You are on page 1of 18

Renewable and Sustainable Energy Reviews 141 (2021) 110801

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews


journal homepage: http://www.elsevier.com/locate/rser

Towards the future of smart electric vehicles: Digital twin technology


Ghanishtha Bhatti a, Harshit Mohan b, R. Raja Singh c, 1, *
a
Department of Electrical and Electronics Engineering, Vellore Institute of Technology, Vellore, India
b
Department of Electrical Engineering, Indian Institute of Technology, Roorkee, India
c
Advanced Drives Laboratory, Department of Energy and Power Electronics, Vellore Institute of Technology, Vellore, India

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

Keywords: Worldwide, transportation accounts for 18% of global carbon dioxide emissions (as of 2019). In order to battle
Digital twin the impending threat of climate change, consumers and industry must adopt sustainable transport that complies
Smart electric vehicles with the United Nations Sustainable Development Goals of increased energy efficiency and reduced greenhouse
IoT
gas emissions. To fulfil these objectives, a new class of vehicles has recently emerged, smart electric vehicles,
Vehicle health monitoring
Battery management system
which is forecasted to reduce carbon dioxide emissions up to 43% as compared to diesel engine vehicles.
Intelligent charging However, to bring these vehicles to the mainstream, supporting architecture is needed to optimize them in a
Power converter sustainable manner. One such novel architecture is Digital Twin Technology, which is a virtual mapping tech­
Sustainable transportation nology, extending from it, capable of investigating the lifecycle of multisystem bodies in a digital environment. In
recent years, digital twin technology is becoming an underpinning area of research globally. As a result, novel
individual research covering digital twin implementation on various aspects of smart vehicles has transpired in
research and industrial studies, consequently allowing digital twin technology to evolve over the years. This
work aims to bridge the gap between individual research to provide a comprehensive review from a technically-
informed and academically neutral standpoint. Conceptual groundwork of digital twin technology is built sys­
tematically for the reader, to allow insight into its inception and evolution. The study sifts the digital twin
domain for contributions in smart vehicle systems, exploring its potential and contemporaneous challenges to
realization. The study then proceeds to review recent research and commercial projects for innovation within this
domain. To the knowledge of the authors, this is the first extensive review of the application of digital twin
technology in smart electric vehicles. The review has been systematically classified into specific domains within
the smart vehicle system such as autonomous navigation control, advanced driver assistance systems, vehicle
health monitoring, battery management systems, vehicle power electronics, and electrical power drive systems.
An in-depth discussion of each vehicle subsystem is undertaken to present this review as an eclectic panorama of
the smart vehicle system. This review further facilitates appreciation of the role of digital twin technology within
each classification from a holistic technical perspective. Finally, the work ends with an inspection of the techno-
socio-economic impact of digital twin technology that will revolutionize mainstream vehicle technology and the
obstacles for further development.

1. Introduction different as IoT is characterized solely by the physical implementation,


whereas CAD focuses exclusively on stand-alone representation in the
As can be deduced from its etymology, a Digital Twin (DT) is defined digital domain. Only the digital twin technology uniquely focuses on
as ‘an integrated multi-physics, multiscale, probabilistic simulation of a bilateral interdependency between the virtual and physical representa­
complex product, which functions to mirror the life of its corresponding tions [2]. This poses various inherent benefits as the physical product
twin’ [1]. The major objective of this technology is to serve as a panoptic can adapt to modify its real-time behaviour concurrently to the feedback
reflection of a physical body in the digital world. While it is sometimes generated by the digital twin. Conversely, the bridging allows the
mistaken as IoT or Computer-Aided Design (CAD), it is fundamentally simulation to be able to precisely mirror the real-world condition of the

* Corresponding author.
E-mail address: rrajasingh@gmail.com (R. Raja Singh).
1
This work was supported by Vellore Institute of Technology (VIT), India through SEED Grant. Dr. R. Raja Singh is with the Danfoss Advanced Drives Laboratory of
Department of Energy and Power Electronics, Vellore Institute of Technology, Vellore, Tamil Nadu, India (rrajasingh@gmail.com)

https://doi.org/10.1016/j.rser.2021.110801
Received 16 August 2020; Received in revised form 13 January 2021; Accepted 8 February 2021
Available online 17 February 2021
1364-0321/© 2021 Elsevier Ltd. All rights reserved.
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

physical body. For the realization of a true cyber-physical system, the 1.2. Potential of digital twin technology in the automotive industry
digital twin must have a dynamic interconnected nature with the
physical model, which can be achieved through real-time sensory data. Aspects of the most technology can be comprehensively modelled by
In this process, the environment of the machine must also be dynami­ other techniques, but systems involving integrated mechanical, elec­
cally replicated in the digital world, including all compatible parameters trical, and software constituents can only be replicated in the virtual
that affect its function [3]. Fig. 1 visualizes these ingredients and fea­ workspace via digital twin technology. A primary exemplar of such an
tures of the digital twin ecosystem. interdisciplinary system is electric vehicles (EV). Electric vehicles pre­
sent major incentives from an environmental standpoint. They allow
1.1. Internet of Things and the foundations of digital twin technology easy switching to renewable energy sources, lowered greenhouse gas
emissions, and increased energy harvesting efficiency, even when
Internet of Things (IoT) is evolving to be giant within the electronics powered by coal-fuelled power plants. There are many aspects of an
industry in recent times. Due to its expansive nature, a consensus can be electric vehicle that have been automated in research to give rise to the
made that there must be no single definition of what IoT encompasses automotive phenomenon known as smart electric vehicles or intelligent
[4]. However, to aid in the understanding of its structure, IoT can be electric vehicles. The exponential growth of smart control systems has
imagined as a network of networks of uniquely distinguishable nodes or brought forth several developments in the automotive industry and led
“things” that collect, transmit and process the data [5]. A study by to autonomous or self-driving cars being built. Self-driving vehicles use a
Asghari P et al. [6] shows that IoT and spinoff technologies have range of vision sensors, as well as a navigation system and actuators to
penetrated various domains as shown in Fig. 2 (data retrieved from monitor mechanical devices and processors. These interprets the data to
Ref. [7]). Nevertheless, with their increasing use, it becomes crucial that solve traffic and driving issues. This innovative technology involves an
IoT infrastructures comply with the Quality of Service metrics (QoS), interdisciplinary mix of information technology, electrical systems, and
which ensure that any visionary framework provides sufficient security, mechanical models [10]. Perhaps the most well know intelligent tech­
reasonable cost, increased reliability, controlled energy consumption, nology in the automotive industry is autonomous steering and has made
short service time, and widespread availability [6]. With security being for various studies. Zhao et al. [11] have in their study built an auton­
a major challenge for all recent technologies, not only IoT, academia has omous driving platform for electric vehicles using the Controller Area
presented various novel authentication schemes for the IoT framework. Network (CAN). The system is made intelligent by its depth perception
During its nascent stage, the authors of [7] forecasted that the IoT so­ ability as well as path planning capability, which eliminates the need of
lutions would constitute a global market owing to a net worth of 7.1 a human driver. Other intelligent transportation systems are based on
trillion dollars in 2020. assisting the driver for increased energy savings and a more comfortable
Just as IoT has been used to revamp many existing or conventional driving experience. MyEco-ACC is an adaptive cruise control platform,
technologies, there has been much advancement that served to improve conceptualized in Ref. [12], that uses the Hammerstein identification to
the utility offered by IoT. One of these monumental innovations is dig­ capture driving style along with regenerative braking to provide the
ital twin technology. The relationship between the DT and IoT repre­ most fuel-efficient and smooth drive for the user. In comparison, the
sents a mutually advantageous architecture. Where IoT generates droves researchers in Ref. [13] use predictive modelling to manipulate the
of real-time data, digital twin technology allows this data, and the data characteristics of an electric motor, such as torque, to maximize the
from exclusively the real-world system, to be analysed simultaneously. energy savings of the vehicle. Intelligent electric vehicles may also be
This is immensely beneficial in optimizing manufacturing, boosting focused on other essential requirements such as increased safety of the
service-oriented processes, and identifying needed design improvements passengers. Guo et al. [14] propose in their research, direct yaw control
[8]. This extends the ability of IoT exponentially, building the founda­ which enables the electric vehicle to keep within lanes and provides the
tions for self-improving devices. Furthermore, basing digital twins on steering system with increased levels of stability. Another safety-based
semantic data allows prognostic development, a key requirement in the intelligent EV technology, proposed in Ref. [15], uses fuzzy adaptive
Industrial Internet of Things (IIoT), forming a cornerstone for Industry control for car following, which reduces the possibility of mid lane
4.0 [9]. driving and, therefore, lowers accident rates. Another desirable appli­
cation of intelligent control is ride-comfort, which is a key factor to
improve market penetration of EVs. The recurrent menace of vehicle

Fig. 1. Ingredients and features of the digital twin ecosystem.

2
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

Fig. 2. Utilization of IoT and sister technologies.

noise and vibration present a hindrance to this objective, and combat­ produced from the conceptual definition, modelling and maturation,
ting it is a profitable avenue in smart vehicle research that has been and industrial assembly production is used as data mines for the DT
employed by various leading automotive manufacturers [16]. Smart [18]. The virtual representation is then formed by the amalgamation of
electric vehicle objectives also stem from the requirement of efficient 3D CAD models or geometric sketches, body characteristic profiles like
energy management. The use of machine learning algorithms, such as computational fluid dynamics or thermal models, and physics-based
backpropagation neural network and support vector machine, are archetype to represent the motion of the automotive [19]. For
common within this paradigm, however, Tang et al. [17] proposes a comprehensive representation, the environment of the vehicle must also
novel approach utilizing Extreme Machine Learning to boost precision be dynamically replicated in the digital world, including all compatible
and computational efficiency. While the algorithms and technologies parameters that affect vehicle function. Other human operator records
that make up an EV are vast, their research and development would be such as historical product progress, fleet data, and maintenance logs can
greatly optimized through the predictive testing and smart modelling provide the digital archetype with relevant context to represent the
capabilities offered by digital twin technology. current or future condition of its surroundings and inner mechanisms. It
Electric vehicles are very well suited for virtual representation, but is evident that digital twin technology poses great potential for creating
the complex subsystem interdependencies prove to be a challenge when a resource-friendly testing and optimization approach for electric vehi­
a realistic model is being implemented. The primary facet of digital twin cles, and smart vehicles generate an abundance of sensory feedback and
technology is that is largely data-driven and the dataset from the vehicle real-time data, making them the ideal candidates for replication as
life cycle enables it to replicate even interdisciplinary vehicle charac­ digital twins. Fig. 3 visualizes an overview of the applications of DT in
teristics. This dataset constitutes the product information passing from the automotive industry.
its Beginning of Life (BoL) to the End of Life (EoL), and the information

Fig. 3. Perspective of the digital twin technology in automotive industry.

3
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

1.3. Significance of the study: digital twin technology within the smart categorized into three phases: formation, incubation, and growth. This
vehicle domain progression and the major technological milestones are shown in Fig. 4.
The primary appearance of the digital twin can be attributed to
With the emerging research on digital twin technology, it becomes Grieves in 2003. Scarcely any other mentions of the digital twin were
crucial to evaluate the extent of penetration of this novel technology in documented in this formative period of digital twin technology. Up until
the smart electric vehicle sector. Therefore, this work aims to bridge the 2011, the fast advancement of the communication technologies added to
gap of individual research to provide a comprehensive review of this the ascent of digital twin technology. In 2011, the initial paper was
pioneering field. More specifically, this work yields the following distributed, which explained how digital twin technology could aid the
contributions: estimation of airplane operational life-span. Later in 2012, NASA
First, this work aims to build the readers foundational knowledge on formally defined the concept of digital twins and envisioned the possi­
the principles and inception of digital twin technology. The idea of bilities it could entail in the aerospace industry [20,21]. Following its
digital twin stems from research in the aerospace sector, spanning formal definition, a sharp growth is observed in academic examinations
various uses and stages of evolution. The first part of this paper distils of digital twin technology, from conceptualizations to rudimentary
fundamental information for the reader, to allow a concrete under­ proof of concept. Therefore, this era can be described as the cultivation
standing of the conception and evolution of digital twin technology. The or incubation stage. The main white paper was published in 2014, which
section continues by offering an insightful investigation into the phases mirrored the development of digital twin surpassing its predominantly
of implementation in the digital twin paradigm. It discusses in detail the avionics-oriented foundations to an array of diverse real-world appli­
workflow of building and utilizing a complex digital twin and overview cations. In 2018, Gartner characterized digital twin technology as one of
the factors affecting every stage of realization. the leading emerging innovations to dominate industries in the future
Next, the focus is streamlined to the application of digital twin [20]. As indicated in Ref. [3], the digital twin is an exceptional simu­
technology in smart vehicle systems. The study presents an in-depth lation framework, dependent on the professional understanding and
systematic review and synopsis of recent advancements of digital twin real-time data gathered from the current build, to understand a pro­
technology in smart electric vehicles. The review covers the novelty and gressively precise replication in diverse measures of time and space. The
contributions of the research and commercial projects that are available authors in Ref. [21] suggested that digital twins are advanced in­
in recent literature. As the scope of this review covers an emerging field, terpretations to define construction and product performance. There are
a comprehensive review of the technology holds great benefits and is the various understandings of digital twin [22]. A few specialists accept that
first of its kind to the knowledge of the authors. The review has been the digital twin examination should focus on simulation. Others contend
systematically classified into specific domains within the smart vehicle that the digital twin contains three dimensions: physical,
system such as, autonomous navigation control, advanced driver assis­ computer-generated or virtual, and linking parts [22,23].
tance systems, vehicle health monitoring, battery management systems, Six aspects are particularly significant to the formation of the digital
vehicle power electronics, and electric power drive systems. An in-depth twin, which are project data, construction data, sensor data, as-built
outlook of each vehicle subsystem provides the reader with a realistic data, data insights, and artificial intelligence. Preliminarily, the model
panorama of the smart vehicle system and the various novel imple­ is founded upon existing project data, which is used for the creation of
mentations of digital twin technology that are made possible by it. data-driven simulations of the physical entity. To expand the scope of
Finally, the work discusses the impact of this study and digital twin these simulations to a functional digital twin, construction data and as-
technology as a whole in the sphere of smart electric vehicles. The po­ built data are incorporated through horizontal and vertical integration
tential for the technology in the future is deliberated and the implica­ topologies. The complete amalgamation of these three entities forms the
tions of these developments are examined from a theoretical viewpoint. contemporary formulation of the digital twin. The crucial inter-link with
The emergence of intelligent and sustainable vehicles and the potential the real-world entities is enabled by the addition of sensory elements
they pose to revolutionize transportation is illustrated. Furthermore, procuring an abundance of raw data. Information exists in the focal
several of notable future applications of digital twin technology within point of the digital twin since it is a precondition for making new-
the smart vehicle domain are explored with their modalities and extent fangled information. However, such information is often obfuscated
of realization. The benefits that digital twin technology may offer these by several trivial data, which must be suitably filtered. The aspect of big
elements are highlighted, and the section continues by outlining the data analytics emerges from this objective of extracting pertinent data
challenges to the growth of this novel architecture. The final section insights which can be utilized by the system’s intelligence for feedback
concludes the study and confers the perspectives gained from this to the digital twin. Once the significant data has been acquired, it is run
extensive research on the development of digital twin technology. through the trained Artificial Intelligence (AI) or deep learning model to
provide critical insight into the system’s performance or decisions
2. Conceptual framework of digital twin technology regarding future utilization of the physical model. The above three as­
pects form the real-time segment of the digital twin. At long last, the
Following the rapid growth of data communication, data acquire­ linking framework connects all the parts along with information, and
ment, data simulation and IoT, digital twin technology has received administration [24,25]. The overview of these contemporary and
increased interest in the research community. Digital twin systems can real-time entities with their interconnections have been illustrated in
assimilate the physical and virtual statistics through product develop­ Fig. 5.
ment, which leads to an enormous volume of data that can be concur­
rently managed by cutting-edge analytics. These data analytics can also 2.2. Strategies for implementing digital twin systems
be correlated with digital twin technology for use in smart
manufacturing. This cyber-physical integration can be traced from its The formation of virtual prototype models can be a challenging task
inception to the current extent of development, which is delineated due to the complicated architecture of the present-day systems. Because
below. of the high significance of the system’s accessibility in the diverse cre­
ation plants, it is required to characterize a nonexclusive demonstrating
2.1. Inception of digital twin technology and its evolution procedure. In order to lessen the modelling efforts and give a typical
structure to the process of demonstrating various assets that empower
The historical backdrop of digital twin technology is fairly short, digital twins, three stages of modelling are recommended. These aim to
which can be attributed to the technological limitations during its diminish demonstrating exertion and manage the high multifaceted
nascent period. The establishment of digital twin technology is nature of machine parts without bringing about very point by point

4
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

Fig. 4. History of digital twin technology.

Fig. 5. Conceptual overview of digital twin technology.

models [26–29]. The main stage consists of the kinematics and dynamic i. The definition of the archetype’s segments that must be dis­
behaviour of an archetype. The succeeding stage is for played. This assignment includes the determining which ma­
computer-generated sensors to assemble information through the chine’s modules are critical to the framework’s functionality and
re-enactment. The final stage consists of identifying the boundaries to be are exposed to persistent wear and tear. The models of non-
refreshed for guaranteeing that the advanced archetype speaks to the critical sub-frameworks are separated to keep the virtual
genuine state of the demonstrated asset. Each stage can be further iso­ model’s computational complexity low.
lated into activities and sub-activities, which are detailed in Fig. 6. ii. The definition of levels at which the system’s modelling shall take
place, including a comprehensive outlook at the available data
2.2.1. Archetype modelling for each level. In the ideal scenario, all archetypes would be
In this stage, the developer should decide first, which archetype demonstrated as white boxes, however, due to real-time dis­
segments are to be displayed. Every advanced archetype must cover the crepancies in information a few segments are realized as black
principle somatic phenomena that influence the archetype’s behaviour boxes. A more comprehensive perspective prompts the use of dim
such as motionless distortions, pulsations or backfire. These phenomena boxes and secret elements, which would necessitate an adding
will be affected by the state or condition of the demonstrated asset, and computational time, resources, and archetypal boundaries to be
thus it can be utilized for predictive support of its future condition. With solved.
respect to the demonstrating level, three approaches must be expoun­ iii. The formation of the preliminary model and then synthesizing all
ded, namely black box (with no information on the inner activity), dim components to manufacture the total advanced archetype. This is
box (hypothetical information available) or white box (comprehensively where the kinematic chain of the archetype is conceptualized by
depicted part). The final development of the primary stage is the associating the constituent parts. The kinematic sequence of the
arrangement of the subsystems so that they may re-enact the dynamic computerized prototype comprises of guidelines and limitations
behaviour of the physical archetype [27,28]. Specifically, this stage is which intend to parameterize and bound the virtual prototype to
categorized by the following processes: the physical body.

5
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

Fig. 6. Stages of digital twin development.

2.2.2. Modelling of virtual sensors ii. The choice of computer-generated sensors and the formation of
In this stage, the selection and implementation of computer- their corresponding replicas. Consequently, the information
generated sensors are undertaken. Computer-generated sensors can observed by the computer-generated sensors is distinguished and
dictate the design of components by accumulating information from the subsequently, the required actuation mechanisms must be
archetype, similar to the function of a real sensor to the physical body. selected.
The information from these entities may then be assembled through the iii. The amalgamation of computer-generated sensors with the multi-
computation aimed at anticipating the Remaining Useful Life (RUL) for physics based archetype. The validated information and the
unprecedented prescient support. As the quantity of soft sensing devices observed parts shall be categorized as past activities, and conse­
increases, the calculation period of the model’s reproduction is quently, the association of the computer-generated sensor is
commensurately expanded. Therefore, it is essential to extract and made with the prototype as part of the operational progression.
indicate which information should be processed by the model and which From that point onward, the client will be able to gauge and
may be ignored without substantial consequence [27–29]. The activities assemble information, which emerges during the model’s
that constitute this stage are: reproduction.

i. Establishing the virtual data which must be tapped by the arti­ 2.2.3. Parameter update
ficial sensory mechanism which is to be developed. This will be The fundamental objective of this stage is to define the pertinent
utilized subsequently and classified as “straightforwardly acces­ boundaries which shall be periodically updated to synchronize the
sible by genuine sensors” and “assessed during recreation archetype with the physical machine’s present condition. These
through virtual sensors”, which will complete the data definition boundaries are flexible and will be related to coordinated model tuning
and acquisition process. [27–30]. The organisational structure and activities of this stage are as
stated below:

6
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

i. Selecting the DT units that need to be tuned by the simultaneous technology poses huge impacts to the evolving field of Autonomous
reproduction tuning component, which empowers the concept of Guided Vehicles (AGV). Advanced machine systems that are capable of
the DT as a real-time synchronized entity. This determination is intelligent decision making, make up one of the pillars of Industry 4.0.
made by mulling over the accessible information originating from Autonomous machine decision making can be realized by implanting
outer sensors or machine’s controller. The criticality of the CPS architecture with different configurations and algorithms that are
segment (its impact to the machine’s performance), and the dependent on the objective of the model. With this superior system in
modelling level of the segment (white, dim, or black) must be place, the machine can proactively carry out and enhance tasks using
extensively formulated. self-adaptive behaviour. The control for an AGV algorithm operates
ii. The definition of the accessible information to be utilized for incrementally, with odometry computation conducted at every step to
regulation is a significant design boundary. This must be delib­ form a viable path to the required target waypoint. The resultant control
erated for the definition of the computerized archetype to be signal is dynamic and even once the traversal is completed the algorithm
regulated alongside the wellspring of the previous information. continues to operate, evaluating the performance metrics of the AGV for
iii. The determination of the updatable demonstrating boundaries is future reference. Using this incremental algorithm with feedback-based
the last stage when the computerized archetype to be regulated, reinforcement the AGV is able to traverse the most efficient path to
and the necessary information for the regulation is established. destination. This workflow is illustrated in Fig. 8.
The archetype model regulation implies the regulation of The use of digital twin technology is varied within the works of
numerous displaying boundaries to empower the modification of autonomous vehicle control within academia. Bottani et al. [32] show
the computerized model. the use of an AGV digital twin for preparing the AGV control system with
control responsess to an array of case scenarios. Such ability to introduce
The previously mentioned activities and sub-processes are the pri­ the machine to scenarios in which its critical decision making ability is
mary workflows by which the advanced archetype of a product is con­ exposed without the possibility of simultaneously introducing a biasing
structed. Through the course of these phases, a concise methodology stochastic factor is a major proponent of realizing an accurate autono­
should be used to empower web-based fine-tuning of the digital twin. mous model using DT technology. Furthermore, the vast amount of
This system is founded upon information trading mechanisms employ­ simulated scenarios that the machine can be exposed to using this
ing the genuine system, its advanced virtual model, and the tuning of the method. This gives the machine learning model its biggest utility:
latter as per the formulated workflow. In the real workspace, the sys­ vehicle big data, which is impossible at the same scale using a physical
tem’s controller and outer sensors are utilized to accumulate the model. A similar approach has been employed by Rassõlkin et al. [33],
necessary information. The accumulated information from these sources where the vehicle big data for training of the automated control of the
is then organized, examined, and contrasted with the insights gained by vehicle propulsion drive is generated by exposing the system DT to case
the virtual sensory components. Subsequently, the data is updated for frameworks and recording the drive responses. This exhaustive frame­
synchronous operation of the virtual archetype. This in turn enables the work, while still not immersing the entire smart vehicle system, provides
continued accurate operation of the digital twin and may be used to precise groundwork for training and testing future autonomous vehicles.
provide insights into the future operation of the physical model in the The studies covered so far use the DT to train the learning algorithm, but
real workspace. Ruba et al. [34] instead deploy a Field Programmable Gate Array (FPGA)
propulsion drive DT as a viable and compact alternative to a physical
3. Synopsis of smart vehicle digital twin technologies test platform for drive control. The multi-faceted applications of this
model are strong indicators of novelty and commercial potential as an
In the automotive industry, digital twins are primarily utilized for industrial test-bed. An abundance of vehicle control systems are arbi­
complete or partial construction of the vehicle’s virtual model. This trary and the control parameters must be defined explicitly while testing
involves the collection of the vehicle’s behavioural and functional in­ the machine. Guerra et al. [35] propose the optimization using P-PI
formation to analyse the actual performance to provide customer- control, where the objective is to reduce the maximum absolute position
specific support. The digital twin is presented either as a model error. This technique paves the way for utilization of DT for tuning
replica or as a collection of interwoven components. Engineers explore control systems in-situ the AGV design process, while conventionally
AI to design the ideal method of producing automotive products before such processes require rigorous physical assessments.
they enter the production line. By analysing the simulation models, the To derive a reliable replication of an AGV, its motion must be
potential problems such as breakdowns, wear and tear can be predicted. completely synchronized to the physical model. This may be achieved by
Unlike road testing and maintenance, the use of digital twins for real-time interconnection or predictive modelling. For the latter, the DT
autonomous vehicles can potentially save many unforeseen expendi­ EV model requires the capability of accurately modelling future motion
tures. The effects of digital twin technology are substantial to smart using archival datasets. A rudimentary concept of a DT based on pre­
electric vehicles, as various aspects of the system may be modelled and dictive modelling was studied by Yang et al. [3]. The DT is used to
optimized. To be able to imbibe the novelty of digital twin vehicle deploy a statistical framework to generate the future motion of the
modelling it becomes crucial to understand the aspects of the digital vehicle based on the historical data. While the technique is novel and
twin ecosystem. We can broadly classify the components of digital twin dynamic, the dissonance introduced by using statistical modelling rather
technology as physical products in the real-world workspace, digital than learning ensues more erroneous results in the long term. The
products in the virtual workspace, and the bidirectional connections of objective pursued by Zhang et al. [36] is similar, though rather than a
system information that bind the digital and physical products together statistical prediction, time series modelling is used to build use of digital
[31]. It is indeed possible to model the complete EV in the virtual twin technology to dynamically compute motion parameters for reali­
workspace, however, when the objective of a study is streamlined, the zation of an autonomous ultra-precision motion system. They present a
experimentation may be bespoke and less computationally intensive. As procedure for optimizing moving trajectory of the vehicle using DT
a result, a study of academia shows some prominent objectives and the technology which can collate and execute motion without backlash such
subsequent realization of the digital twin technology on certain smart as cruising and curving instead of spasmodic zigzag motion (as with
electric vehicle constituents, as summarized in Fig. 7. most other mobility models). This procedure presents immense potential
in adding intelligent seamless cruising and drifting as a salient feature to
3.1. Predictive mobility and autonomous motion control AGVs, which is traditionally attributed to human driving skill. Table 1
summarizes various implementations of DT technology for modelling,
Within the vast paradigm of Cyber-Physical Systems (CPS), DT controlling, and predicting mobility in literature.

7
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

Fig. 7. Overview of digital twin technology for smart electric vehicles.

Fig. 8. Decision algorithm for autonomous guided vehicles.

3.2. Advanced driver assistance systems data can augment the response or threat detection capabilities of
existing ADAS architectures. Cloud-based ADAS interface with a DT has
There are many factors that shape consumer outlook of a vehicle like also been studied by Wang et al. [43], though their sophisticated
energy savings, comfortable driving and efficiency, but the most approach uses a Vehicle-to-Cloud interface for interconnection between
fundamental factor is always safety. Advanced Driver Assistance Sys­ multiple vehicles, creating a cooperative safety system. This model is
tems (ADAS), is the progressing number of security corollaries built to exceedingly pertinent to the urban traffic paradigm as multi-vehicle
enhance driver, passenger and pedestrian safety by reducing both the networks are crucial to expanding the use of smart vehicles beyond in­
amount and impact of motor vehicle accidents. The main objectives of dividual optimization and safety. While both works provide a novel
an ADAS is to alert drivers when possible oncoming dangers are detec­ approach for DT-based ADAS, the systems remain independent of any
ted, to intercede in critical situations to ensure that the driver remains in learning algorithms, which limits their applications. Alam et al. [44]
control of the vehicle, and to avert accidents or decrease the intensity of expand the horizons for use of Cloud-based Cyber Physical Systems
an accident if it cannot be circumvented [37,38]. The systems archi­ (C2PS) using DT with the integration of fuzzy logic and Bayesian
tecture of a typical ADAS is shown in Fig. 9. Table 2 provides a collation inference to maximize the efficacy of the derivative telematics-based
of the application of DT technology to ADAS systems surveyed in driver assistance. The use of fuzzy logic in ADAS DT systems is funda­
literature. mental to facilitate the shift from semi-autonomous vehicles to
Conventionally ADAS rely on sensors to determine future threats, but self-driving vehicles. While the general focus towards any
the perceived information only includes their dynamic status without human-computer interaction involves physical constraining, it is equally
historical information pool. Newer models proposed by academia show important to consider standards, regulation, policy and the technology’s
the integration of historical data to optimize the decision-making compliance with it. To realize this objective, Damjanovic-Behrendt [45],
[39–41]. However, the centralized nature of this historical data pool creates a composite architecture where the digital twin can make ana­
hedges its applications, and this is where the need for cloud-based DT lytics for driver safety using real-time sensor information from the
ADAS arises. Liu et al. [42] suggest using DT technology to build virtual physical smart car along with the General Data Protection Regulation
duplicates of real driving decision factors such as other drivers or pe­ knowledge base. This allows an intelligent ADAS which does not use
destrians. The data and interaction of these entities are cloud processed personal information to an extent that is beyond the jurisdiction of the
and fed to the decision-making algorithm. The insight gained from this law, a quality that can effectively boost public perception and

8
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

Table 1 hefty and incoherent to sleek vehicle designs. Tao et al. [48] propose a
Predictive mobility and autonomous motion control. similar paradigm with increased sophistication in what they name as the
Sl. Main Objective Methodology Used Role of Digital Year “Five Dimensional DT model for prognostics and health management.”
No Twin Technology [Ref. No.] This DT model allows interaction that can discriminate the interference
1. Autonomous Discrete Event To compile the big 2017 [32] from the environment, defects in the models, and potential anomalies in
Guided Vehicle Simulation, data from the the machinery. By utilizing equipment data, digitally produced data,
(AGV) motion Machine-to- prototype and and fused data the DT can accurately assess vehicle health and also the
planning. Machine Logic. provide future need for maintenance. Other works highlight vehicle subsystems
navigational
insights for path
where health monitoring would be more substantial and hence DT
planning and technology would play a crucial role. Magargle et al. [49] focus their
machine research on health monitoring and predictive maintenance of a vehicle
interaction. braking system. The developed DT uses electronic simulation, electro­
2. Automated Convolution Used to provide 2019 [33]
magnetic effect, brake pad material properties, vehicle dynamics, and
Electric Neural Networks, drive response
Propulsion Drive Transfer Learning. data in diverse speed sensing to give accurate health insights. It can be deduced that
Control. case scenarios to utilizing such a vast plethora of multi-physical braking systems data
the CNN. Also used contributes substantially to the success of their model in terms of pre­
as an evaluation diction accuracy. Another system which is subject to constant degra­
platform for the
machine learning
dation in EVs is the electric motor, and its health monitoring is studied
model output. by Venkatesan et al. [50]. The developed DT provides onboard health
3. Autonomous EV Energetic Used to create a 2019 [34] monitoring and prognosis, with the creation of a novel Remote Health
Propulsion Drive. Macroscopic virtual test bench Monitoring and Prognosis Centre (RHMPC). This centre is developed for
Representation for EPDS control
at-a-distance monitoring using cloud-based transmission by the EV
using FPGAs algorithms.
4. Autonomous Computerized For empirical 2019 [35] management authority. Using this methodology, the remaining useful
ultraprecision Numerical parameter life of the motor is deduced computationally without the need for
motion to reduce Control, P-PI optimization detaching the motor from the smart vehicle interface, which cements its
backlash. control. without a physical novelty and enhances commercial usability. Zakrajsek & Mall [51] uti­
model.
lize health monitoring for the vehicle subsystem with maximum
5. Motion Statistical To implement the 2017 [3]
Predictive Modelling from framework and contact-based degradation: tires. Wide-spread tire maintenance or
Modelling using correlation to juxtapose the replacement strategy is failure-based, which should be obsolete in the
Historical Data. historical datasets. model prediction era of smart vehicles. The developed DT was able to compute Probability
with physical
of Failure (PoF) for distributed ranges of tire conditions and speeds.
prototypes.
6. Mobility Time series To create the EV 2019 [36] Consequently, it provided insight into the tire’s estimated life and
Behaviour behaviour fleet integrated maintenance requirement. Table 3 summarizes the various novel con­
Modelling of an modelling. with behaviour tributions in the field of DT health management systems by academics.
EV fleet. framework and
simulate its
behaviour in
3.4. Battery management systems and intelligent charging
various case
scenarios. Battery Management Systems, or BMS, are vital in the consideration
of battery life, reliability and safety in EVs. The BMS is a microprocessor-
based unit that integrates multiple functions and submodules that are
market-value.
not actually distinct physical units but are seen for clarification sepa­
rately shown in Fig. 11. Various types of sensors, actuators, including
3.3. Vehicle health monitoring and management controls are in EV BMS. BMS measures the capacity of the battery, the
battery depreciation during charging/discharging and the right battery
Integrated Vehicle Health Management (IVHM) is the architecture performance and provides users with this information in real-time [52,
that utilizes the sensor input from a vehicle executing prognosis and 53]. To measure the electrical quantities of the entire string as well as
diagnosis to evaluate and forecast the health of its critical frameworks that of a single cell, a voltage and current measurement unit is mounted.
[46]. Originally, NASA employed IVHM to collate data, perform diag­ To calculate the temperature of the battery and coolant, the temperature
nosis, forecast issues, resolve faults, and provide platform maintenance control unit is added, hence it is possible to monitor and control the
in space crafts. Later, citing its success in the aerospace domain, it temperature of the battery system through this unit. Some analog and
proliferated to other vehicles in the terrestrial and marine fields. Instead digital inputs are also included in this system, such as an accelerating
of relying on temporal or fault-based maintenance, vehicle health and brake pedal sensor, charging control and an ON/OFF engine switch
monitoring has enabled Condition-Based Maintenance (CBM) of the [54]. The control unit for balancing consists of capacitors and resistors
vehicles. IVHM ascertains that the subsystem functions in its normal for power dissipation, which is used in battery packs to equalize the
state, without any unforeseen failure. Within this paradigm digital twins charging state of the batteries. The high voltage safety unit is used to
offer automotive manufacturers an increased capacity to diagnose prevent the battery packs from sustaining physical damage. The device
aberrant states and anticipate remaining useful life of vehicle’s ablative also protects the battery packs from situations of overcharging and
components without any need for field testing, thereby improving safety over-discharge. The BMS determines the number of running batteries
and owner satisfaction. Fig. 10 shows how DT technology augments the utilized in the electric vehicle with higher-precision while also tracking
process of vehicle health management. and control of batteries. The BMU controls charging and discharging,
Ezhilarasu et al. [47] discusses how the digital twin can be utilized to cell balance, connection to the high voltage line, and cooling [55]. An
optimize Maintenance Repair and Overhaul (MRO) and build a smart efficient BMS is one of the primary components in EVs to guarantee the
interface for revealing the age-centered degradation of vehicle health to efficient, and long-lasting operation of a battery while dealing with the
support condition-based maintenance. It is lucid in their discourse that a electric grid and challenging driving conditions [56].
significant challenge faced by DTs in IVHM systems is the trade-off be­ Information on battery conditions, such as the state of available
tween weight constraints and the need for a surfeit of sensors, often power (SoP), state of charge (SoC), state of life (SoL) and state of health

9
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

Fig. 9. Advanced driver assistance system architecture.

period and lessening the value from the fully charged battery rating. It
Table 2
can be safely charged/discharged at levels suitable for battery lifecycle
ADAS digital twin systems.
enhancement by calculating the battery’s current capacity. A battery’s
Sl. Methodology Used Role of Digital Twin Year energy capacity relies on its charging current, discharge current, span,
No Technology [Ref. No]
temperature of operation, cut-off voltage, and user profile. Various
1. Human-in-the-loop Generating historical data of 2020 [42] methods have been developed for the estimation of battery SoC [59,60].
simulation, time to collision and speed
These estimation methods can be categorized into different forms:
Sensor Fusion, Distance variance by exposing it to
Matching Algorithm. case scenarios.
electrochemical, electrical and adaptive based. While highly specific,
2. Vehicle-to-Cloud, Ramp To create data and provide 2020 [43] electrochemical techniques are considered hard to implement in soft­
Merging Assistance, Data- predictions using a ware or hardware because they explicitly refer to the physical properties
Cleaning and Sensor Fusion. collaborative approach by of the batteries [61]. A battery equivalent model and a solution algo­
replicating several smart
rithm are used in adaptive methods, e.g., Kalman filter, fuzzy logic and
vehicle nodes and their
interactions. neural network algorithm [62–64]. The residual amount of charge is
3. Internet-of-Vehicles, To implement a telematics 2017 [44] determined from the present SoC and the energy expended and the miles
Deep Integration of Sensor based ADAS with the real- travelled, depending on the latest driving trends. For electric vehicle
Fusion, Bayesian Network, time data influx from battery is the only source, hence the precision of the level calculation is
Fuzzy Logic. physical vehicle and provide
safety insights using an
more important. A basic and important issue in the design of BMS in EVs
integrated machine learning has always been reliable SoC estimates. Not only can reliable and ac­
model. curate calculations be used to determine a battery’s reliability, but also
4. Edge Computing, Cloud To process historical data, 2018 [45] to provide some important statistics, such as the remaining energy
Computing, Temporal and sensor fusion and correlate
and/or the remaining useable time [65]. In other words, the SoC dis­
Location-Based Behavioural with privacy policy
Analysis. requirements to generate plays the driving range of the vehicle or the battery’s remaining capacity
predictions for driver safety in EVs. These insights may be used for maintenance purpose or to alter
based on model behaviour. operational processes which help extend its useful life. A BMS ensures
the fault-resistant operation of a battery pack and to realize it’s an
equivalent cell model must be solved. However, present BMS technology
(SoH), is also given by an effective BMS. Battery SoC model is an algo­
is not customized to the perspective of the automotive industry, as it is
rithm for determining the behaviour of the battery in response to
far from supplying accurate results in a real-time scenario for wide
different internal and external circumstances. SoC model logs history to
operating ranges. Using complex equivalent cell diagnosis, the internal
predict the battery state at any time for service purposes or to estimate
condition of individual battery cells may be computed, in addition to
the travelling distance of the electric vehicle until the battery needs to
measuring sensor data. Through collation of big data in one database,
recharge [57]. The distance travelled is extracted from data made
the battery system predictions and data-driven optimization can be
accessible on the CAN Bus by sensors. Calculating the battery SoC for the
accomplished by deep learning algorithms, which will develop an un­
above-noted functions is an integral feature of the battery model [58].
derstanding of battery life stages.
The SoC is calculated basically by averaging the current flow over a
With this methodology in mind, Li et al. [66] suggest a cloud-based

10
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

Fig. 10. Digital Twin Technology Integration with Integrated vehicle health management frameworks.

such as Health Indicators (HI) shown by Qu et al. [71]. This work


Table 3
demonstrates a long short-term memory (LSTM) modelled Li-Ion battery
Vehicle health management digital twin systems.
DT, which provides dynamic SoC indication as the DT is virtually dis­
Sl. Focus of Health Role of Digital Twin Technology Year charged. Peng et al. [72] achieve a similar aim with respect to spacecraft
No Management [Ref. No.]
Li-Ion batteries, simulating their degradation via a Kalman Filter - Least
1. Entire Vehicle Provides data to carry out condition 2019 [47] Squares Support Vector Machine realized in LabVIEW. While Li-Ion
based maintenance to improve the
batteries are a common subject of these studies, all modern appliances
industrial process of maintenance,
repair and overhaul.
are gradually shifting to solid-state configurations citing their energy
2. Entire Vehicle or Processing physical, digital and fused 2018 [48] density and performance metrics. Park et al. [73] explore this area by
Complex Subsystem data to provide insights of current devising a digital twin-driven model for an all-solid-state battery using
health and future degradation rate. voxel-microstructures.
3. Vehicle Braking Simulate various dynamic factors 2017 [49]
Directly correlational to battery management systems are smart
System such as material properties and
vehicle dynamics to study the effect charging architectures that tie into the overall efficiency of the smart EV
on a braking system and provide system. Electronic Control Units (ECU) and EV power trains form the
predictive maintenance information. cornerstone of this intelligent technology. Therefore, the optimization of
4. Electric Vehicle Provides stable cloud based 2019 [50]
these components using DT technology plays a critical role in realizing
Motor interconnection for real-time health
statistics as well as in house prognosis
an efficient charging architecture and is widely studied by academics.
of the electric motor operation. Fig. 12 illustrates the harmonious interconnection between the digital
5. Vehicle Tires Utilize and simulate tire conditions 2017 [51] twin and charging behaviour. Shikata et al. [74] realizes a digital twin
and speeds to compute the durability, based electrical, mechanical co-simulation (virtual car) and present its
health status, and probability of
behaviour while charging. The virtual vehicle DT made it possible to
failure.
develop an EV automatic charging system where simulation-based
verification made improvements to the algorithm even before the real
system, by utilizing the IoT frameworks, to build a digital twin for system was deployed. EV charging can often have a detrimental effect on
battery systems in the cloud which promotes improvements to processor the public grid due to its heavy load demands. To counteract this effect
capability, data storage capacity, and robustness of the BMS. With a Korotunov et al. [75] use a genetic algorithm to optimize a digital twin
kindred objective, Ramachandran et al. [67] utilizes equivalent cell of EV charging infrastructure to realize a viable model of a smart grid.
circuit modelling and real-time RLS updating to build a DT that provides Table 4 summarizes the novel methodologies and use of digital twin
precise insights of the battery charge estimation and condition to the technology for BMS and smart charging systems.
BMS. Sancarlos et al. [68] upscales the DT paradigm with
physics-focused cause-effect relationships, known as hybrid twin to­
pology. Within this framework, the BMS is based on electrochemistry 3.5. Vehicle power electronic converters
modelled DT providing a self-correcting function to the battery. Most
BMS systems are designed to come into operation once the product is Power electronic devices form the cardinal core of major electric
functional, however, Merkle et al. [69] enhance the application domain vehicle systems such as EV propulsion, battery charging, and power
of battery modelling and management to product design and accessories [76]. Distinctly, power electronic converters present an
manufacturing stages using the digital twin technology to provide in­ array of use cases within an electric vehicle due to the need for constant
sights to stakeholders on the design, manufacturing and operational and regulated power flow [77–79], as highlighted in Table 5. The
metadata. A study of literature also unravels a promising area to interconnected layout of various power electronic devices to the vehicle
enhance the regiment of battery management systems, which is the use subsystems is further illustrated in Fig. 13. However, for the ideal
of DT to produce exhaustive virtual models of batteries. These virtual functionality of these converters in automotive applications various
models, otherwise called battery digital twins, have emerged quite designs and physical constraints must be considered [78], some of which
recently in academic literature, but their applications are vast and can are as follows:
revolutionize the utilization of BMS algorithms. They can also aid in
intricate studies that correlate nano-scale effects to macroscopic con­ • High efficiency
sequences, providing a never before look into the electrochemical but­ • Lightweight
terfly effect within batteries [70]. A salient feature of Battery DTs is that • Compact sizing
they can be directly embedded with battery management technologies • Low EM noise-based interference
• Voltage step up the property

11
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

Fig. 11. Modern battery management systems framework.

• Low current ripple drawn from the battery system perspective. Capital expense is often the first feature of propulsion
• Converter power flow control for varying voltage distributions quality [85,86]. The main challenge mentioned in Refs. [87,88] is the
amount of drive power arising from the scale and the weight of the
The DT technology can be effectively utilized to evaluate whether machine. The third integral parameter closely linked to the total expense
new power electronics meet the above criterion, whether their design of the whole transportation network is the efficiency of the motor.
could result in an unforeseen setback during operation and for real-time Studies of test benches built in various testing centres include energy
condition prognosis. Milton et al. [80] offer the design of DT based storage [89], optimum design [85,90,91], and combination of different
diagnostic system for power electronic converters. The DT is designed to Battery Electric Vehicle (BEV) energy sources [92], such as batteries
be controller-embeddable, allowing probabilistic diagnosis through [93], super capacitor [94–99] flywheels [92], and fuel cells [100] as
real-time data collection. The embedded nature is a major asset to this shown in Fig. 14. There is also an advent of technologies that show
model as it waxes the compactness of the implementation, a factor promise in managing energy judiciously in EPDS and preventing typical
commonly missing in practical DT realizations. Using a similar approach noise-based losses as demonstrated in Refs. [16,17,97]. Many research
Peng et al. [81,82] propose a condition monitoring digital twin setup for studies [101–106], address effectiveness as a critical element in the
DC/DC converters. Cluster-data on the estimated status monitoring consistency of the EPDS. The New European Driving Cycle (NEDC) is
under different testing conditions using the digital twin can be used to widely used to verify several of those markers in the test benches [92,
predict faults or damage to the internal IGBTs or MOSFETs in the con­ 107]. In comparison, the current technologies primarily focused on
verter. It is evident through this work that the use of DT makes condition physical sensors and are geared towards the more robust applications
monitoring and fault diagnosis non-invasive and thus much more [108]. A few key works summarizing the objectives and technologies
cost-effective. M. Biglarbegian [83] focuses his DT-based architecture on used in EPDS are presented in Table 6. There is a limited volume of
high frequency Gallium Nitride power converters using machine research on monitoring work, which requires more complex operational
learning. The resultant DT model is used to conduct life-accelerated tests structures. Contrastingly, important aspects of power have not received
on the power converter as well as various exposures to varying strain proper coverage in the literature.
profiles. Life-accelerated modelling is one of the salient applications of The elements of modern electric power drive systems is shown in
DT for smart manufacturing that have been effectively realized in this Fig. 14, the quality of the control mainly affects the mobility of the
work, albeit at a small-scale through power converters. Jain et al. [84] vehicle, driving suitability, parking accuracy and stability of the
expands the possibilities for application of DT technology in power cruising. Consequently, the focus is mainly on the requirements related
electronics by creating a complete Photovoltaic Energy Conversion Unit to the EPDS control restrictions and tuning possibilities. The commit­
(PVECU) digital twin for fault diagnosis. This model includes a photo­ ment and tracking of critical or hottest spots in energy systems are the
voltaic energy source, source-level power converter, and controller to be main factors that allow the device to function efficiently and safely. By
able to effectively troubleshoot a complex system for faults without estimating the drive responses from EPDS′ digital twin, the methodology
physical analysis. discussed helps measure the quality of the steady-state and transient
motion. These are recommended approaches to drive regulator tuning,
feedback arrangements, sensor allocation and control looping. The
3.6. Electric power drive system
electric car can be paired with its digital twin [109] in electric vehicles.
On the vehicle, IoT sensors are installed to allow the sensors to send or
Modern Electric Power Drive Systems (EPDS) majorly measure drive
receive information from their digital twin. The data collected from the
output from the economic and the technological performance

12
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

Fig. 12. Integrating digital twin technology with battery management systems.

sensors helps to keep track of the car’s operating state and even to spot which vary due to customer-oriented modifications, product design, and
faults early to prevent high maintenance costs. processing technologies, adding to the inherent complexity of compre­
hensive modelling. Furthermore, the smart vehicle paradigm consists of
4. Outlook for the future for digital twins technology in smart numerous heterogeneous systems, which are characteristic of any in­
electric vehicles dustrial cyber-physical system, and involve modelling based on
extremely disparate parameters. There is still a lack of integration
The previous section presented a comprehensive state-of-art review capability in these diverse parameters which gives rise to data owner­
for the use of digital twin technology in the perspective of predictive ship, cleansing, and fusion challenges. The heterogeneity of the system
mobility, advanced driver assistance systems, vehicle health monitoring, may also include the involvement of third-party stakeholders, adding to
battery management systems, power electronic converters, and power the complexity of designing an all-encompassing digital twin for smart
drive system, all of which are crucial to advancing smart electric vehi­ vehicles. The ultimate goal of all organizations is to leverage the avail­
cles. The technologies and models proposed by these various researchers able data to build faster, more cost-effective and high-quality products.
have been monumental in establishing the foundation of DT technology However, the fact is that in terms of the successful utilization of their
in the automotive industry, however, this is merely the tip of the iceberg data, automotive manufacturers are at different maturity rates.
for the potential of DTs. Similarly, a plethora of constraints and obstacles Digital twin technology has strong ties to IoT and Industry 4.0
still exist before DTs can be added into the mainstream development making the technology extremely versatile and flexible. However, both
toolkit. This section aims to bring into consideration these pertinent these architectures are still set back by a lack of global standards of
opportunities and challenges that will govern the objectives of future performance, and this in turn hinders advancement in digital twin
research in this domain. technology. Furthermore, the inclusion of IoT leads to numerous con­
cerns with respect to data and system security [111]. For a system such
4.1. Challenges, obstacles and limitations as a smart vehicle, which is closely interlinked to humans, data security
risks can be potentially fatal and must be carefully examined when
Though the framework of digital twin technology is pioneering, the constructing a digital twin. With human linked systems, there exists
development of the system is equally challenging due to its abundant another element of human behaviour that affects the accuracy of system
prerequisites with respect to engineering, markets, technology and data life and mobility prediction in digital twin systems. Where digital twin
[110]. In terms of engineering challenges, perhaps the most striking is systems are used in real-time vehicle scenarios such as health moni­
the modelling of extremely intricate and interconnected systems. With toring or driver assistance there are numerous issues that can be caused
technology like smart vehicles, there are many subsystem uncertainties to the framework by lost connectivity or power outages. A loss of stream

13
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

Table 4 Table 5
BMS and EV charging Digital Twin Systems. Overview of power electronic device applications in electric vehicles.
Sl. Methodology Used Role of Digital Twin Year Sl. EV Subsystem Power Role of the Power Year
No Technology [Ref. No.] No. Electronic Electronic Device [Ref. No.]
Device
1. SoC and SoH estimation Replicating the battery in the 2020 [66]
Configuration
using extended Thevenin virtual space and providing
model and cloud computing. cloud integration to improve 1. AC motor drives Soft-switching For conversion of 2016 [77]
the computational power, inverters and storage power
data storage capability, and multilevel from battery
reliability of the BMS. inverters subsystem to AC
2. Static battery model using Realistic modelling of an EV 2018 [67] actuation system
least-squares and dynamic battery pack and utilizing it to while maintaining
updating using the RLS estimate the battery minimal EM losses.
algorithm for finding SoC. parameters when linked to 2. DC motor drives Soft-switching To control driving 2011 [78]
physical model as well as converters power supply with
when purely independent. minimal losses
3. Hybrid twin and regression Allows for real-time 2020 [68] during switching.
using sparse-Proper simulations as well as 3. EV Charging AC/DC For conversion of 2011 [79]
Generalized Decomposition adaptation of predictions by converter commercial grid
(s-PGD) considering the real driving power supply to
conditions and the real EV compatible
driving cycle to change. battery DC power.
Integrates reduced order 4. Electric propulsion DC/DC Is between battery 2019 [81]
models to evolve into a hybrid system converter and inverter to
twin. maintain the
4. Unified modelling language Provides accurate progression 2019 [69] voltage of the DC-
meta model to simulate a of battery system from bus and reduce
high voltage lithium ion manufacturing to operational current ripple.
BMS. stage for BMS-based 5. Vehicle lighting, DC/DC Provide the 2020 [82]
predictions. microprocessor, converter required low
5. HI-based LSTM lithium ion Evaluates actual battery 2020 [71] and luxury voltage for tertiary
battery digital twin. capacity and degradation systems. loads with light
based on virtual discharge demand.
profile and dynamic health 6. Regenerative Bidirectional To transfer power 2018 [83]
computation. braking DC/DC towards the
6. Kalman filter - least squares Battery DT computes SoC, 2019 [72] converter actuators during
support vector machine and SoH, and RUL through motoring and from
artificial neural networks. learning models taking data the actuators
from remote sensing links. during
7. Solid sulfide electrolyte all- DT provides insight into 2020 [73] regeneration.
solid-state battery DT. electrochemistry like dead
particles, specific contact
area, effective electronic/ 4.2. Future scope, trends, and opportunities
ionic conductivities, and
charge distribution in the 3D
domain.
There are many challenges faced by the automotive sector, specif­
8. Vehicle 3D rendering Integrates the electrical and 2019 [74] ically concerned with the electric vehicle market. One of the main
engine, controller area mechanical dynamic barriers faced by electric vehicles is the consumer outlook that restricts
network using automatic subsystems to create and consumption in comparison to conventional vehicles. In the future, the
parking, charge control and optimize an automatic EV
role of digital twin technology may be expanded from vehicle technol­
positioning ECUs. charging model with
emphasis on smart charge ogy innovation into a tool to encourage sales of smart electric vehicles.
control and charge utilization. Digital twin simulations of smart vehicles may be visualized through
9. Genetic algorithm Used to design the optimal 2020 [75] interactive VR or AR technology to entice potential users with person­
scheduling, Vehicle to Grid smart grid by genetic alisation utility as well as virtual drive experiences. The digital twin
(V2G) and Grid to Vehicle algorithm based correction.
(G2V) strategy. Used to evaluate the impact of
framework may be customized to any vehicle model and in the future
EV charging on the grid and optimize research and development with the capability to model hun­
ameliorated to reduce this dreds of EV models within seconds once the initial architecture is setup.
impact. Digital twin technology also poses the capacity to streamline the study
and testing of self-driving vehicles and other forms of autonomous
mobility. This will be the catalyst for the mainstream realization of self-
from onboard vehicle sensors and consequential loss of feedback from
the digital twin to the physical vehicle can severely affect the reliability driving cars.
As a cornerstone of Industry 4.0, the future scope of digital twin
of a predictive or AI model. Consequently, the global infrastructure for
wireless technologies and power transmission must be simultaneously technology presents the possibility to revolutionize the design and
testing of smart vehicles. The digital twin represents theoretically
developed to match the requirements of real-time digital twin imple­
mentations. While the components of the digital twin have plain sailing infinite reusability for testing in extreme scenarios. Not only will this
reduce expenses made on elaborate physical testing setups, but will also
objectives, their enabling mechanisms are numerous, ranging from
sensors, machine networking, artificial intelligence as well as sometimes result in safer, more efficient EVs that have been tested more compre­
hensively than ever before. Furthermore, the future scope for continuous
scaled-down physical prototypes (in case of some models, for instance,
large vehicles, data acquisition from the physical body may not be data analysis provides the potential to quantitatively evaluate vehicle
performance metrics. With further integration of AI into the framework,
feasible) [19]. Perhaps accurate acquisition and rigorous data logging
are the biggest challenges to implementing a sophisticated digital twin, the performance may not only be evaluated for the present scenario but
accurately predict the future performance metrics of the EV. This pre­
however, the results of this technology are commensurate in virtues.
sents future scope for intelligent manufacturing processes, which will

14
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

Fig. 13. Utilization of power electronic converters in electric vehicles.

Fig. 14. Elements of modern electric power drive systems (EPDS).

ameliorate the predicted bottlenecks to the life of the vehicle. By elim­ 5. Conclusions
inating future fault conditions presciently during the design phase,
digital twin technology has the potential to open the gates for durable, Beginning from the design to operation, digital twin technology
cost-effective, safer, and self-adaptive smart electric vehicles. opens the avenues for cost-effective and efficient development of sus­
Looking further into the future of technology, the prospect of digital tainable electric vehicle technologies. Digital twin technology is rela­
twin technology surpasses Industry 4.0 towards Industry 5.0 and 6.0. tively recent to the automotive industry, due to the lack of development
The era of Industry 5.0 would represent the integration of human in its sister technologies, such as IoT, delocalized wireless connectivity,
thought and speech into the control process of technology. To accom­ and artificial intelligence, during its conception. However, the scientific
plish this stage, the need for storing immense volumes of speech data for world has now entered a prime era for the development of digital twin
natural language processing is crucial. Cloud-based digital twin tech­ technology and other smart development methodologies. Furthermore,
nology is the ideal enabler as data can be stored and processed in a the coming decade will be a major turning point in human history due to
decentralized manner, reducing the quanta of storage technology unprecedented environmental challenges. Therefore, the primary aim of
needed on mobile systems like vehicles. Following this, civilisation en­ the research community must be to nurture sustainable technology and
ters into Industry 6.0, where design insights made by humans need to be its enablers. With this objective in mind, this review exhibits the
communicated instantly and artificial intelligence manages most tech­ application of digital twin technology in the automotive industry, dis­
nology. The role of digital twins becomes much more significant in this cussed with the backdrop of smart electric vehicles. In this paper, the
stage as creative vehicle designs can be presented and implemented on inception of digital twin technology with the different implementation
the digital twin as fast as thought itself, provided the necessary wireless stages are explicated. The study highlights the digital twin technologies
transfer capacity exists. Furthermore, it is digital twin technology that adapted for smart electric vehicle use cases, such as predictive mobility,
interlinks otherwise disparate elements, which can then be controlled as autonomous motion control, driver assistance systems, vehicle health
a single entity by AI. It is evident that digital twin technology shall be a management systems, battery management systems, intelligent
key tool even as civilisation and industry continue to evolve. charging, vehicle power electronic converters, and electric power drive
systems. Furthermore, the implementation’s challenges and key
enabling technologies are identified, thus aiding future work that takes

15
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

Table 6 [11] Zhao Junqiao, Ye Chen, Wu Yan, Guan Linting, Cai Lewen, Sun Lu, Yang Tao,
Overview of enabling technologies for electric power drive systems. He Xudong, Li Jun, Ding Yongchao, Zhang Xinglian, Wang Xinchen, Huang Jinlin,
Zhang Enwei, Huang Yewei, Jiang Wei, Zhang Shaoming, Xiong Lu,
Sl. Enabling Technology Role in Optimizing the EPDS. Year Feng Tiantian. TiEV: the tongji intelligent electric vehicle in the intelligent
No. [Ref. No.] vehicle future challenge of China. 2018.
[12] Sun B, He R, Deng W, Wu J, et al. Personalized eco-driving for intelligent electric
1. Motor drive integrated Improves regenerative energy 2007 [96] vehicles. SAE Technical; 2018. https://doi.org/10.4271/2018-01-1625. Paper
battery/ultra-capacitor recycling efficiency by using 2018-01-1625.
energy storage. novel drive modulation and [13] Zhang S, Zhang D, Luo Y, Wang J, Li K. Predictive energy management strategy
direct ultra-capacitor for electric vehicles based on estimation of preceding vehicle future movements.
connections. In: IEEE international conference on advanced intelligent mechatronics (AIM),
2. Systems integrated Provides improved power flow 2008 [97] banff, AB, 2016; 2016. p. 1436–41. https://doi.org/10.1109/AIM.2016.7576972.
supercapacitors and management and dynamic [14] Guo J, Luo Y, Hu C, et al. Robust combined lane keeping and direct yaw moment
batteries. performance through control for intelligent electric vehicles with time delay. IntJ Automot Technol
integrated drivetrain and 2019;20:289–96. https://doi.org/10.1007/s12239-019-0028-5.
[15] Li W, Chen T, Guo J, Wang J. Adaptive car-following control of intelligent electric
energy storage design.
vehicles. In: 2018 IEEE 4th international conference on control science and
3. Intelligent powertrain Provides real-time noise, 2020 [16]
systems engineering (ICCSSE), wuhan, China; 2018. p. 86–9. https://doi.org/
stabilization through vibration, and harshness
10.1109/CCSSE.2018.8724753.
passive and active suppression during EV motion [16] Qin Yechen, Tang Xiaolin, Tong Jia, Duan Ziwen, Zhang Jieming, Li Yinong,
methods. to provide seamless cruising. Zheng Ling. Noise and vibration suppression in hybrid electric vehicles: state of
4. Naturalistic data-driven Provides a computationally 2020 [17] the art and challenges. Renew Sustain Energy Rev 2020;124. https://doi.org/
extreme machine learning efficient method for on-board 10.1016/j.rser.2020.109782. 109782, ISSN 1364-0321.
algorithm. energy management in plug-in [17] X. Tang, T. Jia, X. Hu, Y. Huang, Z. Deng and H. Pu, "Naturalistic data-driven
hybrid electric vehicles. predictive energy management for plug-in hybrid electric vehicles," in IEEE
5. Variable voltage control of Optimizing performance and 2012 Trans. Transport. Electrif., doi: 10.1109/TTE.2020.3025352.
dc-dc bidirectional reducing net drive energy [103] [18] Yousefnezhad Narges, Malhi Avleen, Keyriläinen Tuomas, Huotari Matti,
converter. consumption by using an Främling Kary. Product lifecycle information management with digital twin: a
efficient drivetrain control case study. 2020.
strategy. [19] Rios J, Morate FM, Oliva M, Hernandez JC. Framework to support the aircraft
digital counterpart concept with an industrial design view. Int J Agile Syst Manag
6. Distributed drives Provides highly efficient drive 2018
2016;9(3).
utilization. demand management whilst [106]
[20] Grieves M. Digital twin: manufacturing excellence through virtual factory
curbing costs by reducing over- replication. White paper. 2014 [Online]. Available: http://www.apriso.com/li
dimensioning. brary/Whitepaper_Dr_Grieves _DigitalTwin_ManufacturingExcellence.php.
[21] Glaessgen E, Stargel D. The digital twin paradigm for future NASA and U.S. Air
Force vehicles. in Proc. 53rd AIAA/ASME/ASCE/AHS/ASC Struct. Struct. Dyn.
steps towards a future of cleaner and sustainable transportation. Mater. Conf. 2012 [Online]. Available: https://arc.aiaa.org/doi/pdf/10.2514/6.
2012-1818.
[22] Gabor T, Belzner L, Kiermeier M, Beck MT, Neitz A. A simulation-based
Declaration of competing interest architecture for smart cyber-physical systems. In proc. IEEE int. Conf. Autonomic
comput. Germany: Wurzburg; 2016. p. 374–9. https://doi.org/10.1109/
The authors declare that they have no known competing financial ICAC.2016.29.
[23] T. Maurer, “What is a digital twin?” 2017. [Online]. Available: https://
interests or personal relationships that could have appeared to influence community.plm.automation.siemens.com/t5/Digital-Twin-Knowledge- Base/
the work reported in this paper. What-is-a-digital-twin/ta-p/432960.
[24] Tuegel EJ. The airframe digital twin some challenges to realization. In proc. 53rd
AIAA/ASME/ASCE/AHS/ASC struct. Honolulu, HI, USA: . Struct. Dyn. Mater.
Acknowledgement Conf.; 2012. Art no. 1812.
[25] Schluse M, Rossmann J. From simulation to experimentable digital twins:
The authors thank Vellore Institute of Technology (VIT) Vellore, simulation-based development and operation of complex technical systems. In:
Proc. IEEE int. Symp. Syst. Eng., edinburgh, U.K.; 2016. p. 1–6. https://doi.org/
India for providing ‘VIT SEED GRANT’ for carrying out this work. 10.1109/SysEng.2016.7753162.
[26] Tao F, Zhang M. Digital twin shop-floor: a new shop-floor paradigm towards
References smart manufacturing. IEEE Access 2017;5:20418–27. https://doi.org/10.1109/
ACCESS.2017.2756069.
[27] Knapp GL, et al. Building blocks for a digital twin of additive manufacturing. Acta
[1] Glaessgen EH, Stargel D. The digital twin paradigm for future NASA and US air
Mater 2017;135:390–9. https://doi.org/10.1016/j.actamat.2017.06.039. Aug.
force vehicles. In: 53rd struct. Dyn. Mater. Conf. Special session: digital twin,
[28] Alam KM, El Saddik A. C2PS: a digital twin architecture reference model for the
honolulu, HI, US 1–14; 2012.
cloud-based cyber-physical systems. IEEE Access 2017;5:2050–62. https://doi.
[2] Tao Fei, Sui Fangyuan, Liu Ang, Qi Qinglin, Zhang Meng, Song Boyang,
org/10.1109/ACCESS.2017.2657006.
Guo Zirong, Lu Stephen, Nee Andrew. Digital twin-driven product design
[29] GE Look ahead, “The digital twin Could this be the 21st-century approach to
framework. Int J Prod Res 2018:1–19. https://doi.org/10.1080/
productivity enhancements?”. 2015 [Online]. Available: http://gelookahead.eco
00207543.2018.1443229.
nomist.com/the-digital-twin/.
[3] Yang Wenhe, Tan Yifei, Yoshida Kohtaroh, Takakuwa Soemon. Digital twin-
[30] AUCOTEC. 3 Industries being transformed by digital twins. 2017 [Online].
driven simulation for a cyber-physical system in industry 4.0 era. In: Katalinic B,
Available: http://news.aucotec.com/3-industries-transformeddigital-twins/.
editor. Chapter 18 in DAAAM international scientific book 2017. Vienna, Austria:
[31] Kaur Maninder, Mishra Ved P, Maheshwari Piyush. The convergence of digital
DAAAM International; 2017. p. 227–34. https://doi.org/10.2507/daaam.
twin, IoT, and machine learning: transforming data into action. 2020. https://doi.
scibook.2017.18. ISBN 978-3-902734-12-9, ISSN 1726-9687.
org/10.1007/978-3-030-18732-3_1.
[4] Wortmann F, Flüchter K. Internet of things. Bus. Inform. Syst. Engineering 2015;
[32] Bottani Eleonora, Assunta Cammardella, Murino Teresa, Vespoli Silvestro. From
57(3):221–4. https://doi.org/10.1007/s12599-015-0383-3.
the cyber-physical system to the digital twin: the process development for
[5] Nord Jeretta, Koohang Alex, Paliszkiewicz Joanna. The Internet of things: review
behaviour modelling of a cyber guided vehicle in M2M logic. 2017.
and theoretical framework. Expert Syst Appl 2019;133. https://doi.org/10.1016/
[33] Rassõlkin Anton, Vaimann Toomas, Kallaste Ants, Kuts Vladimir. Digital twin for
j.eswa.2019.05.014.
propulsion drive of autonomous electric vehicle. 2019. https://doi.org/10.1109/
[6] Asghari Parvaneh, Rahmani Amir, , Haj Seyyed Javadi, Hamid. Internet of things
RTUCON48111.2019.8982326.
applications: a systematic review. Comput Network 2018;148. https://doi.org/
[34] Ruba Mircea, Nemes Raul-Octavian, Ciornei Sorina, Martis Claudia,
10.1016/j.comnet.2018.12.008.
Bouscayrol Alain, Hedesiu Horia. Digital twin real-time FPGA implementation for
[7] Lund D, MacGillivray C, Turner V, Morales M. Worldwide and regional Internet of
light electric vehicle propulsion system using EMR organization. 2019. p. 1–6.
Things (IoT) 2014–2020 forecast: a virtuous circle of proven value and demand.
https://doi.org/10.1109/VPPC46532.2019.8952428.
Int. Data Corp. Tech. Rep. 2014;248451. Framingham, MA, USA.
[35] Guerra Haber, Elias Rodolfo, Quiza Ramon, Villalonga Alberto, Arenas Javier,
[8] Madni AM, Madni CC, Lucero SD. Leveraging digital twin technology in model-
Castaño Fernando. Digital twin-based optimization for ultraprecision motion
based systems engineering. Systems 2019;7:7.
systems with backlash and friction. IEEE Access; 2019. PP. 1-1. 10.1109/
[9] Kamath Vignesh, Morgan Jeff, Ali Muhammad Intizar. Industrial IoT and Digital
ACCESS.2019.2928141.
Twins for a Smart Factory: an open source toolkit for application design and
[36] Zhang Tianle, Liu Xiangtao, Luo Zongwei, Dong Fuqiang, Jiang Yu. Time series
benchmarking. 2020.
behavior modeling with digital twin for Internet of Vehicles. EURASIP J Wirel
[10] Sell R, Coatanéa E, Christophe F. Important aspects of early design in
Commun Netw 2019. https://doi.org/10.1186/s13638-019-1589-8. 2019.
mechatronic. In: 6th int. DAAAM balt. Conf. "Industrial eng.; 2008. no. April.

16
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

[37] Butakov VA, Ioannou P. Personalized driver/vehicle lane change models for [65] Shen WX, Chan CC, Lo EWC, Chau KT. Adaptive neuro-fuzzy modeling of battery
ADAS. IEEE Trans Veh Technol 2015;64(10):4422–31. https://doi.org/10.1109/ residual capacity for electric vehicles. IEEE Trans Ind Electron 2002;49:677–84.
TVT.2014.2369522. Oct. [66] Li Weihan, Kwiecien Monika, Badeda Julia, Jöst Dominik, Schulte Dominik,
[38] https://www.oxts.com/what-is-adas/. Sauer Dirk. Digital twin for battery systems: cloud battery management system
[39] Orlovska Julia, Wickman Casper, Söderberg Rikard. Design of a data-driven with online state-of-charge and state-of-health estimation. J Energy Storage 2020.
communication framework as personalized support for users of ADAS. Procedia 30. 101557. 10.1016/j.est.2020.101557.
CIRP 2020;91:121–6. https://doi.org/10.1016/j.procir.2020.02.156. [67] Ramaraj Ramachandran, Subathra B, Srinivasan Seshadhri. Recursive estimation
[40] Pérez S, Pérez J, Arroba P, Blanco R, Ayala JL, Moya JM. Predictive GPU-based of battery pack parameters in electric vehicles. 2018. p. 1–7. https://doi.org/
ADAS management in energy-conscious smart cities. IEEE international smart 10.1109/ICCIC.2018.8782329.
cities conference (ISC2), casablanca, Morocco. 2019. p. 349–54. https://doi.org/ [68] Sancarlos Abel, Cameron Morgan, Abel Andreas, Cueto Elias, Duval Jean-Louis,
10.1109/ISC246665.2019.9071685. 2019. Chinesta Francisco. From ROM of electrochemistry to AI-based battery digital and
[41] Rabinowitz Aaron, Gaikwad Tushar, White Samantha, Bradley Thomas, hybrid twin. Arch Comput Methods Eng 2020:1–37. https://doi.org/10.1007/
Asher Zachary. Synchronous and open, real world, vehicle, ADAS, and s11831-020-09404-6.
infrastructure data streams for automotive machine learning algorithms research. [69] Merkle Lukas, Segura Alexander, Grummel Jan, Lienkamp Markus. Architecture
2020. https://doi.org/10.4271/2020-01-0736. of a digital twin for enabling digital services for battery systems. 2019. p. 155–60.
[42] Liu Yongkang, Wang Ziran, Han Kyungtae, Shou Zhenyu, Tiwari Prashant, https://doi.org/10.1109/ICPHYS.2019.8780347.
Hansen John. Sensor fusion of camera and cloud digital twin information for [70] Wu Billy, Dhammika Widanage W, Yang Shichun, Liu Xinhua. Battery digital
intelligent vehicles. 2020. twins: perspectives on the fusion of models, data and artificial intelligence for
[43] Wang Ziran, Liao Xishun, Zhao Xuanpeng, Han Kyungtae, Tiwari Prashant, smart battery management systems. Energy AI 2020;1. https://doi.org/10.1016/
Barth Matthew, Wu Guoyuan. A digital twin paradigm: vehicle-to-cloud based j.egyai.2020.100016. 100016, ISSN 2666-5468.
advanced driver assistance systems. 2020. https://doi.org/10.1109/VTC2020- [71] Qu X, Song Y, Liu D, Cui X, Peng Y. Lithium-ion battery performance degradation
Spring48590.2020.9128938. evaluation in dynamic operating conditions based on a digital twin model.
[44] Alam Kazi Masudul, El Saddik, Abdulmotaleb. C2PS: a digital twin architecture Microelectron Reliab 2020;114. https://doi.org/10.1016/j.
reference model for the cloud-based cyber-physical systems. IEEE Access; 2017. microrel.2020.113857. 113857, ISSN 0026-2714.
PP. 1-1. 10.1109/ACCESS.2017.2657006. [72] Peng Y, Zhang X, Song Y, Liu D. A low cost flexible digital twin platform for
[45] Damjanovic-Behrendt Violeta. A digital twin-based privacy enhancement spacecraft lithium-ion battery pack degradation assessment. Auckland, New
mechanism for the automotive industry. 2018. p. 272–9. https://doi.org/ Zealand: IEEE International Instrumentation and Measurement Technology
10.1109/IS.2018.8710526. Conference (I2MTC); 2019. p. 1–6. https://doi.org/10.1109/
[46] Ezhilarasu CM, Skaf Z, Jennions IK. The application of reasoning to aerospace I2MTC.2019.8827160. 2019.
integrated vehicle health management (IVHM): challenges and opportunities. [73] Park Joonam, Kim Kyu, Oh Dae Yang, Jin Dahee, Kim Dohwan, Jung Yoon Seok,
Prog Aero Sci 2019;105:60–73. Lee Yong Min. Digital twin-driven all-solid-state battery: unraveling the physical
[47] Ezhilarasu, Mattuvarkuzhali Cordelia, Jennions IK, Skaf Zakwan. Understanding and electrochemical behaviors. Adv Energy Mater 2020;10. https://doi.org/
the role of a digital twin in the field of integrated vehicle health management 10.1002/aenm.202001563.
(IVHM). 2019. https://doi.org/10.1109/SMC.2019.8914244. [74] Shikata Hayato, Yamashita Tetsuo, Kouji ARAI, Nakano Takayuki,
[48] Tao Fei, Zhang Meng, Liu Yushan, Nee Andrew. Digital twin driven prognostics Hatanaka Kenichi, Fujikawa Hiroyuki. Digital twin environment to integrate
and health management for complex equipment. CIRP Annals 2018;67. https:// vehicle simulation and physical verification. Sei technical review • number 88.
doi.org/10.1016/j.cirp.2018.04.055. 2019. p. 18–21.
[49] Magargle Ryan, Johnson Lee, Mandloi Padmesh, Davoudabadi Peyman, [75] Korotunov Sergiy, Tabunshchyk Galyna, Okhmak Vyacheslav. Genetic algorithms
Kesarkar Omkar, Krishnaswamy Sivasubramani, Batteh John, Pitchaikani Anand. as an optimization approach for managing electric vehicles charging in the smart
A simulation-based digital twin for model-driven health monitoring and grid. CEUR Workshop Proceed 2020;2608.
predictive maintenance of an automotive braking system. 2017. p. 35–46. [76] Chan CC, Chau KT. An overview of power electronics in electric vehicles. IEEE
https://doi.org/10.3384/ecp1713235. Trans Ind Electron 1997;44:3–13.
[50] Manickavasagam K, Venkatesan Suchitra, Nagendran Vijayalakshmi, [77] Sadek Shady. Power electronic converter topologies used in electric vehicles.
Tengenkai Nikita. Health monitoring and prognosis of electric vehicle motor 2016.
using intelligent-digital twin. IET Electr Power Appl 2019;13. https://doi.org/ [78] Monzer Al Sakka, Van Mierlo Joeri, Gualous Hamid. DC/DC converters for
10.1049/iet-epa.2018.5732. electric vehicles, electric vehicles - modelling and simulations, seref soylu,
[51] Zakrajsek Andrew, Mall Shankar. The development and use of a digital twin IntechOpen. September 12th 2011, Available from: https://www.intechopen.com
model for tire touchdown health monitoring. 2017. https://doi.org/10.2514/ /books/electric-vehicles-modelling-and-simulations/dc-dc-converters-for-electric
6.2017-0863. -vehicles.
[52] Ali MU, Zafar A, Nengroo SH, Hussain S, Alvi MJ, Kim HJ. Towards a smarter [79] Hirota M, Baba T, Zheng X, Nii K, Ohashi S, Ariyoshi T, Fujikawa H. Development
battery management system for electric vehicle applications: a critical review of of new AC/DC converter for PHEVs and EVs. 2011. p. 68–72.
lithium-ion battery state of charge estimation. Energies 2019. [80] Milton Matthew, de la O, Castulo, Ginn Herbert, Benigni Andrea. Controller-
[53] Chen Z, Xia B, You C, Mi CC. A novel energy management method for series embeddable probabilistic real-time digital twins for power electronic converter
plugin hybrid electric vehicles. Appl Energy 2015;145:172–9. diagnostics. IEEE Trans Power Electron 2020. PP. 1-1. 10.1109/
[54] Shareef H, Islam MM, Mohamed A. A review of the stage-of-the-art charging TPEL.2020.2971775.
technologies, placement methodologies, and impacts of electric vehicles. Renew [81] Peng Yingzhou, Wang Huai. Application of digital twin concept in condition
Sustain Energy Rev 2016;64:403–20. monitoring for DC-DC converter. 2019. p. 2199–204. https://doi.org/10.1109/
[55] Hannan MA, Lipu MSH, Hussain A, Mohamed A. A review of lithium-ion battery ECCE.2019.8912199.
state of charge estimation and management system in electric vehicle [82] Peng Yingzhou, Zhao Shuai, Wang Huai. A digital twin based estimation method
applications: challenges and recommendations. Renew Sustain Energy Rev 2017. for health indicators of DC-DC converters. IEEE Trans Power Electron 2020.
[56] Manzetti S, Mariasiu F. Electric vehicle battery technologies: from present state to https://doi.org/10.1109/TPEL.2020.3009600. PP. 1-1.
future systems. Renew Sustain Energy Rev 2015;51:1004–12. [83] Biglarbegian Mehrdad. High frequency gan power converters digital twin. 2018.
[57] Yanhui Z, Wenji S, Shili L, Jie L, Ziping F. A critical review on state of charge of https://doi.org/10.13140/RG.2.2.12149.22249.
batteries. J Renew Sustain Energy 2013;5:1–11. [84] Jain Palak, Poon Jason, Singh Jai Prakash, Spanos Costas, Sanders Seth,
[58] Barcellona S, Grillo S, Piegari L. A simple battery model for ev range prediction: Panda Sanjib. A digital twin approach for fault diagnosis in distributed
theory and experimental validation. In: Proceedings of electrical systems for photovoltaic system. IEEE Trans Power Electron 2019. https://doi.org/10.1109/
aircraft, railway, ship propulsion and road vehicles & international transportation TPEL.2019.2911594. PP. 1-1.
electrification conference (ESARS-ITEC), toulouse, France; 2016. 2–4 November. [85] Marra F, Sacchetti D, Pedersen AB, Andersen PB, Larsen E. Implementation of an
[59] Campestrini C, Horsche MF, Zilberman I, Heil T, Zimmermann T, Jossen A. Electric Vehicle test bed controlled by a Virtual Power Plant for contributing to
Validation and benchmark methods for battery management system regulating power reserves. In: 2012 IEEE power and energy society general
functionalities: state of charge estimation algorithms. J Energy Storage 2016;7: meeting; 2012. p. 1–7.
38–51. [86] IKA Institut für Kraftfahrzeuge, “Test Facilities for Automotive Research and
[60] Cuma MU, Koroglu T. A comprehensive review on estimation strategies used in Development. Germany: ” Aachen; 2011.
hybrid and battery electric vehicles. Renew Sustain Energy Rev 2015;42:517–31. [87] Lee W, Sarlioglu B. Thermal analysis of lateral GaN HEMT devices for high power
[61] Truchot C, Dubarry M, Liaw BY. State-of-charge estimation and uncertainty for density integrated motor drives considering the effect of PCB layout and parasitic
lithium-ion battery strings. Appl Energy 2014;119:218–27. parameters. In: 2018 IEEE transportation electrification conference and expo
[62] He W, Williard N, Chen C, Pecht M. State of charge estimation for Li-ion batteries (ITEC); 2018. p. 471–6.
using neural network modeling and unscented Kalman filter-based error [88] de Lillo L, Ahmadi B, Empringham L, Johnson M, Espina J, Abebe R. Next
cancellation. Int J Electr Power Energy Syst 2014;62:783–91. generation integrated drive, ngid: a novel approach to thermal and electrical
[63] Chen Z, Fu Y, Mi CC. State of charge estimation of lithium-ion batteries in electric integration of high power density drives in automotive applications,. IEEE Energy
drive vehicles using extended Kalman filtering. IEEE Trans Veh Technol 2013;62: Conversion Congress and Exposition (ECCE); 2018. p. 1228–32. 2018.
1020–30. [89] Liu Jun, Wang Li-fang, Yang Jian, Liu Gui-dong. “Research of a novel flexible load
[64] Xu L, Wang J, Chen Q. Kalman filtering state of charge estimation for battery for electric vehicle test bench,” in 2010. Int Conf Comput Commun Technol Agric
management system based on a stochastic fuzzy neural network battery model. Eng 2010;1:223–6.
Energy Convers Manag 2012;53:33–9.

17
G. Bhatti et al. Renewable and Sustainable Energy Reviews 141 (2021) 110801

[90] Alcala I, Claudio A, Guerrero G. Test bench to emulate an electric vehicle through [100] Thounthong P. Control of a three-level boost converter based on a differential
equivalent inertia and machine dc. In: 2008 11th IEEE international power flatness approach for fuel cell vehicle applications. IEEE Trans Veh Technol 2012;
electronics congress; 2008. p. 198–203. 61(3):1467–72. Mar.
[91] Zhao Hui, Cheng Li, Zhang Guojiang. “Design of a versatile test bench for hybrid [101] International Standard, “IEC 60034-30-1:2014 | Rotating electrical machines -
electric vehicles,” in 2008 IEEE Vehicle Power and Propulsion Conference. 2008. Part 30-1: efficiency classes of line operated AC motors (IE code) 2014;50.
p. 1–4. [102] International Standard, “IEC 60034-2-1:2014 Rotating electrical machines - Part
[92] Lungoci C, Bouquain D, Miraoui A, Helerea E. Modular test bench for a hybrid 2-1: standard methods for determining losses and efficiency from tests (excluding
electric vehicle with multiples energy sources. In: 2008 11th international machines for traction vehicles). 2014.
conference on optimization of electrical and electronic equipment; 2008. [103] Estima JO, Marques Cardoso AJ, “. Efficiency analysis of drive train topologies
p. 299–306. applied to electric/hybrid vehicles. IEEE Trans Veh Technol 2012;61(3):1021–31.
[93] Khatun P, Bingham CM, Schofield N, Mellor PH. An experimental laboratory Mar.
bench setup to study electric vehicle antilock braking/traction systems and their [104] Graebener S, Tarnowski M, Goehlich D. “Commercial vehicle drive train
control. In: Proceedings IEEE 56th Vehicular Technology Conference. 3; 2002. technology and topology pre-selection,” in 2015 tenth international conference
p. 1490–4. on ecological vehicles and renewable energies. EVER); 2015. p. 1–5.
[94] Cheng Y, Joeri VM, Lataire P. Research and test platform for hybrid electric [105] Krasopoulos CT, Beniakar ME, Kladas AG. Multicriteria PM motor design based on
vehicle with the super capacitor based energy storage. In: 2019 IEEE 60th ANFIS evaluation of EV driving cycle efficiency. IEEE Trans Transp Electrif Jun.
international scientific conference on power and electrical engineering of riga 2018;4(2):525–35.
technical university (RTUCON) 2007 European conference on power electronics [106] Sharma N, Liu Y. Multiple electrical machines applied for high drive train
and applications; 2007. p. 1–10. efficiency. In: 2018 IEEE international conference on electrical systems for
[95] Schaltz E, Khaligh A, Rasmussen PO. Influence of battery/ultracapacitor energy- aircraft, railway, ship propulsion and road vehicles & international transportation
storage sizing on battery lifetime in a fuel cell hybrid electric vehicle. IEEE Trans electrification conference. ESARS-ITEC); 2018. p. 1–7.
Veh Technol 2009;58(8):3882–91. Oct. [107] Rodic M, Jezernik K, Trlep M. Use of dynamic emulation of mechanical loads in
[96] Lu S, Member S, Corzine KA, Member S, Ferdowsi M. A new battery/ the testing of electrical vehicle driveline control algorithms. In 2007 European
ultracapacitor energy storage system design and its motor drive integration for conference on power electronics and applications. 2007. p. 1–10.
hybrid electric vehicles. Vehicle Technol IEEE Trans 2007;56(4):1516–23. [108] Senanayaka JSL, Van Huynh K, Robbersmyr KG. Multiple classifiers and data
[97] Yoo H, Member S, Sul S, Park Y, Jeong J. System integration and power-flow fusion for robust diagnosis of gearbox mixed faults. IEEE Trans Ind Inform 2018.
management for a series hybrid electric vehicle using supercapacitors and pp. 1–1.
batteries. IEEE Trans Ind Appl 2008;44(1):108–14. [109] Sharma M, George JP. Digital twin in the automotive industry: driving physical-
[98] Drabek P, Streit L, Los M. The energy storage system with supercapacitor. In: digital convergence. 2018.
Proceedings of 14th international power electronics and motion control [110] Singh Sumit, Shehab Essam, Higgins Nigel, Fowler Kevin, Tomiyama Tetsuo,
conference EPE-PEMC 2010; 2010. pp. T9-39-T9- 43. Fowler Chris. Challenges of digital twin in high value manufacturing. 2018.
[99] Iannuzzi D. Improvement of the energy recovery of traction electrical drives using https://doi.org/10.4271/2018-01-1928.
supercapacitors. In: 2008 13th international power electronics and motion [111] Fuller A, Fan Z, Day C, Barlow C. Digital twin: enabling technologies, challenges
control conference; 2008. p. 1469–74. and open research. IEEE access, vol. 8; 2020. p. 108952–71. https://doi.org/
10.1109/ACCESS.2020.2998358.

18

You might also like