-time data fusion,
online data and offline data fusion, physical data and simulation data
fusion, structured and unstructured data fusion, big data fusion, object-
based data fusion, similarity fusion, cross-language data fusion, etc.
Currently, it is difficult to visualize explicitly the large-scale and high-
dimensional data. In the future, multiple models should be adopted to
customize the data visualization outcomes through parallel visualiza-
tion technologies, complex data dimensionality reduction visualization
technologies, unstructured data visualization techniques, etc. The re-
commendation of key technologies for data lifecycle management in-
cludes sensors and other IoT technologies for data collection, 5 G
technology for data transmission, NewSQL technology for data storage,
edge-cloud architecture computing technology for data processing, ar-
tificial intelligence technology for data fusion. The data visualization
technologies vary depending on the applications.
4.4. Enabling technologies for digital twin services
DT integrates multiple disciplines to achieve advanced monitoring,
simulation, diagnosis and prognosis. Monitoring requires computer
graphics, image processing, 3-D rendering, graphics engine, virtual-
reality synchronization technologies, etc. Simulation involves structural
simulation, mechanics (e.g., fluid dynamics, solid mechanics, thermo-
dynamics, and kinematics) simulation, electronic circuit simulation,
control simulation, process simulation, virtual test simulation, etc.
Diagnosis and prognosis are based on data analysis, which involves
statistical theory, machine learning, neural network, fuzzy theory, fault
tree, etc.
As shown in Fig. 10, some hardware and software resources and
even knowledge can be encapsulated into services. The lifecycle of re-
source services can be divided into three stages: service generation,
service management, and on-demand use of services [60]. Service
generation technologies include resource perception and assessing (e.g.,
sensors, adapters, and middleware), resource virtualization, and re-
source encapsulation technologies (e.g., SOA, web services, and se-
mantic services), etc. Service management technologies include service
searching, matching, collaboration, comprehensive utility evaluation,
quality of service (QoS), scheduling, fault tolerance technologies, etc.
On-demand use technologies consisted of transaction and business
management technologies, etc. to provide supporting for realizing au-
tomatic matching, transaction process monitoring, comprehensive
evaluation, optimal scheduling of services and users’business. Knowl-
edge services involve the process of knowledge capture, storage,
sharing, and reuse, etc. Common technologies for knowledge capturing
include association rule mining, statistical methods, artificial neural
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
10

network, decision tree, rough set method, case-based reasoning
method, etc. Knowledge storage, sharing, and reuse are implemented in
the form of services.
Resource and knowledge services, application services can be
managed through the industrial IoT platform. Platform provides some
supporting functions such as service publishing, querying, searching,
smart matching and recommendation, online communication, online
contracting, service evaluation, etc. Platform-related technologies in-
clude platform architecture, organization mode, operation and main-
tenance management, security technologies, etc.
In addition, the creation of virtual models is complex and specia-
lized project, so are the data fusion and analysis. For users who do not
have relevant knowledge, it is difficult to build and use the DT.
Therefore, it is imperative that models and data are able to be shared
and used by users. Because service can shield the underlying hetero-
geneity, the DT components can be encapsulated to services to be
managed and used in service platform. As a result, the DT components
that cannot be easily developed in-house can be purchased, shared, and
reused in a convenient “pay-as-you-go”fashion through services [12].
Benefited from comprehensive servitization, the DT can be managed in
a service platform uniformly. Among the enabling technologies for di-
gital twin services, service-oriented architecture is the most important.
In the future, moving target detection is significant for smart
monitoring. Moving target detection, classification, tracking and other
high-level behavioral analysis algorithm improvements are research
route. For multi-state, multi-physics, multi-scale and complex coupling
simulations, they are required to be more precise, more detailed, and
have continuous dynamic optimization capabilities. As DT is complex
system integrating multiple engineering disciplines, the future research
includes multi-domain simulation, joint simulation of multi-simulation
systems coupling. Besides, future simulations also need to enhance
high-performance computing and parallel scalability capabilities. The
generation of massive operational data poses new challenges for diag-
nosis and prognosis. Big data-based diagnosis and prognosis will be the
mainstream research, including algorithm design, feature extraction,
performance improvement, etc. Service transactions involve the service
provider, demander and operator. How to consider the interests of all
participants, and balance their utility is the bottleneck, which still
needs to be solved. Services work together to complete tasks.
Collaborative services are exposed to more uncertainties, which affect
the smooth completion of the task. For knowledge extraction, there is a
lack of resources in natural language processing, especially dictionaries,
which is worth of future research. Efficiency and security are the two
basic elements of the platform. Architecture, algorithms and standards
of security and reliability without compromising performance are the
top priority of research.
4.5. Enabling technologies for connections in digital twin
As shown in Fig. 11, based on the real-time data exchange through
CN_PV, not only the running state of the physical entities is reflected
dynamically in the virtual world, but also the analysis results of the
virtual models are sent back to control the physical entities. Through
CN_PD, DT is used to manage the entire product lifecycle, which laid
data foundation for analysis, prediction, quality tracing, and product
planning. Through CN_PS, services (e.g., monitoring, diagnostics, and
prognosis) are linked to physical entities to receive data and feed the
service outcome back. In the connections of physical entities with
models, data and services, identification, sensing and tracking of phy-
sical entities are crucial. Therefore, RFID, sensor, wireless sensor net-
work, and other IoT technologies are necessary. Data exchange requires
communication technology [61], unified communication interfaces,
and protocol technologies, including protocol parsing and conversion,
interface compatibility, and common gateway interface, etc. Since
human interacts with DT in both physical and virtual worlds, human-
computer interaction technologies (e.g., VR, AR, MR) should be in-
corporated, as well as human–robot interaction and collaboration [62].
Given many different models, CN_VD needs communication, interfaces,
protocols, and standard technologies to ensure smooth data interaction
between virtual models and data. Similarly, the connections between
services and virtual models (CN_VS) as well as data (CN_SD) also re-
quire communication interface, protocol, standard technologies, and
collaboration technologies. Finally, security technologies (e.g., device
security, network security, information security) must be incorporated
to protect the security of DTs. In the connections in digital twin, the
communication interfaces and protocol technologies, human-computer
Fig. 10. Enabling technologies for digital twin services.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
11

interaction technologies, as well as security technologies should be pay
more attention.
The connection part serves to ensure real-time interaction among
different parts of DTs. At present, the inconsistency of interfaces, pro-
tocols, and standards is the bottleneck of DT connection. It is necessary
to investigate the general interconnection theories, standards, and de-
vices with heterogeneous multi-source elements. As data traffic con-
tinues to grow exponentially, research hotspots such as multi-dimen-
sional multiplexing (e.g., time division, wavelength division, frequency
division, code division, and modularization) and coherent technologies,
can provide more bandwidth and lower latency access services. Facing
the massive incoming data, a promising solution is to build ultra-large-
capacity routers with tens of millions of small routing entries to provide
end-to-end communications. It is necessary to develop new network
architectures in order to achieve the flexible control of network traffic
and make the network (as pipelines) more intelligent. Given the in-
crease of communication bandwidths and energy consumption, it is
necessary to develop new strategies and approaches toward green
communication.
5. Tools for digital twin
As shown in Fig. 12, based on the 5-dimension digital twin model,
functional requirements and enabling technologies of digital twin, some
tools are prescribed, including tools for cognizing and controlling
physical world, tools for digital twin modeling, tools for digital twin
data management, tools for digital twin services applications, and tools
for connections in digital twin.
5.1. Tools for cognizing and controlling physical world
The tools for physical part of DT can be divided into tools for cog-
nizing physical world and tools for controlling physical world.
Cognizing different aspects of the physical world is the foundation of
digitalization. IoT is one of the drivers of digital twin. When the phy-
sical entities are hooked up to data sensing and gathering systems, di-
gital twin turn the data into insights and ultimately into optimized
processes and business outcomes. For example, Ali Cloud IoT provides
secure and reliable device sensing capacity, enabling fast access to
multi-protocol, multi-platform, multi-regional devices. Besides, the
virtual models run in parallel to the physical assets. Driven by sensors
data, digital twin flags operational behavior that deviates from simu-
lated behavior. For example, a petroleum company may stream sensor
data from offshore oil rigs that operate continuously. IoTSyS is an IoT
middleware, which provides a communication protocol stack for the
communication between smart devices. IoTSyS supports multiple
standards and protocols, including IPv6, oBIX, 6LoWPAN, and efficient
XML exchange formats. Moreover, most tools for cognizing the physical
world are vision-related. For example, in an uncharted workshop en-
vironment, AGV (automated guided vehicle) cars can use LIDAR (light
detection and ranging), depth camera, GPS (global positioning system),
and maps established through the ROS (robot operating system) soft-
ware architecture, to optimize the path [63]. Similar software tools are
shown in Fig. 13.
A tool for controlling the physical world can make physical entities
run more efficiently and securely based on feedback information, which
is analysis and processing of perceived physical entity state information
Fig. 11. Enabling technology for connections.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
12

in virtual world. Digital twin is to adjust the physical world mainly
through controlling the operations with feedback. Therefore, the tools
for changing the physical world mostly are control-related. For ex-
ample, TwinCAT software system can turn almost any compatible PC
into a real-time controller with a multi-PLC system, NC axis control,
programming environment and operating station [64]. SAP provides
vehicle maintenance and remote diagnosis services for Trenitalia (i.e.,
the primary train operator in Italy) through real-time data analysis.
Besides, it provides an optimal operation plan for the health state and
train running state through the dispatching system [65]. Similar soft-
ware tools are shown in Fig. 13.
5.2. Tools for digital twin modeling
ANSYS Twin Builder containing extensive application-specific li-
braries and features third-party tool integration is an appropriate soft-
ware tool for digital twin modeling, which allows for multiple modeling
domains and languages. Twin Builder can enable engineers to quickly
build, validate and deploy the digital models of physical assets. Twin
Builder's built-in libraries provide rich components to create the desired
system dynamics models at an appropriate level of detail, which include
models from multiple physical domains and multiple levels of fidelity.
Besides, Twin Builder couples with ANSYS' physics-based simulation
technology to bring the detail of 3D into the systems context. Moreover,
Fig. 12. Framework of tools for digital twin.
Fig. 13. Tools for cognizing and controlling physical world.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
13

Twin Builder readily integrates embedded control software and HMI
design to support testing the performance of embedded controls with
models of the physical system. In addition, Siemens NX software, a
flexible and powerful tool, can enable companies to realize the value of
the digital twin. NX software can deliver the next generation of design,
simulation, and manufacturing solutions through integrated toolset, to
support every aspect of product development, from concept design
through engineering and manufacturing.
Moreover, virtual models reproduce the physical geometries, prop-
erties, behaviors, and rules. The models include geometry models,
physical models, behavior models and rule models. Therefore, the tools
for DT modeling include geometry modeling tools, physical modeling
tools, behavior modeling tools, rule modeling tools.
The geometric modeling tools serve to describe the shape, size,
position and assembly relationship of entities, based on which, to per-
form structural analysis and production planning. For example, a per-
formance test device for the DT model of CNC machine tool is estab-
lished in SolidWorks. Besides, 3D Max is software for 3D modeling,
animation, rendering and visualization. 3D Max is used to shape and
define detailed environments, objects (person, place, or thing) and
widely used in advertising, film and television, industrial design, ar-
chitectural design, 3D animation, multimedia production, games, and
other engineering fields. The common geometric modeling tools are
shown in Fig. 14.
The physical modeling tools are used to build physical model by
endowing physical characteristics of physical entities into geometric
models, then physical state of physical entities can be analyzed through
this physical model. For example, through the finite element analysis
(FEA) software by ANSYS, sensor data can be used to define real-time
boundary conditions for the geometric models and integrate wear
coefficient or performance degradation into the models [66]. Besides,
Simulink can be used to create physics-based model using multi-domain
modeling tools. Physics-based modeling with Simulink involves mul-
tiple models, including mechanical, hydraulic, and electrical compo-
nents. Similar software tools for physical modeling are shown in
Fig. 14.
The behavior modeling tools are used to establish a model that re-
sponds to external drivers and disturbance factors and improves the
simulation service performance of DT. For example, based on the soft
PLC platform CoDeSys, the motion control system of CNC machine tool
can be designed. The motion control system can interact information
with the multi-domain model of three-axis CNC machine tool estab-
lished in the software platform MWorks, through the socket commu-
nication. In this way, it can realize the motion control of single-axis and
three-axis interpolation of CNC machine tool. Besides, the multi-do-
main model can respond to the external drive. Similar software tools
are shown in Fig. 14.
The rule modeling tools can improve the service performance by
modeling the logics, laws, and rules of physical behaviors. For example,
the machine learning ability by PTC’s Thingworx upon the HP EL20
edge computing system can monitor sensors to automatically learn the
normal state of the pump while it is running. Based on the learned rules,
DT can recognize abnormal operating conditions, detect abnormal
patterns, and predict future trends [67]. Similar software tools are
shown in Fig. 14.
Fig. 14. Tools for digital twin modeling.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
14

5.3. Tools for digital twin data management
Data is the carrier of information and the key driver of DT. As shown
in Fig. 15, the tools for DT data management include data collection
tools, data transmission tools, data storage tools, data processing tools,
data fusion tools, and data visualization tools.
Data collection tools can obtain complete, stable and effective data
through reasonable sensor placement. For example, DHDAS signal ac-
quisition and analysis system is a set of signal analysis and processing
software. The software can be used with a variety of models to complete
the real-time acquisition and analysis of different signals. The software
also has signal analysis processing capabilities. Similar software tools
are shown in Fig. 15.
The purpose of data transmission is to realize real-time data trans-
mission while ensuring that data information is not missing or da-
maged, and to maintain the authenticity of data to the greatest extent.
With the advent of the big data era, traditional FTP solutions are in-
adequate to meet the data transmission needs in terms of speed or re-
liability. A representing tool is Aspera that is known for the ability to
transmit large size file, over long transmission distance, and under poor
network condition. Aspera uses the existing WAN infrastructure to
transmit data in a much faster speed than FTP and HTTP. Without
changing the original network architecture, it supports the Web inter-
face, client, command line and API for transmission, as well as PC,
mobile devices, MAC, and Linux devices. Alternative data transmission
tools are shown in Fig. 15.
Data storage is the guarantee of subsequent operations, which rea-
lizes the classification and preservation of data, and responds to data
calling in real time through efficient read-write mechanism. The data
storage technology has experienced a rapid development in recent year.
A representing example is HBase based on the Hadoop platform. HBase
is a highly reliable, high performance, column-oriented, scalable, real-
time read-write distributed database. It can support the storage of both
semi-structured and unstructured data, as well as independent in-
dexing, high availability, and large instantaneous writes. The alter-
native data storage systems are shown in Fig. 15.
Data processing eliminates interference and contradictory in-
formation, making data available for efficient use. For example, Spark is
an open source cluster computing software, which has the real-time
data processing ability. Spark supports applications written in multiple
languages such as Java, Scala and Python, greatly reducing the user's
threshold. Spark also supports SQL and Hive SQL for data query. Similar
data processing software is shown in Fig. 15.
Data fusion integrates, filters, correlates and synthesizes the
processed data to aid in judgment, planning, verification and diagnosis.
For example, Spyder is a commonly used data fusion tool that supports
Python programming. Another data fusion software, Pycharm, can
improve productivity in debugging, syntax highlighting, project man-
agement, code jumping, smart prompting, auto-completion, unit
testing, and version control. Other tools that are equipped with the data
fusion ability are shown in Fig. 15.
Data visualization provides neat, intuitive and clear data informa-
tion to personnel for real-time monitoring and rapid capture of target
information. For example, the open-source software Echarts can run
smoothly on PCs and mobile devices, and is compatible with most
current browsers. Echarts provides intuitive, vivid, and customized data
visualizations for huge-volume and dynamic data. It can accommodate
a variety of data formats without extra conversions. Similar tools are
shown in Fig. 15.
5.4. Tools for digital twin service applications
The tools for digital twin service applications can be classified into
platform service tools, simulation service tools, optimization service
tools, diagnostic and prognosis service tools, as shown in Fig. 16.
Service platform tools integrate emerging technologies such as the
Internet of things, big data, artificial intelligence. For example, the
Thingworx platform can connect the DT model to the products in op-
eration, to display sensor data, and analyze results through web ap-
plications. ThingWorx platform could provide industrial protocol con-
version, data acquisition, device management, big data analysis and
other services. HIROTEC, a premier automation manufacturing equip-
ment and parts supplier, realized the connection between CNC machine
operation data and ERP system data based on the ThingWorx platform,
effectively reducing equipment downtime. Siemens launched the
MindSphere platform. The platform can transmit industrial field devices
data collected by sensors, controllers and various information systems
to the cloud in real time through secure channels, and provide big data
analysis and mining, industrial APPs, and value-added services for en-
terprises. Similar tools are shown in Fig. 16.
The application service tools include monitoring service tools, op-
timization service tools, diagnostic and prognosis service tools, etc. The
diagnostic and prognosis service tools can provide intelligent predictive
maintenance strategy for equipment and reduce equipment downtime,
etc., by analyzing and processing the twin data. For example, the
ANSYS simulation platform helps customers design the IIoT-connected
assets themselves and analyze the operational data created by these
smart devices, versus design data to enable troubleshooting and
Fig. 15. Tools for digital twin data management.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
15

predictive maintenance. In addition, integrated with data-driven
methods (machine learning, deep learning, neural networks, and
system identification, etc.), MATLAB can be used to determine re-
maining useful life to inform operations on the most opportune time to
service or replace equipment. For example, Baker Hughes, a large ser-
vice company providing products and services to the oil development
and processing industry, has developed predictive maintenance alarm
system based on MATLAB. Similar diagnostic and prognosis tools are
shown in Fig. 16.
Using twin data like sensor data, energy costs, or performance fac-
tors, the optimization service tools are triggered to run hundreds or
thousands of what-if simulations to evaluate readiness or necessary
adjustments to current system set-points. This enables system opera-
tions to be optimized or controlled during operation to mitigate risk,
reduce cost and energy consumption, and increase system efficiencies.
For example, the Plant Simulation software by Siemens can optimize
the production line scheduling and factory layout [68]. And in digital
twin electric grid, Simulink receives measured data from the grid, then
runs thousands of simulation scenarios to determine if the energy
reserve is sufficient and whether grid controllers need adjustment. Si-
milar tools are shown in Fig. 16.
Advanced simulation tools not only can perform diagnostics and
determine the best benefits of maintenance, but also capture informa-
tion to refine the next-generation design. For example, if lack of ap-
propriate FEM simulation analysis, in the design of CNC machine tool,
the machine will fail in vibration. On the other hand, if extra material is
added to increase the strength and to reduce vibration then the cost of
machine would escalate due to the over-designing of the CNC machine
tool. However, carry out corresponding structure simulation analysis in
the finite element software ANSYS and then auxiliary to the appropriate
evaluation function, which will take into account the performance and
cost, and meet the lean design requirements of CNC machine tool [69].
Similar simulation tools are shown in Fig. 16.
5.5. Tools for connections in digital twin
The tools for DT connections are used to connect the physical and
virtual worlds, as well as to connect different parts of DT. The core of
Fig. 16. Tools for digital twin services applications.
Fig. 17. Tools for connections in digital twin.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
16

any DT is to map between physical and virtual worlds and break the
boundaries between physical and virtual realities. For example, PTC
Thingworx can act as a gateway between sensors and digital models to
connect various smart devices to the IoT ecosystem [67]. MindSphere is
a cloud-based, open IoT operating system from Siemens that connects
products, plants, systems, and machines. MindSphere uses advanced
analytics to enable the wealth of data generated by the IoT. Jasper
Control Center from Cisco Jasper can better manage connected devices
using NB-IoT technology. Jasper Control Center continuous monitor
network conditions, device behavior, and IoT service status to ensure
high service reliability through real-time diagnostics and proactive
monitoring of connection status. The connections within DT mean the
communication, interaction, and exchange of information among phy-
sical entity, data center, service, and virtual model. These information
connections are necessary to help develop problem diagnostics and
troubleshooting, determine the ideal maintenance plan based on the
characteristics of each physical asset, and optimize the performance of
physical assets, etc. For example, the Azure IoT Hub by Microsoft en-
abled Rolls-Royce to build engine models and perform data analysis
based on machine learning. In this way, it can detect anomalies of
about-to-fail components and prescribe suitable solutions [70]. Similar
tools are shown in Fig. 17.
There are a number of comprehensive tools that play multiple roles
in DT applications, such as FEA software ANSYS not only modeling, but
also providing simulation services, troubleshooting services, and so on.
Similar comprehensive tools include Predix, Siemens’MindSphere,
ANSYS, Dassault's 3D Experience, Foxconn's Beacon, PTC’s Thingworx,
etc., as shown in Table 1.
Table 1
Comprehensive tools and their roles in different aspects of DT (√denotes it can be used in this part).
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
17

Implementing digital twin is a complex system and long-drawn
process, which requires multiple technologies and tools to work to-
gether. For example, reproducing a wind turbine requires monitoring of
various data (e.g., vibration signals, acoustic signals, electrical signals,
etc.) of the gearbox, generator, blades, bearings, shafts, tower and
power converter, as well as environment conditions (e.g. wind speed,
wind direction, temperature, humidity and pressure). In addition, di-
gital twin includes virtual representation of physical asset. Many
models need to be built to reproduce a wind turbine, including geo-
metric models, functional models, behavioral models, rules models, fi-
nite element analysis models, fault diagnosis models, life prediction
models, and so on. All of the above need one of the enabling technol-
ogies and tools. For example, the data collection of various signals from
wind turbine need sensor technologies. The data transmission, storage,
processing and fusion may use 5 G, NewSQL, edge-cloud architecture
and artificial intelligence technologies, etc. And the geometric models
can be built through tools such as SolidWorks, UG, AutoCAD, CATIA,
etc. Finite element analysis models can run in ANSYS, MARC, ADINA,
etc. Moreover, Dymola, MWorks, SimulationX and others can support
system modeling and simulation. From the above, digital twin involves
a wide range of technologies and tools that are invented or developed
by different companies. There are different protocols and standards,
about these technologies and tools. To enable these technologies and
tools to work together, data and models should be standardized and
delivered in common formats, protocols and standards. Through
common formats, protocols and standards, these technologies and tools
work together for a particular objective.
6. Conclusion
DT represents an advancement of digitalization. It is increasingly
applied in more and more areas, such as smart manufacturing, building
management, smart city, healthcare, oil & gas, and many more. Because
DT is a complex system integrating multiple engineering disciplines,
many companies and researchers are unfamiliar with the key technol-
ogies and tools of DT. The 5-dimension digital twin model has good
practicability and scalability, and can provide a common reference
model support for applications of digital twin in different fields.
Combined with 5-dimension digital model, this paper investigated and
summarized the enabling technologies and tools for DT, which could
provide guidance for DT practices. However, due to different formats,
protocols and standards, current tools may not be integrated and used
simultaneously for a particular objective. Therefore, in the future, the
universal design and development platforms and tools for digital twin
are required to be developed. Besides, infrastructure that is suitable for
industrial practices and has high reliability, is required to meet the
requirements of digital twin. Besides, the practice of DT is closely re-
lated to specific objects. For example, DT city and DT shop-floor are
quite different in terms of model size, operational rules, data manage-
ment, etc. Therefore, this paper provides general directions for enabling
technologies and some examples of tools. The technology research
about DT and the selection of tools require the participants in academia
and industry to judge and decide according to specificfields and ob-
jects.
Declaration of Competing Interest
Qinglin Qi, Fei Tao, Tianliang Hu, Nabil Anwer, Ang Liu, Yongli
Wei, Lihui Wang, and A.Y.C. Nee declare that they have no conflict of
interest or financial conflicts to disclose.
Acknowledgments
This work is financially supported in part by National Natural
Science Foundation of China (NSFC) under Grant 51875030 and
51705014, and in part by the National Key Research and Development
Program of China under Grant 2016YFB1101700.
References
[1] Tao F, Qi Q. Make more digital twins. Nature 2019;573:490–1.
[2] Tao F, Zhang M, Nee AYC. Digital twin driven smart manufacturing. Elsevier-
Academic Press; 2019. ISBN: 9780128176306.
[3] Botkina D, Hedlind M, Olsson B, Henser J, Lundholm T. Digital twin of a cutting
tool. Procedia CIRP 2018;72:215–8.
[4] Parviainen P, Tihinen M, Kääriäinen J, Teppola S. Tackling the digitalization
challenge: how to benefit from digitalization in practice. Int J Inf Syst Project
Manage 2017;5(1):63–77.
[5] Legner C, Eymann T, Hess T, Matt C, Böhmann T, Drews P, et al. Digitalization:
opportunity and challenge for the business and information systems engineering
community. Bus Inf Syst Eng 2017;59(4):301–8.
[6] Tao F, Qi Q, Wang L, Nee A. Digital twins and cyber–physical systems towards smart
manufacturing and industry 4.0: correlation and comparison. Engineering
2019;5:653–61.
[7] Cheng Y, Zhang Y, Ji P, et al. Cyber-physical integration for moving digital factories
forward towards smart manufacturing: a survey. Int J Adv Manuf Technol
2018;97(1–4):1209–21.
[8] Tao F, Cheng J, Qi Q, Zhang M, Zhang H, Sui F. Digital twin-driven product design,
manufacturing and service with big data. Int J Adv Manuf Technol
2018;94(9–12):3563–76.
[9] Tuegel EJ, Ingraffea AR, Eason TG, Spottswood SM. Reengineering aircraft struc-
tural life prediction using a digital twin. Int J Aerosp Eng 2011:1–14.
[10] Glaessgen E, Stargel D. The digital twin paradigm for future NASA and US air force
vehicles. Proceedings of the 53rd AIAA/ASME/ASCE/AHS/ASC Structures,
Structural Dynamics and Materials Conference 20th AIAA/ASME/AHS Adaptive
Structures Conference 14th AIAA; 2012 Apr 23-26 2012. p. 1818.
[11] Grieves M. Digital twin: manufacturing excellence through virtual factory replica-
tion. White paper. Melbourne, FL: Florida Institute of Technology; 2014.
[12] Qi Q, Tao F, Zuo Y, Zhao D. Digital twin service towards smart manufacturing.
Procedia CIRP 2018;72:237–42.
[13] Söderberg R, Wärmefjord K, Carlson JS, Lindkvist L. Toward a digital twin for real-
time geometry assurance in individualized production. CIRP Ann-Manuf Technol
2017;66(1):137–40.
[14] Zaccaria V, Stenfelt M, Aslanidou I, Kyprianidis KG. Fleet monitoring and diag-
nostics framework based on digital twin of aero-engines. Proceedings of ASME
Turbo Expo 2018: Turbomachinery Technical Conference and Exposition; 2018,
June 11-15 2018.
[15] Cai Y, Starly B, Cohen P, Lee YS. Sensor data and information fusion to construct
digital-twins virtual machine tools for cyber-physical manufacturing. Procedia
Manuf 2017;10:1031–42.
[16] Goossens P. Industry 4.0 and the power of the digital twin Available:https://
www.maplesoft.com/ns/manufacturing/industry-4-0-power-of-the-digital-
twin.aspxAccessed: 2019-09-24 2019.
[17] Rosen R, Von Wichert G, Lo G, et al. About the importance of autonomy and digital
twins for the future of manufacturing. IFAC-PapersOnLine 2015;48(3):567–72.
[18] Boschert S, Rosen R. Digital twin—the simulation aspect. Mechatronic futures.
Cham: Springer; 2016. p. 59–74.
[19] Tao F, Zhang H, Liu A, Nee AYC. Digital twin in industry: state-of-the-art. IEEE
Trans Industr Inform 2018;15(4):2405–15.
[20] Tao F, Zhang M, Cheng J, Qi Q. Digital twin workshop: a new paradigm for future
workshop. Comput Integr Manuf Syst 2017;23(1):1–9.
[21] Tao F, Liu W, Zhang M, et al. Five-dimension digital twin model and its ten ap-
plications. Comput Integr Manuf Syst 2019;25(1):1–18.
[22] Negri E, Fumagalli L, Macchi M. A review of the roles of digital twin in CPS-based
production systems. Procedia Manuf 2017;11:939–48.
[23] Shafto M, Conroy M, Doyle R, Glaessgen E, Kemp C, LeMoigne J, et al. Modeling,
simulation, information technology & processing roadmap. National
Aeronautics and Space Administration; 2012 Available: https://www.nasa.gov/
sites/default/files/501321main_TA11-ID_rev4_NRC-wTASR.pdf, Accessed: 2019-
09-24.
[24] Qi Q, Tao F. Digital twin and big data towards smart manufacturing and industry
4.0: 360 degree comparison. IEEE Access 2018;6:3585–93.
[25] Tao F, Zhang M, Liu Y, Nee AYC. Digital twin driven prognostics and health man-
agement for complex equipment. CIRP Ann-Manuf Technol 2018;67(1):169–72.
[26] Tao F, Zhang M. Digital twin shop-floor: a new shop-floor paradigm towards smart
manufacturing. IEEE Access 2017;5:20418–27.
[27] Tao F, Qi Q. New IT driven service-oriented smart manufacturing: framework and
characteristics. IEEE Trans Syst Man Cybern Syst 2019;49(1):81–91.
[28] Schleich B, Anwer N, Mathieu L, Wartzack S. Shaping the digital twin for design and
production engineering. CIRP Ann-Manuf Technol 2017;66(1):141–4.
[29] Zheng Y, Yang S, Cheng H. An application framework of digital twin and its case
study. J Ambient Intell Humaniz Comput 2019;10(3):1141–53.
[30] Dassault Systèmes. Meet Virtual Singapore, the city’s 3D digital twin
Available:https://govinsider.asia/digital-gov/meet-virtual-singapore-citys-3d-
digital-twin/Accessed: 2019-09-24 2019.
[31] Marchal T. Will Digital Twins Improve Healthcare for Their Real-World
Counterparts? Available: https://www.ansys.com/blog/will-digital-twins-improve-
healthcare-real-world-counterparts/, Accessed: 2019-09-24 2019.
[32] The Living Heart Project. 2019 Available: https://www.3ds.com/products-services/
simulia/solutions/life-sciences/the-living-heart-project/, Accessed: 2019-09-24.
[33] The promise of a digital twin strategy. 2019 Available: https://info.microsoft.com/
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
18

rs/157-GQE-382/images/Microsoft%27s%20Digital%20Twin%20%27How-To
%27%20Whitepaper.pdf, Accessed: 2019-09-24.
[34] DNVGL. Digital Twin at Work Available: https://www.dnvgl.com/feature/digital-
twins.html, Accessed: 2019-09-24 2019.
[35] McCannel DC. What is a digital twin? Available: https://www.llamazoo.com/what-
is-a-digital-twin/, Accessed: 2019-09-24 2019.
[36] GE Renewable Energy. Digital Wind Farm-The Next Evolution of Wind Energy
Available: https://www.ge.com/content/dam/gepower-renewables/global/en_US/
downloads/brochures/digital-wind-farm-solutions-gea31821b-r2.pdf, Accessed:
2019-09-24 2019.
[37] Grieves M, Vickers J. Digital twin: mitigating unpredictable, undesirable emergent
behavior in complex systems. Transdisciplinary perspectives on complex systems.
Cham: Springer; 2017. p. 85–113.
[38] Guo J, Zhao N, Sun L, Zhang S. Modular based flexible digital twin for factory
design. J Ambient Intell Humaniz Comput 2019;10(3):1189–200.
[39] Tao F, Sui F, Liu A, Qi Q, Zhang M, Song B, et al. Digital twin-driven product design
frameworkJ]. Int J Prod Res 2019;57(12):3935–53.
[40] Zhang H, Liu Q, Chen X, Zhang D, Leng J. A digital twin-based approach for de-
signing and multi-objective optimization of hollow glass production line. IEEE
Access 2017;5:26901–11.
[41] Zhang Z, Wang X, Wang X, Cui F, Cheng H. A simulation-based approach for plant
layout design and production planning. J Ambient Intell Humaniz Comput
2019;10(3):1217–30.
[42] Zhang H, Zhang G, Yan Q. Dynamic resource allocation optimization for digital
twin-driven smart shopfloor. Proceedings of 2018 IEEE 15th International
Conference on Networking. 2018. p. 1–5.
[43] Sun H, Li C, Fang X, Gu H. Optimized throughput improvement of assembly flow
line with digital twin online analytics. Proceedings of 2017 IEEE International
Conference on Robotics and Biomimetics (ROBIO). 2017. p. 1833–7.
[44] Senington R, Baumeister F, Ng A, Oscarsson J. A linked data approach for the
connection of manufacturing processes with production simulation models.
Procedia CIRP 2018;70:440–5.
[45] Vachálek J, Bartalský L, Rovný O, Šišmišová D, MorháčM, Lokšík M. The digital
twin of an industrial production line within the industry 4.0 concept. Proceedings of
2017 21st International Conference on Process Control (PC) 2017:258–62.
[46] Coronado PDU, Lynn R, Louhichi W, Parto M, Wescoat E, Kurfess T. Part data in-
tegration in the shop floor digital twin: mobile and cloud technologies to enable a
manufacturing execution system. J Manuf Syst 2018;48:25–33.
[47] Zhang Z, Wang X, Zhu X, et al. Cloud manufacturing paradigm with ubiquitous
robotic system for product customization. Robot Comput Integr Manuf
2019;60:12–22.
[48] Wang L, Haghighi A. Combined strength of holons, agents and function blocks in
cyber-physical systems. J Manuf Syst 2016;40:25–34.
[49] Zakrajsek AJ, Mall S. The development and use of a digital twin model for tire
touchdown health monitoring. Proceedings of 58th AIAA/ASCE/AHS/ASC
Structures, Structural Dynamics, and Materials Conference. 2017. p. 0863.
[50] Seshadri BR, Krishnamurthy T. Structural health management of damaged aircraft
structures using digital twin concept. Proceedings of 25th AIAA/AHS Adaptive
Structures Conference 1675. 2017.
[51] Liu Z, Meyendorf N, Mrad N. The role of data fusion in predictive maintenance
using digital twin. Proceedings of AIP Conference Proceedings. 1949. 2018. (1).
[52] Tao F, Qi Q, Liu A, Kusiak A. Data-driven smart manufacturing. J Manuf Syst
2018;48:157–69.
[53] Kritzinger W, Karner M, Traar G, et al. Digital twin in manufacturing: a categorical
literature review and classification. IFAC-PapersOnLine 2018;51(11):1016–22.
[54] Angrish A, Starly B, Lee YS, et al. A flexible data schema and system architecture for
the virtualization of manufacturing machines (VMM). J Manuf Syst
2017;45:236–47.
[55] Adamson G, Wang L, Moore P. Feature-based control and information framework
for adaptive and distributed manufacturing in cyber physical systems. J Manuf Syst
2017;43:305–15.
[56] Robinson S, Brooks RJ. Independent verification and validation of an industrial
simulation model. Simulation 2010;86(7):405–16.
[57] Lei S. Design of data acquisition system based on zigbee for wireless sensor net-
works. Proceedings of MATEC Web of Conferences 246. 2018. p. 3036.
[58] Wang J, Ma Y, Zhang L, et al. Deep learning for smart manufacturing: methods and
applications. J Manuf Syst 2018;48:144–56.
[59] Wu D, Liu S, Zhang L, et al. A fog computing-based framework for process mon-
itoring and prognosis in cyber-manufacturing. J Manuf Syst 2017;43:25–34.
[60] Tao F, Zhang L, Liu Y, et al. Manufacturing service management in cloud manu-
facturing: overview and future research directions. J Manuf Sci Eng
2015;137(4):040912.
[61] Schroeder GN, Steinmetz C, Pereira CE, et al. Digital twin data modeling with au-
tomationml and a communication methodology for data exchange. IFAC-
PapersOnLine 2016;49(30):12–7.
[62] Yao B, Zhou Z, Wang L, et al. A function block based cyber-physical production
system for physical human–robot interaction. J Manuf Syst 2018;48:12–23.
[63] Sprunk C, Lau B, PfaffP, Burgard W. An accurate and efficient navigation system for
omnidirectional robots in industrial environments. Auton Robots
2017;41(2):473–93.
[64] Twin CAT –PLC and motion control on the PC. 2019 Available: http://www.
beckhoff.com/twincat/, Accessed: 2019-09-24.
[65] An update on SAP and trenitalia’s IoT-enabled dynamic maintenance approach.
2019 Available: https://www.arcweb.com/blog/update-sap-and-trenitalias-iot-en-
abled-dynamic-maintenance-approach, Accessed: 2019-09-24.
[66] Magargle R, Johnson L, Mandloi P, Davoudabadi P, Kesarkar O, Krishnaswamy S,
et al. A simulation-based digital twin for model-driven health monitoring and
predictive maintenance of an automotive braking system. Proceedings of the 12th
International Modelica Conference, Prague, Czech Republic, May 15-17, 2017 132.
2017. p. 35–46.
[67] Creating a digital twin for a pump. 2019 Available: https://www.ansys.com/zh-tw/
about-ansys/advantage-magazine/volume-xi-issue-1-2017/creating-a-digital-twin-
for-a-pump, Accessed: 2019-09-24.
[68] Zhang Z, Wang X, Wang X, Cui F, Cheng H. A simulation-based approach for plant
layout design and production planning. J Ambient Intell Humaniz Comput
2019;10(3):1217–30.
[69] Gupta A, Kundra TK. A review of designing machine tool for leanness. Sadhana
2012;37(2):241–59.
[70] How IoT is turning Rolls-Royce into a data-fuelled business. 2019 Available:
https://www.i-cio.com/innovation/internet-of-things/item/how-iot-is-turning-
rolls-royce-into-a-data-fuelled-business, Accessed: 2019-09-24.
Q. Qi, et al. Journal of Manufacturing Systems xxx (xxxx) xxx–xxx
19