See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323107887

A systematic development method for cyber-physical machine tools

Article  in  Journal of Manufacturing Systems · February 2018

DOI: 10.1016/j.jmsy.2018.02.001

CITATIONS READS

71 3,457

4 authors, including:

Chao Liu Ray Y. Zhong

The Hong Kong Polytechnic University The University of Hong Kong
32 PUBLICATIONS   1,178 CITATIONS    199 PUBLICATIONS   6,148 CITATIONS   

SEE PROFILE SEE PROFILE

Xun Xu
University of Auckland
426 PUBLICATIONS   13,128 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Machine Learning for Metal Additive Manufacturing View project

Developing a prototype for automated smart construction View project

All content following this page was uploaded by Chao Liu on 13 March 2018.

The user has requested enhancement of the downloaded file.

G Model
JMSY-638; No. of Pages 12 ARTICLE IN PRESS
Journal of Manufacturing Systems xxx (2018) xxx–xxx

Contents lists available at ScienceDirect

Journal of Manufacturing Systems

journal homepage: www.elsevier.com/locate/jmansys

A systematic development method for cyber-physical machine tools

Chao Liu, Hrishikesh Vengayil, Ray Y. Zhong, Xun Xu ∗
Department of Mechanical Engineering, University of Auckland, Auckland 1010, New Zealand

a r t i c l e i n f o a b s t r a c t

Article history: Machine Tool 4.0 introduces a new generation of machine tools that are smarter, well connected, widely
Received 30 November 2017 accessible, more adaptive and more autonomous. Cyber-Physical Machine Tools (CPMT), based on recent
Received in revised form 21 January 2018 advancements of the Information and Communication Technology, provides a promising solution for
Accepted 2 February 2018
Machine Tool 4.0. This paper proposes a systematic development method for CPMT. Generic system
Available online xxx
architecture is developed to provide guidelines for advancing existing Computer Numerical Control (CNC)
machine tools to CPMT. The proposed architecture allows machine tool, machining processes, real-time
Keywords:
machining data and intelligent algorithms to be deeply integrated through various types of networks. The
Machine tools
Machine tool 4.0
development methodologies for the core of the CPMT, the Machine Tool Cyber Twin (MTCT), are studied
Cyber-physical systems and discussed in detail. MTCT enables different types of feedback loops among the physical world, the
Cyber twin cyber space and humans to be realized. An MTConnect-based CPMT prototype is developed to validate the
MTConnect proposed CPMT. Experimental results have proved great interoperability, connectivity and extensibility
of the proposed CPMT. The potential of implementing artificial intelligence in CPMT is also discussed.
© 2018 Published by Elsevier Ltd on behalf of The Society of Manufacturing Engineers.

1. Introduction of machine tools which is currently in progress triggered by the

recent advancements in Cyber-Physical Systems (CPS), Internet of
Machine tools are machines that are powered to cut, shape or Things (IoT) and cloud-based solutions. In general, Machine Tool
finish metal or other rigid materials [1,2]. They are known as the 4.0 defines a new generation of machine tools that are smarter, well
“mother machines” since they enable the manufacture of other connected, widely accessible, more adaptive and more autonomous
machines which produce virtually everything used in daily life. In [5].
the realm of manufacturing, machine tools have always been play- Recent advancements in Information and Communication Tech-
ing a crucial role in boosting productivity ever since their advent. nology have attracted more and more attention in the domain
Their performances directly affect the product quality as well as the of manufacturing. As a result, research on Cyber-Physical Produc-
production efficiency. The evolutionary history of machine tools tion Systems (CPPS) [6], IoT-enabled manufacturing systems [7,8]
has significantly affected the history of industrialization. As shown and Cloud Manufacturing [9,10] have been extensively studied in
in Fig. 1, while industrialization can be briefly divided into four recent years. Although these topics differ from one another in terms
industrial revolutions, i.e. Industry 1.0 (mechanization, end of 18th of technological focus, implementation strategy, system frame-
century), Industry 2.0 (mass production, start of 20th century), work and so forth, a common trend in the development strategy
Industry 3.0 (automation and IT, start of 1970s) and Industry 4.0 for manufacturing systems has been clearly indicated – the deep
(Cyber-Physical Systems-based digitization, present time) [3]; the integration of the cyber/virtual world and the physical manufac-
technological evolution of machine tools up to now can be corre- turing devices. Previous research in this area mainly focused on
spondingly summarized as four stages, namely Machine Tool 1.0 the development of CPS-based manufacturing systems in terms of
(mechanically driven but manually operated, end of 18th century), the design principles [6,11], system frameworks [12] and imple-
Machine Tool 2.0 (electronically driven and numerically controlled, mentation strategies [13,14], few studies on the development of
middle of 20th century), Machine Tool 3.0 (computer numerically CPS-based machine tools have been done. Owing to the unique and
controlled, late 20th century) and Machine Tool 4.0 (Cyber-Physical essential role that machine tools are playing in any manufacturing
Machine Tools and cloud-based solutions, present time) [4]. The system, as we step into the new era of Industry 4.0, current machine
term “Machine Tool 4.0” stands for a new technological evolution tools must be advanced to a higher level of intelligence, autonomy
and connectivity, i.e. Machine Tool 4.0. Due to the lack of a sys-
tematic development method for CPS-based machine tools in the
new era of Machine Tool 4.0, Cyber-Physical Machine Tools (CPMT)
∗ Corresponding author.
is proposed to bridge the research gap [5]. CPMT refers to the
E-mail address: xun.xu@auckland.ac.nz (X. Xu).

https://doi.org/10.1016/j.jmsy.2018.02.001
0278-6125/© 2018 Published by Elsevier Ltd on behalf of The Society of Manufacturing Engineers.

Please cite this article in press as: Liu C, et al. A systematic development method for cyber-physical machine tools. J Manuf Syst (2018),
https://doi.org/10.1016/j.jmsy.2018.02.001
G Model
JMSY-638; No. of Pages 12 ARTICLE IN PRESS
2 C. Liu et al. / Journal of Manufacturing Systems xxx (2018) xxx–xxx

Fig. 1. Evolutionary history of industrialization and machine tools.

integration of the machine tool, machining processes, compu- machine tools more flexible, adaptable and intelligent has never
tation and networking, where embedded computations monitor stopped. The development and implementation strategies of intel-
and control the machining processes, with feedback loops in ligent CNC machining systems have been a topic if research for
which machining processes can affect computations and vice versa some time. Cheng et al. [17] developed an intelligent CNC sys-
[4]. With CPS technology deeply implemented, CPMT has dis- tem to improve the intelligence and communication ability of
tinct advantages in comparison with traditional machine tools. conventional NC machine tools under the distributed network
On one hand, embedded computations take full advantage of the manufacturing mode. The system comprised a tool monitoring
real-time manufacturing data, providing efficient decision-making module, a mechanical actuator error compensation module, an
support for humans, meanwhile endowing CPMT with advanced intelligent control module and a communication module. Aim-
intelligence and autonomy. Consequently, machining performance ing at improving the intelligence of machine tools, Kim et al.
and efficiency can be significantly improved. On the other hand, [18] proposed an agent-based knowledge evolution system in
ubiquitous networking and advanced connectivity enable the servi- the machine-to-machine environment. A sensory agent, a dia-
tization of CPMT, hence CPMT becomes product-service systems logue agent and a decision support agent were developed to
[15] as well as cloud manufacturing resources [16] that provide autonomously gather knowledge, produce knowledge and make
significant benefits for both the machine tool manufacturers and decisions during machining processes. Work-offset compensation
the users. and recommendation of cutting condition based on thermal change
This paper proposes a systematic development method for were performed on a tapping machine to verify the proposed
CPMT. Generic system architecture for CPMT is developed. The key system. In order to achieve the intelligent control of production
components and functions included in a typical CPMT are pro- processes, Zhang et al. [14] proposed a data-driven add-on CPS
posed and explained. The proposed CPMT architecture provides architecture of a CNC machine tool. Three key issues of the con-
guidelines for advancing existing CNC machine tools to CPMT. The figuration design of the add-on CPS system were studied, i.e., node
development methodologies for the Machine Tool Cyber Twin, the configuration of the add-on cyber-physical system, interconnection
core of a CPMT, are studied and discussed to provide directions for technology of the heterogeneous nodes and data-driven adap-
future research. An MTConnect-based CPMT prototype developed tive configuration method of the sensor networks. Ridwan and Xu
in our lab is introduced. The experimental results validated the fea- [19] developed an intelligent machine condition monitoring sys-
sibility and advantages of the proposed CPMT. The remainder of this tem that allows machining parameter optimization before, during
paper is organized as follows. Section 2 provides an overview of the and after machining to shorten machining time and increase prod-
state-of-the-art work related to the development of CPMT in both uct quality. Real-time cutting power, vibration and feed-rate data
academia and industry. Section 3 introduces the generic system collected using different sensors were streamed in MTConnect for-
architecture for CPMT. Section 4 studies and discusses the devel- mat and fuzzy logic was utilized to realize autonomous in-process
opment methodologies of each key component of the Machine Tool feed-rate optimization. Morgan and O’Donnell [20] presented the
Cyber Twin. Section 5 introduces an MTConnect-based CPMT pro- design and development of a reconfigurable, flexible and extensi-
totype developed in our lab. Section 6 concludes the paper and ble CPS monitoring system for machine tools. An advanced signal
provides directions for the future work. processing chain was developed to realize the semi-autonomous
process characterization of a CNC turning machine tool. Atluru
2. State of the art et al. [21] proposed a system framework for a smart machine
supervisory system which integrates individual process monitoring
Although there have been few studies on the generic develop- and control modules to realize a globally optimal machining solu-
ment methods for CPS-based machine tools, the effort to make tion. The communication mechanism of the system was based on

Please cite this article in press as: Liu C, et al. A systematic development method for cyber-physical machine tools. J Manuf Syst (2018),
https://doi.org/10.1016/j.jmsy.2018.02.001
G Model
JMSY-638; No. of Pages 12 ARTICLE IN PRESS
C. Liu et al. / Journal of Manufacturing Systems xxx (2018) xxx–xxx 3

MTConnect protocol. The decision-making mechanism of the sys- to improve the intelligence, connectivity, accessibility, adaptability
tem was demonstrated through a LabView application which and autonomy of CPMT.
integrates process planning, health monitoring and tool condi-
tion monitoring. Focused on data-sharing capabilities, Li et al. 3. Generic system architecture for CPMT
[22] developed an intelligent CNC machining system consisted
of a shared machining parameters database, an open and This section provides generic system architecture for CPMT
high-performance CNC system, an online monitoring module, a (Fig. 2). CPMT is a CPS-based machine tool featuring ubiquitous
parameters optimization module, an energy efficiency analysis connectivity, advanced intelligence and autonomy. Real-time man-
module and a machine tool health diagnosis platform. ufacturing data from the physical devices are collected by diverse
Practitioners from industry are also making significant efforts types of sensors and transferred into the cyber space via different
to improve the connectivity, the digitization and the intelligence networks. In the cyber space, a Machine Tool Cyber Twin (MTCT) is
of machine tools. Recently, STEP Tools, Inc. is developing a digital modelled by integrating the manufacturing data with an informa-
thread solution for CNC machining processes to keep the design, tion model, a database and different intelligent algorithms. MTCT
manufacturing, and inspection of a product connected around a is a digital abstraction of the machine tool which represents the
digital twin [23]. A 3D model-based machining simulator which characteristics and real-time status of the machine tool, monitors
fuses STEP (ISO 10303) models of the product, MTConnect sta- and controls the machine tool with built-in computation and intel-
tus of the machine tool and Quality Information Framework (QIF) ligence, and sends the shop-floor manufacturing data to different
metrology feedback was developed to build the digital twin of the Human-Machine Interfaces (HMIs) as well as the cloud to provide
product with integrated machining and measurement while a part efficient decision-making support for different users. In this way,
is being cut. Mazak Cooperation launched a Mazak SmartBox [24] various types of feedback loops among the physical world, the cyber
as a scalable, end-to-end Industrial Internet of Things platform that space and humans are established (indicated by colored arrows in
connects manufacturing equipment, including machines, software Fig. 2). Consequently, the performance and the efficiency of the
and other devices, to a factory’s network and allows the free flow machine tools can be significantly improved. The detailed func-
of information to management systems via MTConnect. The Mazak tions and the features of the proposed CPMT are explained in the
SmartBox provides intelligent data analytics and monitoring of CNC following subsections.
machine tools while considering security issues. Aiming at facilitat-
ing the digitization of machine tools, DMG MORI provides a series
of software solutions for machine tools such as Virtual Machine, 3.1. Components and functions
Service Agent and Messenger, to offer high-fidelity simulation and
estimation of machining processes, web-based real-time machine A CPMT comprises three main components, i.e. Physical devices,
monitoring through mobile devices and intelligent maintenance Networks and MTCT.
services based on real machining data [25]. Shenyang Machine Tool
Company developed an i5 smart numerical control system as an 1) Physical devices: It can be divided into two categories: machining
Internet-based intelligent terminal comprising intelligent control, devices and data acquisition devices. Machining devices include
intelligent operation, intelligent programming, intelligent mainte- each component of the CNC machine tool, the cutting tools and
nance and intelligent management [26]. Based on real-time data the workpieces; while data acquisition devices contain various
collection and cloud-based big data analytics, i5 enables remote types of sensors and measurement devices such as radio fre-
machine monitoring and workshop management services through quency identification (RFID) tags and readers, power meters,
mobile devices. dynamometers, accelerometers, Acoustic Emission (AE) sensors,
The state-of-art work in this area shows that both researchers and Coordinate Measuring Machines (CMM). During machining
and industrial practitioners are striving to implement the CPS processes, the CNC controller and the data acquisition devices
technology in machine tools. However, from the perspective of capture real-time machining data generated by the machining
academia, most of the previous research focuses on the develop- devices and send them to the cyber space for further analysis.
ment of specific intelligent functions for a machine tool, such as Real-time machining data acquisition is the prerequisite for all
CPS-based condition monitoring and prognostics, process planning, the subsequent functions in a CPMT.
adaptive control and machining parameter optimization. These 2) Networks: one of the distinct advantages of CPMT is the
functions are usually developed as standalone CPS, and hence advanced connectivity. In CPMT, various types of networks can
are difficult to integrate or cooperate with other functions of be implemented to achieve reliable, efficient and ubiquitous
the machine tool. Few studies have been attempted to develop a communications among the physical world, the cyber space and
machine tool as a complete CPS (or CPMT) in a systematic way. humans. Real-time feedback from the CNC controllers can be
From the perspective of industry, machine tool manufacturers retrieved through various industrial Ethernet, fieldbus and serial
are making efforts to integrate CPS-based intelligent functions in communication protocols such as Profinet, EtherCAT, Power-
their products. Nonetheless, most of the industrial solutions are link, RS-232 and RS-485. Data collected by various sensors can
vendor-dependent and require huge investments, presenting great be transmitted to the cyber space through different communi-
integration and interoperability challenges for modern manufac- cation protocols such as Ethernet, WiFi, ZigBee and Bluetooth,
turing systems which usually contain machine tools from different with the implementation of low-cost microcontrollers such as
vendors. Arduino and Raspberry Pi. Open communication standards such
To address these issues, this paper proposes a systematic as MTConnect and OPC Unified Architecture (OPC UA) allow
development method for CPMT. A generic, extensible and standard- the data to be uniformly transferred across the data acquisition
based system architecture is developed to provide guidelines for devices, data analysis applications and HMIs through the Inter-
advancing existing a CNC machine tool to a CPMT. The proposed net. Consequently, various types of feedback loops among the
method enables the machine tool, machining processes, real-time CPMT, the cloud and humans can be realized.
machining data and intelligent algorithms to be deeply inte- 3) Machine Tool Cyber Twin: The most significant difference
grated through standardized communication protocols. Machine between a CPMT and a traditional CNC machine tool lies in the
Tool Cyber Twin, as a digital model of the machine tool, is proposed MTCT. The MTCT comprises four main modules, including Data
fusion, Information model, Intelligent algorithms and Database.

Please cite this article in press as: Liu C, et al. A systematic development method for cyber-physical machine tools. J Manuf Syst (2018),
https://doi.org/10.1016/j.jmsy.2018.02.001
G Model
JMSY-638; No. of Pages 12 ARTICLE IN PRESS
4 C. Liu et al. / Journal of Manufacturing Systems xxx (2018) xxx–xxx

Fig. 2. Generic system architecture for CPMT.

• Data fusion: data fusion is a preliminary step to build the MTCT. categorizes the real-time data into different groups and feeds
Firstly, raw data generated by different sensors are cleansed and them into the embedded intelligent algorithms in the MTCT based
preprocessed so that meaningful information can be extracted. on specific data requirements. On the other hand, the database
Secondly, real-time machining data generated by CNC controller stores the timestamped real-time manufacturing data as histori-
and data acquisition devices are converted into a common format, cal data and provides them to the cloud for further data analytics
such that the data can be efficiently manipulated in the follow- and value-added services.
ing processes. Thirdly, data from different data sources (e.g. CNC
controller, RFID tags and various sensors) are grouped to their 3.2. Feedback loops
correlated component, in preparation for the following infor-
mation modelling. As a result, the data fusion module provides Feedback loops are an indispensable part of any CPS since they
unified, accurate and reliable data as the basis for modelling the complete the integration of the physical world, the cyber space
MTCT. and humans. The proposed CPMT enables three types of feedback
• Information model: The information model provides a compre- loops as indicated by the colored arrows in Fig. 2, i.e. autonomous
hensive understanding of the machine tool. It clearly presents the feedback control loop (blue arrows), shop-floor decision-making
logical structure of the machine tool including the relationships support loop (purple arrows) and cloud-based decision-making
between each sub-system and critical component on one hand. support loop (green arrows).
On the other hand, it groups all the available data items related to
a specific component together such that the real-time manufac- 1) Autonomous feedback control loop: in the autonomous feedback
turing data can be effectively manipulated and utilized in further control loop, the MTCT analyzes the real-time manufacturing
analysis. Since the information model is digitally developed, it data with embedded intelligent algorithms and sends con-
is easy to be modified or extended when existing physical com- trol commands directly to the CNC controller, such that the
ponents and data acquisition devices are replaced, or additional machine tool can autonomously adjust machining parameters
devices are implemented. during machining processes to adapt to the changing machin-
• Intelligent algorithms: various data visualization, process opti- ing conditions, while meeting specific requirements such as
mization and Prognostics and Health Management (PHM) surface quality, cutting tool wear and processing time. It is
algorithms can be embedded in the MTCT. These algorithms worth mentioning that open CNC controllers such as STEP-NC
retrieve specific real-time machining data directly from the infor- based controllers [27,28] play an important role in this con-
mation model, endow the machine tool with intelligent and text. With the implementation of STEP-NC technology, data
autonomous functions and provide efficient decision-making related to machine capability, process planning and various
supports for shop-floor technicians. The output of these embed- machining parameters can be optimized in STEP-NC files and
ded algorithms can be directly sent to the CNC controller as directly feedback to the STEP-NC controllers as control com-
control commands to realize autonomous in-process optimiza- mands [29–31]. Consequently, autonomous feedback control
tion, or fed back to machine operators to improve machine loop can be efficiently realized without developing additional
utilization and machining performances. interfaces between MTCT and CNC controllers.
• Database: since the CNC controller and the data acquisition 2) Shop-floor decision-making support loop: The integration of the
devices generate enormous amounts of real-time manufacturing real-time analytical information generated by the embedded
data during machining processes, it is necessary for the MTCT to intelligent algorithms and the advancements of data visualiza-
have a database that is able to safely, reliably and effectively store tion technologies such as AR give rise to various types of smart
and manage all the big industrial data. On one hand, the database HMIs (e.g. smart phones, tablets and wearable devices) that

Please cite this article in press as: Liu C, et al. A systematic development method for cyber-physical machine tools. J Manuf Syst (2018),
https://doi.org/10.1016/j.jmsy.2018.02.001
G Model
JMSY-638; No. of Pages 12 ARTICLE IN PRESS
C. Liu et al. / Journal of Manufacturing Systems xxx (2018) xxx–xxx 5

allow the shop-floor technicians (e.g. machine operators, main- Table 1

Static property data of manufacturing devices in CPMT.
tenance technicians and quality control technicians) to have an
instant, comprehensive and intuitive perception of the machin- Devices Static property data (examples)
ing processes, thus enabling them to make efficient decisions to Machine tool Manufacturer; Date of manufacture; Machine type; CNC
improve the machining performance as well as the equipment controller model; Dimension; Weight; Number of axes;
effectiveness. Working area; Axes travel distance; Clamping area;
3) Cloud-based decision-making support loop: during machining Maximum table load; Rapid motion speed; Maximum
spindle speed; Maximum torque; Tool turret capacity
processes, enormous amounts of field-level historical manu-
Cutting tools Manufacturer; Date of manufacture; Serial number; Tool
facturing data generated from each component and machining type; Material; Various geometries
process are transmitted to the MTCT. These big data will be Workpieces Material; Dimensions; CAD file; Process plan; Machining
further transferred to the cloud database for long-term archiv- program
ing. Various big data analytics algorithms and computational
resources can be utilized to provide cloud-based decision-
4. Development methodologies for machine tool cyber twin
making supports for the designers and planners. For example,
product designers and process planners can optimize the prod-
MTCT is the essential reason that distinguishes CPMT from a
uct designs and the process plans based on the data related to
traditional CNC machine tool. Modelling of MTCT is the process of
the machining performance; production planners can improve
digitizing the physical machine tool and machining processes in
the Overall Equipment Effectiveness based on the statistical data
the cyber space. This section reports on the development method-
related to the machine utilization rate; device providers can
ologies of the core modules in the MTCT, including manufacturing
also take advantage of the data provided by their customers to
data acquisition, data fusion, information modelling, database and
improve the design of their devices and provide value-added ser-
intelligent algorithms.
vices. From another point of view, the MTCT can also be treated
as a composite manufacturing service [32] that can be provided,
managed and consumed in the cloud manufacturing platform. 4.1. Manufacturing data acquisition

In order to provide comprehensive, reliable and accurate data of

3.3. Extensibility the manufacturing devices and the machining processes, the useful
data as well as their related data acquisition techniques must be
Machine tools are usually not reconfigurable or extensible once recognized. In general, the field-level manufacturing data in a CPMT
they have been assembled in the shop floor. However, the pro- can be divided into three categories, i.e. static property data, real-
posed CPMT architecture allows the machine tool to be extensible time machining data and measurement data.
in terms of the available real-time data, the information model, the
autonomous functions, and the decision-making supports. 4.1.1. Static property data
Firstly, different types of data acquisition devices can be Static property data represent the basic properties of the related
deployed on the machine tool to extend the available real-time physical devices. Table 1 lists some static property data of the
data based on specific needs of the users. Other than the feedback machine tool, the cutting tools and the workpieces. These data are
from the CNC controller, additional RFID tags and readers, cam- usually collected from the manuals of the devices. To efficiently
eras, accelerometers, dynamometers, AE sensors and so forth can transmit these data to the information model in the cyber space,
be implemented to collect real-time machining data related to spe- RFID technology can be utilized given the various advantages of
cific components and machining processes. Based on the unified RFID including the remote accessibility, durability, reusability, data
communication standards and different types of networks, these capacity, security, and so forth. Ideally, each machine tool, cutting
extended data can be transferred to the cyber space and appended tool and workpiece should be equipped with an RFID tag contain-
to their associated components as additional data items in the ing its static property data listed in Table 1. RFID readers can then
information model. Secondly, the autonomous functions can be read and transmit these data to the information model in the cyber
extended by embedding customized intelligent algorithms in the space.
MTCT. Various intelligent algorithms such as ANN [33,34], Fuzzy
Logic [35,36], Random Forests [37] and Support Vector Machine 4.1.2. Real-time machining data
(SVM) [38] can be embedded in the MTCT to retrieve multiple real- Real-time machining data represent the status of the machining
time data items directly from the information model, and endows processes from various aspects. Some of the real-time machining
the machine tool with autonomous in-process optimization, tool data have significant influences on the machining performance and
wear prediction, surface roughness prediction, and so forth. Fur- are commonly used as the input data for various data analytics
thermore, the advanced connectivity of CPMT allows diverse forms and process optimization algorithms, thus need to be collected and
of decision-making supports and value-added services to be fur- transmitted to the cyber space for further analysis. In the proposed
ther developed. For example, cutting tool providers can access the CPMT, real-time machining data can be obtained from the CNC con-
machining data to help the users optimize the tool path and cut- troller as well as various sensors that embedded in or attached
ting parameters based on their knowledge. Maintenance service to the machine tools. Table 2 lists the commonly used real-time
providers can access the prognosis data and provide remote main- machining data and the related data acquisition devices.
tenance guidance using AR technology. CNC controllers can usually provide some real-time feedback
Due to the generic and extensible characteristics, the proposed as listed in Table 2. However, owing to the diversity of existing
CPMT system architecture provides strategic guidelines for advanc- CNC controllers, retrieving real-time data from CNC controllers
ing traditional CNC machine tools to CPMT. From the perspective becomes a challenging task. For the open source CNC controllers
of machine tool design, the proposed architecture also provides such as LinuxCNC, it is easy to access the CNC data through
guidelines for the design of the next generation of machine tools pre-defined Application Programming Interfaces (APIs). However,
that thoroughly integrate machine tool, data acquisition devices, for the proprietary CNC controllers (e.g. FANUC, Siemens, Haas)
MTCT and networks from the initial design stage right down to the that predominate the CNC control market, specific adapters (spe-
shop floor deployment. cially designed hardware or software) have to be implemented for

Please cite this article in press as: Liu C, et al. A systematic development method for cyber-physical machine tools. J Manuf Syst (2018),
https://doi.org/10.1016/j.jmsy.2018.02.001
G Model
JMSY-638; No. of Pages 12 ARTICLE IN PRESS
6 C. Liu et al. / Journal of Manufacturing Systems xxx (2018) xxx–xxx

Table 2 Data formatting is responsible for translating all the manufac-

Real-time machining data and data acquisition devices in CPMT.
turing data obtained from the aforementioned data acquisition
Data acquisition device Real-time machining data (examples) devices into a unified data format, such that they can be efficiently
CNC controller Operation status; Alarms; Programs; managed and analyzed in the cyber space. Recently, MTConnect
Tool turret position; Spindle speed; and OPC UA have shown great potential for addressing this issue.
Spindle rotating direction; Feed-rate; MTConnect is a set of open, royalty-free standards designed for
Positions, speeds, accelerations of axes the exchange of data between shop floor equipment and software
Power meter; Wattmeter Motor power and current
applications, developed by the MTConnect Institution [43]. MTCon-
Dynamometer; Strain gauge Cutting forces and torques
AE sensor Acoustic emission nect transforms the data from various data acquisition devices into
Piezoelectric accelerometer; Vibration, chatter a common XML (Extensible Markup Language) format defined by
Vibrometer the standard and uses HTTP (Hypertext Transfer Protocol) as the
Thermocouple; Infrared thermometer Cutting zone temperature
underlying transport protocol. OPC UA is a set of open, royalty-
Cameras Machining process visualization
free standards designed as a universal factory floor communication
protocol developed by the OPC Foundation [44]. OPC UA is able to
Table 3 encode the field-level manufacturing data as the pre-defined OPC
Measurement data and data acquisition devices in CPMT. UA Binary messages or XML messages and deliver them over TCP/IP
Measurement Measurement data (examples) connections. Both two standards are open, royalty-free, platform-
device independent and extensible. In addition, an MTConnect-OPC UA
CMM; Laser Dimensional and geometric accuracy of workpieces Companion Specification has been released to ensure interoper-
scanner ability and consistency between them. Therefore, both of these
Profilometer; CCD Surface roughness, texture, and finish of workpieces two standards can be implemented as the data communication
camera
standard in the proposed CPMT.
Optical sensor; Cutting tool wear
Camera With various data acquisition devices implemented in the CPMT,
the data generated by each data source may be related to multi-
ple components. On the other hand, data generated by different
specific CNC controllers such that the internal real-time data can data acquisition devices may be related to a same component. This
be accessed. It is worth mentioning that more and more recently complexity makes the data manipulation a challenging task. Data
developed CNC controllers have been embedded with their own grouping addresses this issue by assigning each data item with a
adapters that support open communication standards such as unique ID, and grouping the data from different data sources to the
MTConnect and OPC UA. certain component that they are related to. Consequently, the sta-
The implementation strategies of different sensors and signal tus of each component can be comprehensively expressed in the
processing techniques for machining processes have been exten- information model.
sively studied and they are not the focus of this research, thus the
detailed sensing techniques are not discussed in this paper. Some 4.3. Information modelling
comprehensive reviews of the related techniques can be found in
[39–41]. Information model is one of the cores in MTCT. The information
model of CPMT should have three fundamental capabilities. First,
4.1.3. Measurement data it should represent the structure of the machine tool and all the
Measurement data refer to the measurement results obtained critical components in a logical manner. Second, it should contain
from various measurement devices that reflect the actual machin- both the static properties and the dynamic status of each critical
ing performance. Table 3 lists the commonly used measurement component. Third, the information model must be extensible such
data in CPMT. The accuracy and surface integrity of workpieces that additional components and data can be easily integrated. Both
provide the basic requirements for various optimization processes MTConnect and OPC UA provide a feasible information modelling
such as product design optimization, process planning optimiza- method that meets the requirements.
tion and machining parameter optimization. The cutting tool wear
data have significant influence on the cutting tool condition and the 4.3.1. MTConnect-based machine tool information model
machining performance. These measurement data can be obtained MTConnect defines a hierarchical XML data model for machine
during or after the machining processes, based on different user tools. It has two primary types of XML elements, i.e. structural ele-
requirements. In addition, these measurement data should also be ments and data elements [45]. A structural element represents a
stored in the RFID tags attached to the workpieces and cutting tools, physical or logical component. The static properties of that com-
such that they can be utilized by different algorithms in the cyber ponent are expressed by the attributes of the structural element.
space. Each structural element may have some sub-elements based on
their relationships. The data elements refer to all the real-time
4.2. Data fusion data items that can be collected from that component. MTConnect
defines a comprehensive dictionary of the important components
The data fusion module contains four main functions, i.e. data and data items especially for CNC machine tools, thus users only
cleansing, data pre-processing, data formatting and data grouping. need to choose the appropriate attributes (data types, units, etc.)
Fig. 3 shows the functions and the data flows in the data fusion for their specific information model. The XML schema that defines
module. the data model can also be extended to add custom components
Data cleansing and pre-processing ensure the raw data obtained and data based on specific needs [46]. Fig. 4 shows the structure of
by various sensors are reliable, accurate and meaningful. These two an MTConnect-based information model of a simple turning lathe.
functions can be achieved using various low-cost microcontroller As shown in the figure, the Machine tool has three sub-components
boards such as Arduino and Raspberry Pi. Specific signal process- including Controller, Axes and Cutting tool. Axes is an abstract com-
ing algorithms need to be implemented for specific sensor data. ponent that comprises three sub-components including Linear X,
An example of the implementation of data cleansing algorithm in Linear Y and Spindle. Each component has its own attributes and
CPMT was introduced in one of our previous research [42]. data items, representing its static properties and real-time data

Please cite this article in press as: Liu C, et al. A systematic development method for cyber-physical machine tools. J Manuf Syst (2018),
https://doi.org/10.1016/j.jmsy.2018.02.001
G Model
JMSY-638; No. of Pages 12 ARTICLE IN PRESS
C. Liu et al. / Journal of Manufacturing Systems xxx (2018) xxx–xxx 7

Fig. 3. Functions and data flows in the data fusion module.

Fig. 4. Structure of an MTConnect-based information model of a simple turning lathe.

respectively. Take the sub-component Spindle as an example; it 4.4. Database and intelligent algorithms
represents the spindle of the turning lathe. The static property of
the spindle is indicated by its maximum speed, which is obtained Having all the static property data and real-time machining data
from the specification of the machine tool and manually input to represented by the information model, a database must be devel-
the information model. The real-time status of the spindle is rep- oped in the MTCT such that the big data can be safely, reliably
resented by the speed and rotating direction obtained from the and effectively stored, managed, as well as retrieved by external
CNC controller, the motor power obtained from the power meter, intelligent algorithms and data analytics tools. Since the informa-
and the vibration obtained from the accelerometer. In this way, the tion model has already assigned each data item with a unique ID
MTConnect-based information model comprehensively represents and a timestamp, an efficient strategy for developing the database
the status of the entire machine tool. is to divide the database into different categories in accordance
with the structure of the information model. Each category repre-
4.3.2. OPC UA-based machine tool information model sents a component in the information model, containing the static
OPC UA, compared with MTConnect, provides a more generic properties and all the real-time data. Based on the open and royalty-
information modelling method. OPC UA defines an object-based free data communication standards applied in the MTCT, various
address space model as a Meta model, based on which vari- database developing tools such as MySQL, PostgreSQL and SQLite
ous domain-specific information models can be developed. The can be utilized to develop the database. Fig. 6 shows a generic
objects, similar to the concept of Object-Oriented Programming, structure of the proposed database. Based on this structure, a com-
are placeholders of variables, events and methods. Objects and their prehensive performance history of each component and machining
components are represented in the address space as a set of nodes process of the CPMT can be archived. Consequently, various data
described by attributes and interconnected by references [47]. Con- analytics applications in the cloud can retrieve specific historical
sequently, in order to develop a machine tool information model data items in specific time periods.
using OPC UA, the developer needs to build the structure of the Intelligent algorithms embedded in the MTCT make the CPMT
information model and define all the objects, attributes and ref- intelligent and autonomous. The proposed MTCT builds a solid
erences within it according to the actual structure, components foundation for the implementation of artificial intelligence in
and available data of the machine tool. Fig. 5 shows the struc- machine tools. In the MTCT, various types of embedded intelligent
ture of an OPC UA-based information model of a simple turning algorithms can access the real-time data directly from the infor-
lathe. Each component is modelled as an object and may have sub- mation model and extract specific data items to provide real-time
components and property as its child nodes. Each object may have monitor and control functions, such as in-process machining opti-
some data items modelled as its attributes with different data types. mization, high-fidelity simulation, data visualization and PHM. For
It is obvious that OPC UA provides developers with more flexibilities example, Augmented Reality (AR)-based algorithms can be embed-
to build the machine tool information model, but in the meantime, ded to retrieve the real-time machining video, axes positions and
relatively more effort and time are required. speeds, and perform AR-assisted monitoring and simulation [48].

Please cite this article in press as: Liu C, et al. A systematic development method for cyber-physical machine tools. J Manuf Syst (2018),
https://doi.org/10.1016/j.jmsy.2018.02.001
G Model
JMSY-638; No. of Pages 12 ARTICLE IN PRESS
8 C. Liu et al. / Journal of Manufacturing Systems xxx (2018) xxx–xxx

Fig. 5. Structure of an OPC UA-based information model of a simple turning lathe.

Fig. 6. Structure of the database in MTCT.

Artificial Neural Network (ANN) algorithms can be embedded to RFID tags. A Kistler 3-axis dynamometer (type 9273) was mounted
retrieve the cutting speed, feed and depth of cut, and perform tool on the work table to measure the cutting forces in X, Y and Z direc-
wear prediction [33]. Fuzzy inference systems can be embedded to tion. A PCB Piezoelectric accelerometer (model 352C65) was used to
retrieve the spindle speed, feed-rate and depth of cut, and perform measure the vibration of the spindle. An RPM sensor was mounted
surface roughness prediction [49]. The development methodolo- above the spindle to measure the spindle speed.
gies of the intelligent algorithms will not be discussed in detail The overall system architecture of the CPMT prototype is
since it is beyond the scope of this paper. described in Fig. 8. The machine tool is controlled by LinuxCNC
software installed in a Linux based computer. The static property
5. MTConnect- based CPMT prototype data of the machine tool and cutting tool in the RFID tags are read
by the Arduino microcontroller, and transferred to the MTConnect
An MTConnect-based CPMT prototype was developed to val- Adapter through serial connection. The sensor signals from the
idate the feasibility and performance of the proposed CPMT dynamometer, the accelerometer and the RPM sensor are trans-
architecture and development methodologies. MTConnect was ferred to a NI PXI-1031 data acquisition platform and processed
utilized as the communication standard and the information mod- using LabView software. The processed data are then sent to the
elling technology. MTConnect Adapter through TCP/IP connection. Real-time machin-
ing data from the CNC controller are retrieved from the LinuxCNC
software to the MTConnect Adapter directly through APIs since Lin-
5.1. Experimental setup
uxCNC is open sourced and they were installed in the same Linux
computer. The MTConnect Adapter transforms all the received data
The experimental setup of the MTConnect-based CPMT proto-
into SHDR (Simple Hierarchical Data Representation) format with
type is shown in Fig. 7. The prototype was built based on a Sherline
timestamps attached in front of each changed data, and keeps send-
3-Axis Milling machine. The machine tool and a test cutting tool
ing these data to the MTConnect Agent through TCP/IP connection.
were both assigned with a 13.56 MHz RFID tag containing their
The MTConnect Agent, running on a Windows computer, will then
static property information. An Arduino Uno development board
group all the data into the information model, format the data into
with an NFC shield connected was used to read the data from the

Please cite this article in press as: Liu C, et al. A systematic development method for cyber-physical machine tools. J Manuf Syst (2018),
https://doi.org/10.1016/j.jmsy.2018.02.001
G Model
JMSY-638; No. of Pages 12 ARTICLE IN PRESS
C. Liu et al. / Journal of Manufacturing Systems xxx (2018) xxx–xxx 9

Fig. 7. Experimental setup of the MTConnect-based CPMT prototype.

Fig. 8. System architecture of the MTConnect-based CPMT prototype.

XML format and publish them to various applications through the 5.3. CPMT-Cyber twin application
Internet.
A CPMT-Cyber Twin application was developed as the MTCT
of the MTConnect-based CPMT prototype. Fig. 10 shows the User
5.2. Information modelling Interface of the application. It comprises four main modules as
marked in Fig. 10. The application connects to the MTConnect Agent
The MTConnect-based information model for this machine tool through the Internet by specifying the IP address and port number
was developed as a Device.xml file based on the XML schema pro- of the Agent. Once connected, the application reads the Device.xml
vided by MTConnect standard. Fig. 9 shows the structure and all the file and displays the logical structure of the machine tool as a tree
components involved in the information model. All the data items structure in module 1. The tree structure clearly shows the logical
are collapsed due to space limitation. The detailed data items of relationships between each component. All the available data items
the spindle, the CNC controller and the cutting tool are expanded are shown under the component to which they are associated. The
on the right to illustrate how the data from different sources were name and data type of each data item are also displayed in the tree
grouped. As shown in Fig. 9, the spindle speed and vibration data structure. Consequently, this structure provides a comprehensive
collected from different sensors, the execution status information understanding of the static attributes of the machine tool.
retrieved from the LinuxCNC software, and the static property data Module 2 shows the real-time status of each component and
of the cutting tool obtained from the RFID tags were grouped as data machining process. All the available real-time data items are dis-
items under the spindle component, controller component and cut- played and continuously updated in the text boxes under their
ting tool component respectively. Each component and data item related component. The Machine Tool components show the basic
were assigned with a unique ID such that each component and data attributes of the machine tool and the time stamp of all the real-
item can be retrieved and manipulated separately. time data. The Controller component shows the real-time status

Please cite this article in press as: Liu C, et al. A systematic development method for cyber-physical machine tools. J Manuf Syst (2018),
https://doi.org/10.1016/j.jmsy.2018.02.001
G Model
JMSY-638; No. of Pages 12 ARTICLE IN PRESS
10 C. Liu et al. / Journal of Manufacturing Systems xxx (2018) xxx–xxx

Fig. 9. MTConnect-based XML Information model of Sherline 3-axis milling machine.

Fig. 10. User Interface of the CPMT-Cyber Twin application.

of the controller, including the current power status (ON or OFF), Module 3, at the bottom of the interface, provides data visu-
execution status (Active, Interrupted or Ready), controller mode alization and analytics functions based on the real-time data. As
(Automatic or Manual), NC program (file path and name), current shown in Fig. 10, under the Machine Status tab, the historical exe-
line of the CN program, feed rate and feed rate override. The Axes cution status of the machine tool is displayed as a line graph, where
component shows the commanded and actual positions of each the X-axis indicates the time and the Y-axis indicates the execu-
linear axis and the speed and vibration of the spindle. The cutting tion status, i.e. Active (machining), Interrupted (paused) or Ready
forces and cutting tool information are also displayed. The tick-box (stopped). The machine utilization rate is calculated and shown
before the data item indicates that it can be chosen and plotted as as a pie chart accordingly. The current application allows all the
a line graph in module 3. real-time data to be directly visualized as line graphs in the time

Please cite this article in press as: Liu C, et al. A systematic development method for cyber-physical machine tools. J Manuf Syst (2018),
https://doi.org/10.1016/j.jmsy.2018.02.001
G Model
JMSY-638; No. of Pages 12 ARTICLE IN PRESS
C. Liu et al. / Journal of Manufacturing Systems xxx (2018) xxx–xxx 11

Fig. 11. Data visualization of cutting forces and spindle vibration.

domain. Fig. 11 shows another two examples of the visualization It is worth mentioning that there are several limitations of the
of cutting forces and spindle vibration. These data visualization current prototype. First, MTConnect is a read-only standard. The
functions help the machine operators to identify severe changes feedback loops cannot be realized directly through MTConnect.
of machining conditions during machining processes and make Additional interfaces need to be developed to control the machine
quick decisions to prevent further damages. In effect, these data can tool. Second, since the amount of the real-time data is enormous,
be further analyzed in frequency and discrete time domain using the data transmission rate from the MTConnect agent to the appli-
wavelet transform and SVM, to predict and avoid chatter [50]. cation in the current prototype is limited to 100 ms. This could
Module 4 displays the historical data of a specific data item be improved by optimizing the application program or utilizing
selected in the machine tool structure. As shown in Fig. 10, the higher bandwidth networks. Third, the cutting tool is modelled
spindle vibration is selected in the machine tool structure, and the as a sensor component containing some basic data items in the
historical vibration data are displayed as a spreadsheet in mod- information model. To realize a comprehensive cutting tool infor-
ule 4. The first column contains the timestamps of each value. The mation model, the MTConnect Assets model which is compatible
second column shows the values of the data. The third column with ISO 13399 (Cutting tool data representation and exchange) can
indicates the sequence numbers of each value that are defined in be implemented. Last, the current CPMT-Cyber Twin application
MTConnect standard. These historical data can be directly saved as only provides some basic data visualization and analytics functions.
spreadsheets locally or transferred to the cloud database through Advanced intelligent algorithms need to be embedded to improve
the Internet. Various machine learning algorithms can directly use the intelligence of the prototype.
these data as training data or test data and provide the machine
tool with self-learning capabilities.
This application represents the MTCT of the MTConnect-based 6. Conclusions and future work
CPMT prototype. It allows users to remotely monitor the real-
time status of the machine tool, its critical components and the Machine Tool 4.0 introduces a new generation of machine tools
machining processes through the Internet. Based on this applica- that are smarter, well connected, widely accessible, more adap-
tion, machine operators can make efficient decisions to improve tive and more autonomous. CPMT, based on the advancements of
the machining performance; maintenance technicians can per- CPS, IoT and cloud technology, provides a promising solution for
form proactive maintenance to improve the machine utilization Machine Tool 4.0. In this paper, a systematic development method
rate; production planners and managers can remotely monitor the for CPMT is proposed to provide guidelines for advancing current
machine status in real time and optimize the production plans; var- CNC machine tools to CPMT. Generic system architecture for CPMT
ious service providers can access the historical data in the cloud to is developed. Based on this architecture, machine tool, machin-
provide value-added services using their own applications. ing processes, real-time machining data and intelligent algorithms
are integrated through diverse types of networks. MTCT is pro-
posed as the digital model of the physical machine tool in the cyber
5.4. Discussion space. It monitors and controls the machine tool with embedded
intelligence, and enables various types of feedback loops among
The MTConnect-based CPMT prototype validated the feasibil- the physical world, the cyber space and humans. Development
ity of the proposed CPMT system architecture. Various real-time methodologies for the key modules in MTCT including manufac-
manufacturing data were collected from the CNC controller, RFID turing data acquisition, data fusion and information modelling of
tags and different sensors. Different types of networks were utilized machine tools are proposed and discussed.
to transmit the data to the cyber space. Data fusion was realized The feasibility and advantages of the proposed CPMT architec-
based on the implementation of MTConnect standard across dif- ture and development methodologies are demonstrated through
ferent operation systems. An MTConnect-based information model the development of an MTConnect-based CPMT prototype. Static
was developed to represent the logical structure of the machine tool property data and real-time machining data were collected from
as well as the real-time status of each component and machining the CNC controller and different sensors such as RFID tags,
process. The experimental results shown in the CPMT-Cyber Twin accelerometer, dynamometer and RPM sensor. MTConnect was uti-
application have proved that the MTConnect-based CPMT proto- lized to realize unified and efficient data communication across
type has great potential of improving machining performances and different networks and operation systems. A CPMT-Cyber Twin
increasing machine tool utilization. It not only provides the basic application was developed to demonstrate the advancements of
requirements for implementing artificial intelligence in machine the MTCT. Experimental results validated that MTCT represents a
tools, but also enables various forms of feedback loops proposed in comprehensive digital model of the machine tool, including the
Section 3.2 to be realized. MTConnect has been proved to be capa- structure of the machine tool and the real-time status of each com-
ble of providing CPMT with great interoperability, connectivity and ponent and machining process. Data visualization and analytics
extensibility. functions developed in the application have indicated the great

Please cite this article in press as: Liu C, et al. A systematic development method for cyber-physical machine tools. J Manuf Syst (2018),
https://doi.org/10.1016/j.jmsy.2018.02.001
G Model
JMSY-638; No. of Pages 12 ARTICLE IN PRESS
12 C. Liu et al. / Journal of Manufacturing Systems xxx (2018) xxx–xxx

potential of implementing artificial intelligence in CPMT by embed- [18] Kim DH, Song JY, Lee JH, Cha SK. DevelopMent and evaluation of intelligent
ding various intelligent algorithms in the MTCT. Furthermore, the Machine tools based on knowledge evolution in M2 M environMent. J Mech
Sci Technol 2009;23:2807–13.
advanced connectivity of MTCT allows the real-time data and his- [19] Ridwan F, Xu X. Advanced CNC system with in-process feed-rate
torical data to be accessed by different users through the Internet, optimisation. Robot Comput Integr Manuf 2013;29:12–20.
thus enabling the servitization of the machine tool. [20] Morgan J, O’Donnell GE. Cyber physical process monitoring systems. J Intell
Manuf 2015:1–12.
Future work is summarized as follows. Firstly, our main focus [21] Atluru S, Huang SH, Snyder JP. A smart machine supervisory system
will be on the implementation of artificial intelligence in CPMT. framework. Int J Adv Manuf Technol 2012;58:563–72.
Various data visualization and analytics algorithms will be embed- [22] Li Y, Liu Q, Xiong J, Wang J. Research on data-sharing and intelligent CNC
machining system, 2015. 2015. p. 625–30.
ded into MTCT to take full advantage of the real-time machining
[23] STEP Tools Inc. Digital thread for manufacturing; 2017 https://www.
data. Secondly, mobile applications for the developed CPMT pro- steptools.com/sln/thread/.
totype will be developed on different HMIs such as smart phones [24] Mazak Corporation. Mazak SmartBox; 2017 https://www.mazakusa.com/
machines/technology/digital-solutions/mazak-smartbox/.
and wearable devices, to enhance the human-machine interaction
[25] DMG MORI CO. LTD. DMG MORI software; 2017 https://us. dmgmori. com/
between different users and CPMT. Thirdly, an OPC UA-based CPMT products#Software.
prototype is under development in our lab. A universal CPMT plat- [26] SHENYANG MACHINE TOOL CO. LTD. i5 smart numeric control; 2017 http://
form that integrates these two CPMT prototypes will be introduced www.i5cnc.cn/pages/i5. html.
[27] Wang H, Xu X, Des Tedford J. An adaptable CNC system based on STEP-NC and
in our future work. Furthermore, the integration of CPMT and cloud function blocks. Int J Prod Res 2007;45:3809–29.
manufacturing platform will be investigated in the future. [28] Rauch M, Laguionie R, Hascoet JY, Suh SH. An advanced STEP-NC controller for
Although the development method proposed in this paper intelligent machining processes. Robot Comput Integr Manuf
2012;28:375–84.
focuses on machine tools, the basic philosophy also applies to other [29] Ridwan F, Xu X, Liu G. A framework for machining optimisation based on
manufacturing equipment such as industrial robots and 3D print- STEP-NC. J Intell Manuf 2012;23:423–41.
ers. In the era of Industry 4.0, manufacturing equipment is expected [30] Um J, Suh S-H, Stroud I. STEP-NC machine tool data model and its
applications. Int J Comput Integr Manuf 2016;29:1058–74.
to become CPS-based systems, such that autonomy and Machine- [31] Xu XW, Newman ST. Making CNC machine tools more open, interoperable
to-Machine communications can be realized through their Cyber and intelligent—a review of the technologies. Comput Ind 2006;57:141–52.
Twins, thus the envisioned CPPS and smart factories can be even- [32] b Li B-H, Zhang L, Wang S-L, Tao F, Cao J-W, Jiang X-D, et al. Cloud
manufacturing: a new service-oriented networked manufacturing model.
tually realized.
Jisuanji Jicheng Zhizao Xitong/Computer Integr Manuf Syst CIMS 2010;16,
1-7+16.
References [33] Palanisamy P, Rajendran I, Shanmugasundaram S. Prediction of tool wear
using regression and ANN models in end-milling operation. Int J Adv Manuf
Technol 2008;37:29–41.
[1] Rosenberg N. Technological change in the machine tool industry, 1840–1910.
[34] Zain AM, Haron H, Sharif S. Prediction of surface roughness in the end milling
J Econ Hist 1963;23:414–43.
machining using Artificial Neural Network. Expert Syst Appl
[2] Geoffrey B, Winston AK. Fundamentals of Machining and Machine Tools;
2010;37:1755–68.
1989.
[35] Aliustaoglu C, Ertunc HM, Ocak H. Tool wear condition monitoring using a
[3] Kagermann H, Wahlster W, Helbig J. Securing the future of German
sensor fusion model based on fuzzy inference system. Mech Syst Signal
manufacturing industry: recommendations for implementing the strategic
Process 2009;23:539–46.
initiative INDUSTRIE 4.0. Final Rep Ind 2013;40:1–84. Work Gr.
[36] Lo S-P. An adaptive-network based fuzzy inference system for prediction of
[4] Xu X. Machine Tool 4.0 for the new era of manufacturing. Int J Adv Manuf
workpiece surface roughness in end milling. J Mater Process Technol
Technol 2017;92:1893–900.
2003;142:665–75.
[5] Liu C, Xu X. Cyber-physical machine tool – the era of machine tool 4.0.
[37] Wu D, Jennings C, Terpenny J, Gao RX, Kumara S. A comparative study on
Procedia CIRP 2017;63:70–5.
machine learning algorithms for smart manufacturing: tool wear prediction
[6] Monostori L. Cyber-physical production systems: roots, expectations and R&D
using random forests. J Manuf Sci Eng 2017;139:71018.
challenges. Procedia CIRP 2014;17:9–13.
[38] Widodo A, Yang B-S. Support vector machine in machine condition
[7] Zuehlke D. SmartFactory-towards a factory-of-things. Annu Rev Control
monitoring and fault diagnosis. Mech Syst Signal Process 2007;21:2560–74.
2010;34:129–38.
[39] Liang SY, Hecker RL, Landers RG. Machining process monitoring and control:
[8] Tao F, Cheng Y, Da Xu L, Zhang L, CCIoT-CMfg Li BH. Cloud computing and
the state-of-the-art. J Manuf Sci Eng 2004;126:297.
internet of things-based cloud manufacturing service system. IEEE Trans Ind
[40] Teti R, Jemielniak K, O’Donnell G, Dornfeld D. Advanced monitoring of
Informatics 2014;10:1435–42.
machining operations. CIRP Ann – Manuf Technol 2010;59:717–39.
[9] Xu X. From cloud computing to cloud manufacturing. Robot Comput Integr
[41] Abellan-Nebot JV, Subirón FR. A review of machining monitoring systems
Manuf 2012;28:75–86.
based on artificial intelligence process models. Int J Adv Manuf Technol
[10] Wu D, Greer MJ, Rosen DW, Schaefer D. Cloud manufacturing: strategic vision
2009;47:237–57.
and state-of-the-art. J Manuf Syst 2013;32:564–79.
[42] Deng C, Guo R, Liu C, Zhong RY, Xu X. Data cleansing for energy-saving: a case
[11] Monostori L, Kádár B, Bauernhansl T, Kondoh S, Kumara S, Reinhart G, et al.
of cyber-physical machine tools health monitoring system. Int J Prod Res
Cyber-physical systems in manufacturing. CIRP Ann – Manuf Technol
2017;7543:1–16.
2016;65:621–41.
[43] MTConnect Institution. MTConnect; 2008 http://www.mtconnect.org/.
[12] Lee J, Bagheri B, Kao HA. A cyber-physical systems architecture for industry
[44] Foundation OPC. OPC unified architecture; 2006 https://opcfoundation.org/
4.0-based manufacturing systems. Manuf Lett 2015;3:18–23.
about/opc-technologies/opc-ua/.
[13] Wang L, Torngren M, Onori M. Current status and advancement of ®
[45] Sobel W. MTConnect standard part 2–device information model; 2014.
cyber-physical systems in manufacturing. J Manuf Syst 2015;37:517–27.
[46] Lei P, Zheng L, Wang L, Wang Y, Li C, Li X. MTConnect compliant monitoring
[14] Zhang C, Jiang P, Cheng K, Xu XW, Ma Y. Configuration design of the add-on
for finishing assembly interfaces of large-scale components: a vertical tail
cyber-physical system with CNC machine tools and its application
section application. J Manuf Syst 2017;45:121–34.
perspectives. Procedia CIRP 2016;56:360–5.
[47] Mahnke W, Leitner SH, Damm M. OPC unified architecture; 2009.
[15] Isaksson O, Larsson TC, Ronnback AO. Development of product-service
[48] Liu C, Cao S, Tse W, Xu X. Augmented reality-assisted intelligent window for
systems: challenges and opportunities for the manufacturing firm. J Eng Des
cyber-physical machine tools. J Manuf Syst 2017;44:280–6.
2009;20:329–48.
[49] Ho W-H, Tsai J-T, Lin B-T, Chou J-H. Adaptive network-based fuzzy inference
[16] Lu Y, Xu X, Xu J. Development of a hybrid manufacturing cloud. J Manuf Syst
system for prediction of surface roughness in end milling process using hybrid
2014;33:551–66.
Taguchi-genetic learning algorithm. Expert Syst Appl 2009;36:3216–22.
[17] Cheng T, Zhang J, Hu C, Wu B, Yang S. Intelligent machine tools in a
[50] Yao Z, Mei D, Chen Z. On-line chatter detection and identification based on
distributed network manufacturing mode environment. Int J Adv Manuf
wavelet and support vector machine. J Mater Process Technol
Technol 2001;17:221–32.
2010;210:713–9.

Please cite this article in press as: Liu C, et al. A systematic development method for cyber-physical machine tools. J Manuf Syst (2018),
https://doi.org/10.1016/j.jmsy.2018.02.001

View publication stats