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

net/publication/325292161

Digital twin modeling method for CNC machine tool

Conference Paper · March 2018

DOI: 10.1109/ICNSC.2018.8361285

CITATIONS READS

56 915

4 authors, including:

Weichao Luo Tianliang Hu

Peng Cheng Laboratory Shandong University
13 PUBLICATIONS   630 CITATIONS    71 PUBLICATIONS   2,194 CITATIONS   

SEE PROFILE SEE PROFILE

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

Digital twin modeling method for CNC machine tool View project

Digital Twin in Industry View project

All content following this page was uploaded by Weichao Luo on 26 February 2021.

The user has requested enhancement of the downloaded file.

Digital Twin modeling method for CNC machine tool
Weichao Luo1,2. Tianliang Hu1,2*. Wendan Zhu1,2. Fei Tao3.
1 School of Mechanical Engineering, Shandong University, Jinan 250061, China
2 Key Laboratory of High Efficiency and Clean Mechanical Manufacture (Shandong University), Ministry of Education, Jinan, China
3 School of Automation Science and Electrical Engineering, Beihang University, Beijing 100191, China
Email: tlhu@sdu.edu.cn

Abstract—CNC machine tool (CNCMT) is the mother method is needed; a consistent model between design and
machine of industry, which plays an important role in the coming actual environment of machine tool should be established,
smart manufacturing. The intelligence of CNCMT has a big which needs the real-time and accurate data mapping method;
significance, which will enables its self-sensing, self-prediction an effective machine learning algorithm to mine the data
and self-maintenance without user concerns. In order to realize gathered from sensors and control system is also necessary.
the intelligence of CNCMT, a Digital Twin (DT) modeling
method for CNCMT is researched㸪including a multi-domain The rest of this paper is organized as follows. In Section 2,
unified modeling method, a mapping method and an autonomous the concept of DT and related researches are introduced.
strategy. This paper provides a demonstration of DT modeling Section 3 introduces the application scenario of DT for
method for CNCMT. CNCMT. Section 4 discusses the description, mapping and
intelligence of DT. Section 5 concludes this paper and points
Keywords—CNC machine tool (CNCMT); Digital Twin (DT); out the future works.
Smart manufacturing; Fault predict and diagnosis
II. RELATED WORKS
I. INTRODUCTION
Smart manufacturing is the fourth industrial revolution A. Modeling and simulation
driven by new technologies such as Internet of Things (IoT), DT is a complete virtual prototype of an entire system, a
big data, data mining, and machine learning[1]. Smart new era in modeling and simulation. At present, multi-domain
manufacturing and its application are essential to society, and it modeling and simulating methods consists of software
is the development trend of industry. CNC Machine tool interface, High Level Architecture (HLA) and UML (Unified
(CNCMT) is the mother machine of industry [2], which plays Modeling Language).
an important role in smart manufacturing. So the intelligence of
CNCMT is necessary to manufacturing. (1) Software interface is a modeling method based on the
corresponding interface of commercial software for
However, in traditional modelling process of CNCMT, interdisciplinary fields [11], such as ANSYS, ADAMS,
developers often design digital model based on design MATLAB/Simulink. The above software is very mature but
parameters in the software[3]. This redundant design pattern the interface compatibility between different software cannot
will cause a lot of design iterations. In maintenance stage, fault be ignored.
diagnosis and prediction of equipment have been drawing more
and more attention[4]. Many scholars have studied fault (2) HLA is an advanced simulation structure, and has many
diagnosis and prediction methods [5, 6]. Data-driven approach advantages in the standardization, openness, scalability, and
can obtain more effective and hidden information from the vast support for distributed simulation [12].
data, based on relevant machine-learning algorithm[7, 8]. For (3) UML method mainly focuses on the multi-physics
example, S Cho [9] proposed an intelligent tool breakage unified modeling language Modelica [13]. Modelica is an
detection system which used a support vector machine (SVM) equation modeling language put forward in 1997, based on a
learning algorithm, JS Zhang [10]realized the CNC fault variety of modeling language. Modelica language has many
diagnoses by RBF neural network algorithm and program. But, advantages in model reusability, simple and convenient
running or faults data always rely on user's feedback or manual modeling [14, 15].
records, which will inevitably result in the loss of accuracy and
instantaneity. That cannot realize objective and real-time B. DT concept
design improvement according to the CNCMT application.
With the development of IoT, big data, and machine
The important difference of smart manufacturing is learning, many companies and scholars have put forward the
interaction between cyber system and physical system. In DT concept. The DT concept is a new idea about CPS (Cyber-
recent years, the appearance of digital twin (DT) technology physical systems) model of physical assets.
provides an effective solution to integrate the physical world
and the information world. To enable the intelligence of Many scholars and companies have explored the
CNCMT, a DT of it is necessary, which acts as a complete application of DT. NASA studied the health maintenance and
virtual prototype of an entire system and a one-to-one mapping guarantee of aerospace aircraft with the guidance of DT [16,
relationship. Therefore, a multi-domain digital modeling 17]. Lee J proposed a unified 5-level architecture as a guideline

National Natural Science Foundation Of China(Grant No.51405270)

978-1-5386-5053-0/18/$31.00 ©2018 IEEE

for implementation of CPS [18]. Greyce N. Schroeder [19] model stores and analyzes the running status data, then
proposed using AutomationML to model attributes related to makes decision using machine learning algorithm.
DT and pointed out that this model is very useful for data
exchange between different systems connected with DT. Aitor • The connection layer is a DT mapping model, which
Moreno presented the process of constructing a DT for a sheet consists of a mapping model server, a mapping model
metal punching machine to support the interactive design of client, and a data-mapping dictionary. The DT mapping
optimal NC machining programs [20]. TaoFei put forward the model makes the DT interact with physical CNCMT
DT Workshop (Digital Twin Workshop) concept, which real-timely. The DT mapping model is designed as
consists of the DT Workshop system composition, operating interface to communication drivers, allowing a
mechanism, and implementation method [21]. Many leading standardized read and written access to data in
companies such as PTC, ANSYS, GE, and Siemens also CNCMT.
explored many applications under the guidance of DT concept.

III. APPLICATION SCENARIO OF DT FOR CNCMT

A. Characteristics of CNCMT

Fig.1 System composition of CNCMT

CNCMT is a set of machine tool equipped with computer
numerical control system, which can be considered as a typical
mechatronic equipment composed of mechanical, electrical and
hydraulic components. The basic components of CNCMT are
CNC system, servo system, functional unit, and the auxiliaries,
as shown in Fig.1. It consists of many interacting parts such as
a spindle, cutting tools, a servo feed system, a hydraulic
system, an electrical control system, a control panel and the
like. Therefore, a multi-domain digital model build method is Fig.2 DT modeling and application framework
needed. By using model-based diagnosis and prediction᧨error
will be found in one component or in other parts which IV. MODELING OF DT
influences that component if a measured or computed value is DT framework contains the physical space, the digital
outside an interval. space and the connection between them. This paper proposed a
building method of DT, which contains DT descriptive model,
B. DT modeling and application framework DT mapping model and DT intelligent model.
To realize the self-sensing, self-prediction and self-
maintenance of CNCMT, a DT framework is proposed. As A. DT descriptive model
shown in the Fig.2, the DT framework contains the physical After analyzing the structure and function of CNCMT, it is
space, the digital space and the connection between them. divided into different subsystems, components and parts. Using
• In the physical space, the device’s running status are object-oriented method, as shown in Fig.3, subsystem
gathered to control system by different type of sensors engineers can determine the structure and mathematics
such as temperature sensor, pressure sensor, velocity equations of CNCMT according to existing experience and
sensor and so on. parameters.
• In the digital space, the DT consists of DT descriptive Different parts connect through interface between them and
model and DT intelligent model. The main function of different subsystems connect through energy convertor. Using
descriptive model is to describe the geometric, physical this method, a parameterized, modular and mathematical DT
and electrical character of CNCMT. The DT intelligent

978-1-5386-5053-0/18/$31.00 ©2018 IEEE

descriptive model can be built in the multi-physics modeling parsing and transformation of the protocol by mapping model
and simulation environment. server, as shown in Fig.4. The solver and post-processor of DT
descriptive model can conduct the working condition loading
and system simulation of DT. Through statistical analysis of
running status data, the design parameter and working
condition can be verified and simulated.

C. DT intelligent model
CNCMT will incur different faults irregularly. It will cause
bad precision and affect the production if the troubleshooting is
not timely. Traditional fault diagnosis method mainly depends
on expert’s experience but it is very expensive and unstable.
In this research, a DT intelligent model is built. It mainly
contains expert system for DT with input from history
maintenance data and fault information as well as fault
solutions output. Expert system contains a knowledge base and
an inference engine. To acquire knowledge autonomously and
simulate people's thinking method, artificial neural network is
used to build expert system. Artificial neural network generates
a number of rules with relevant algorithm, and provides an
Fig.3 Multi-physical modeling of CNCMT based on UML effective method to acquire knowledge, which is very difficult
modelica in traditional method. When the CNCMT incurs failure,
diagnosis results can get from output of neural networks after
B. DT mapping model
inputting failure phenomenon᧨as shown in Fig.5.
Working conditions of CNCMT comprise kinematics,
dynamics, and thermodynamics parameters according to
different subjects. To make the DT descriptive model be alive,
the running data from various sensors and CNC controller
should be uploaded to DT digital space real-timely.

Fig.5 DT intelligent model

The DT intelligent model acts as a brain of CNCMT
comprises fault monitoring, fault analysis and decision-making
functions. Fault monitoring mainly monitors the important
parameters of system; fault analysis is aimed to locate the
Fig.4 DT mapping strategy source of fault and decision-making gives out prediction and
OPC (OLE for Process Control) is used today as standardized fault solution.
interface between automation systems. Recently, OPC
foundation proposed OPC UA (Unified Architecture) as a D. Case study
replacement for all existing COM-based specifications. OPC CNC milling machine (CNCMM) is an important kind of
UA enables a platform-independent interoperability standard CNCMT and has a wide application. Since the component
for moving data between the factory floor and the enterprise. failures seriously affect the production task, DT is necessary
Mapping method is designed according to the OPC UA for CNCMM to predict and diagnose the fault.
standards in this research. Referring to the OPC UA unified Ball screw is an important component of CNCMM, which
architecture, mapping model server is design to collect data transforms the force from motor and guarantees machining
from different sensors, and the data-mapping dictionary is accuracy. The lead screw and bearing of feed system always
designed after analyzing data type and physical meaning. The encounter failure, so it is worth studying the intelligent fault
sensors on the CNCMT connect with the DT model after

978-1-5386-5053-0/18/$31.00 ©2018 IEEE

prediction and diagnosis of them. In order to descript the [2] Schmidt R, Möhring M, Härting R C, et al. Industry 4.0 - Potentials for
essence of it, a multi-domain model is built using modelica in Creating Smart Products: Empirical Research Results: Bis 2015
International Conference on Business Information Systems, Lecture
MWorks software. Notes in Business Information Processing, 2015[C].
[3] The Digital Twin Paradigm for Future NASA and U.S. Air Force
Vehicles[J].
[4] Tao F, Wang Y, Zuo Y, et al. Internet of Things in product life-cycle
energy management[J]. Journal of Industrial Information Integration,
2016,1:26-39.
[5] Ding S X. Model-Based Fault Diagnosis Techniques - Design Schemes,
Algorithms, and Tools[J]. Ifac Papersonline, 2016,49(15):50-56.
[6] Lei Y, Lin J, Zuo M J, et al. Condition monitoring and fault diagnosis of
planetary gearboxes: A review[J]. Measurement, 2014,48(1):292-305.
[7] Li J, Tao F, Cheng Y, et al. Big Data in product lifecycle
management[J]. International Journal of Advanced Manufacturing
Technology, 2015,81(1-4):667-684.
[8] Cheng Y, Chen K, Sun H, et al. Data and Knowledge Mining with Big
Data towards Smart Production[J]. 2017.
[9] Cho S, Asfour S, Onar A, et al. Tool breakage detection using support
Fig.6 multi-domain model of x-axis vector machine learning in a milling process[J]. International Journal of
The mapping model is developed based on XML schema. Machine Tools & Manufacture, 2005,45(3):241-249.
In this paper, B-splines neural network is used for the fault [10] Zhang J S, Zhang Y H. Research of CNC Fault Diagnosis Based on RBF
predict model. Vibration sensors are installed on the two sides Neural Network[J]. Applied Mechanics & Materials, 2012,150:211-216.
of ball screw and temperature sensors are installed on the [11] Wang G G. Definition and Review of Virtual Prototyping.[J]. Journal of
supporting base. They are used to monitor components of ball Computing & Information Science in Engineering, 2002,2(3):232-236.
screw real-timely. The residual lifetime of ball screw can get [12] Pedrielli G, Scavardone P, Tolio T, et al. Simulation of Complex
after analyzing the temperature and vibration values. Through Manufacturing Systems via HLA-Based Infrastructure: IEEE Workshop
on Principles of Advanced and Distributed Simulation, 2011[C].
the application of DT, it gives out alarm and resolution timely,
[13] Elmqvist H, Mattsson S E, Otter M. Modelica --- An International Effort
the fault is recognized and troubleshot before serious failure To Design An Object-Oriented Modeling Language: Summer Computer
occurred. Simulation Conference, Reno, Nevada, July, 1998[C].
[14] Association M. Modelica, a Unified Object Oriented Language for
V. CONCLUSION AND FUTURE WORKS Physical Systems Modeling. Tutorial[J]. 2005.
[15] Modelica U, Gh U, Soest P A. Modeling and Digital Simulation of
In this paper, a modelling method of DT for CNCMT is Hydraulic Systems in Design and Engineering Education using
proposed. This paper gives a demonstration of DT application Modelica and HyLib[J]. Modelica Workshop, 2000.
scenario in CNC machine tool era. However, the research still [16] The Air Force Digital Thread_Digital Twin - Life Cycle Integration and
needs many works to fulfill the entire DT application. For Use of Computational and Experimental Knowledge[J].
future work, we intend to continue works, which conclude: (1) [17] Reengineering Aircraft Structural Life Prediction Using a Digital
applying the TensorFlow to construct the machine-learning Twin[J].
algorithm; (2) improving modelling consistency and accuracy [18] Lee J, Bagheri B, Kao H A. A Cyber-Physical Systems architecture for
according to experiment; (3) studying the smart thinking Industry 4.0-based manufacturing systems[J]. Manufacturing Letters,
2015,3:18-23.
ability, analysis and decision-making ability of modelica.
[19] Digital Twin Data Modeling with AutomationML and a Communication
Methodology for Data Exchange[J].
REFERENCES [20] Virtualisation process of a sheet metal punching machine within the
Industry 4.0 vision[J].
[1] Xu X W, Newman S T. Making CNC machine tools more open, [21] Fei T, Meng Z, Cheng J, et al. Digital twin workshop:a new paradigm
interoperable and intelligent ಧ a review of the technologies[J]. for future workshop[J]. Computer Integrated Manufacturing Systems,
Computers in Industry, 2006,57(2):141-152. 2017.

978-1-5386-5053-0/18/$31.00 ©2018 IEEE

View publication stats