You are on page 1of 27

Enterprise Information Systems

ISSN: 1751-7575 (Print) 1751-7583 (Online) Journal homepage: https://www.tandfonline.com/loi/teis20

The Internet of Things enabled manufacturing


enterprise information system design and shop
floor dynamic scheduling optimisation

Songling Tian, Taiyong Wang, Lei Zhang & Xiaoqiang Wu

To cite this article: Songling Tian, Taiyong Wang, Lei Zhang & Xiaoqiang Wu (2019): The Internet
of Things enabled manufacturing enterprise information system design and shop floor dynamic
scheduling optimisation, Enterprise Information Systems, DOI: 10.1080/17517575.2019.1609703

To link to this article: https://doi.org/10.1080/17517575.2019.1609703

Published online: 05 May 2019.

Submit your article to this journal

Article views: 56

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=teis20
ENTERPRISE INFORMATION SYSTEMS
https://doi.org/10.1080/17517575.2019.1609703

The Internet of Things enabled manufacturing enterprise


information system design and shop floor dynamic
scheduling optimisation
Songling Tiana,b, Taiyong Wanga,b, Lei Zhangc and Xiaoqiang Wua
a
School of Mechanical Engineering, Tianjin University, Tianjin, P. R. China; bKey Laboratory of Mechanism
Theory and Equipment Design, MOE, Tianjin University, Tianjin, P.R. China; cSchool of Mechanical
Engineering, Tianjin University of Commerce, Tianjin, P. R. China

ABSTRACT ARTICLE HISTORY


A production information management system based on industrial Received 29 December 2018
Internet of Things (IoT) technology is designed, which makes the Accepted 16 April 2019
manufactory enterprise have the ability of dynamic shop floor KEYWORDS
scheduling. To solve the rescheduling problem, a hybrid method IoT; dynamic flexible job
combining the game theory method and the rolling horizon shop scheduling problem;
rescheduling strategy (HGR) is proposed. In a computational negotiation mechanism;
experiment, the values of the relative error (RE) and mean ideal game theory; rolling horizon
distance (MID) are 11.3 and 13.74, respectively, which illustrates rescheduling strategy
the feasibility and efficiency of the proposed HGR method. An
engineering experiment shows that the proposed method can
handle the practical engineering problems.

1. Introduction
Production information management is the most important part of an enterprise with
manufacturing as its core content. Shop floor scheduling and control execution are the
main manifestation of production information management in manufacturing enter-
prises (Zhong et al. 2013) (Cochran et al. 2017). Although the shop floor does not directly
produce economic benefits, it guarantees the implementation of enterprise goals and
plans through specific production activities and ensures the survival, operation and
development of the enterprise. Therefore, the operation of the shop floor plays
a decisive role in the development of the manufacturing enterprise (Bitam, Zeadally,
and Mellouk 2018).
In recent years, the business model of manufacturing enterprises gradually move
towards the characteristics of multi-varieties and small batches, which will require the
development of manufacturing system towards automation, high flexibility and effi-
ciency (Cochran et al. 2017). At present, manufacturing enterprises lack real-time collec-
tion of shop floor production field information and production control layer cannot
provide information support for enterprise management decision-making, resulting in
“Information Fault“ and ”Information Island” (Liu et al. 2000).

CONTACT Taiyong Wang tywang@tju.edu.cn School of Mechanical Engineering, Tianjin University, No.135
Yaguan Road, Haihe Education Park, Tianjin, P. R. China; Key Laboratory of Mechanism Theory and Equipment Design,
MOE, Tianjin University, Tianjin, P.R. China
© 2019 Informa UK Limited, trading as Taylor & Francis Group
2 S. TIAN ET AL.

Traditionally, in the manufacturing process of shop floor, the physical and information
flow of shop floor production field intersects with each other, and the information flow
is opaque. Shop floor scheduling system cannot obtain real-time production process
information, which makes the production process forms the phenomenon of
“Information Island”. It is difficult to control and manage the production process in
real time. Moreover, many uncertainties affect the normal operation of production plans,
such as machine failure, shortage of inventory, delivery changes, urgent orders, etc.
(Ouelhadj and Petrovic 2009), which will disrupt the pace of production. Therefore,
production information management of manufacturing enterprises is an important
means to avoid delays in production planning, improve production performance and
meet the diverse needs of customers. In order to cope with the change of demand on
the shop floor production field, people carry out repeated production reforms from the
past. Nowadays, the emergence of the Internet of Things (IoT) has attracted more
attention (Verdouw et al. 2015).
The motivation of this paper is to design a production information management
system based on IoT technology, using sensors, RFID, industrial wireless communica-
tion, automatic identification and other technologies to collect data of shop floor. On
the basis of supervising the material status and manufacturing equipment, process
guidance, resource allocation and execution control are carried out for the shop floor
processing tasks, and finally, a standard production information management system
is formed. At the same time, the production information management system also
includes the big data module, which is responsible for the processing, storage,
analysis and application of the big data in the shop floor production, and provides
data support for the realisation of the shop floor scheduling and execution control
functions. The application of information management system can reasonably assign
and manage the resources of orders, materials and equipment, which makes the
manufactory enterprise have the ability of dynamic shop floor scheduling (Sartal
et al. 2017).
The main contributions of this paper include the following:

(1) A production information management system based on industrial Internet of


Things technology is designed, which enables seamless information exchange
and networked interactions of physical and digital objects.
(2) The rescheduling method formulates a multi-objective FJSP with random machine
breakdowns using an extended game model.
(3) To obtain a more feasible solution, we present a dynamic rescheduling strategy
combining the rolling horizon rescheduling strategy and game theory method.
(4) To obtain an optimal schedule and minimise the changes to the affected opera-
tions in the rescheduling, we use backward induction to obtain a Nash equili-
brium solution.

This paper is organised as follows. In Section 2, we present the related literature reviews;
In Section 3, we describe the perception and fusion of shop floor production information
based on Internet of Things; Section 4 presents rescheduling in a dynamic flexible job
shop with random machine breakdown; Section 5 explains the rolling horizon resche-
duling strategy and dynamic scheduling strategy; in Section 6, we present an extensive
ENTERPRISE INFORMATION SYSTEMS 3

computational study using several benchmark problems, comparing our results with the
state-of-the-art algorithms and conduct experiments on the experimental platform and
present an applied example of a locomotive axle manufacturing enterprise. Some final
remarks and future research directions are given in Section 7.

2 Related literature reviews


This section introduces the background of relevant research. We review the big data-
based information management and shop floor scheduling methods.

2.1. Big data based information management and shop floor scheduling for IoT
enabled manufacturing
The implementation of business processes through the use of information systems (i.e.
product lifecycle management (PLM), customer relationship management (CRM), enter-
prise resource planning (ERP), office automation (OA), etc.) has become a key success
factor for companies (Glaschke and Gronau 2016) (Liu et al. 2018). More and more
manufacturing enterprises begin to pay more attention to the information management
and implementation of manufacturing shop floor. Shop floor is the source of the
ultimate interests of manufacturing enterprises and plays an important role in the
whole enterprise. In the past, the real-time information management in the shop floor
production field has not been given due attention (Sangaiah et al. 2018).
With the development trend of economic globalisation, manufacturing cluster and
information, the IoT technology and its application in manufacturing industry appeal the
attention of scientists and engineers, which is one of the updating area of computer
technology, information engineering, manufacturing information engineering, network
and communication (Bi 2017) (Liu, Liu, and Zhou 2018). One of the main research
focuses on real-time acquisition and management of manufacturing resources and
products, and collaborative application according to the requirements of management
information system (Răileanu et al. 2018). On the basis of system integration technology,
information fusion processing system and big data analysis, the development of the IoT
enabled manufacturing service platform and application system is another area to focus
on (Vishwasrao and Sangaiah 2017).
Perception recognition is the core of IoT enabled manufacturing, and therefore, the
technology of perception recognition is the key. Perception recognition mainly collects
information automatically by means of RFID, networked intelligent sensor equipment,
etc. (Zhang et al. 2016). At the same time, it also collects data by means of intelligent
terminal of man-machine exchange. At present, the manufacturing of sensor technol-
ogy in the IoT enabled manufacturing is developing towards the direction of intelli-
gence and wireless. The main new technologies include radio frequency identification,
wireless sensor network technology, positioning technology and embedded technol-
ogy (Wang et al. 2014). The perceived data are pre-processed by embedded intelligent
processing system and transmitted to the sensor network. With the development of
the perception and recognition technology of the IoT enabled manufacturing, it has
realised the comprehensive, accurate, real-time perception and supervisory of the
4 S. TIAN ET AL.

physical world information (Jiang, Chen, and Duan 2016) (Medhane and Sangaiah
2017).
In summary, most of the current research on the IoT enabled manufacturing mainly
focuses on the theoretical design level of system architecture and control model. In
practical application, it mainly applies to the information collection and supervisory of
manufacturing process, while the application of scheduling and controlling of manu-
facturing execution process is seldom involved. It is still in the stage of research and
exploration.

2.2. Shop floor scheduling methods


The scheduling problem in low-volume, high-variety production system environ-
ments with alternative routing, often referred to in the literature as the flexible job
shop scheduling problem (FJSP), is a generalisation of the traditional job shop
scheduling problem (JSP) in which there are no alternative machines (Shalaby,
Abdelmaguid, and Abdelrasol 2012). The JSP is demonstrated to be an NP-hard
problem, as is the FJSP.
Most previous research has focused on addressing the FJSP under deterministic
conditions, assuming that the parameters are known and that no interruption occurs
(J. Zhang, Yang, and Zhou 2016). However, in practice, scheduling often confronts
abnormal events such as machine breakdown, urgent order arrival, over- or under-
estimation of the processing time, order cancellation, changes to the due date and
lagging behind or being ahead of schedule (Tao, Xiao, and Hao 2007). The inclusion of
such interruptions in the traditional FJSP makes the scheduling problem more practical
but also more complex and challenging to solve (Hasan et al. 2011). Thus, the FJSP
scheduling environment can be classified into two main classes, namely, static or
deterministic (offline) scheduling and stochastic or dynamic (online) scheduling (Al-
Hinai and Elmekkawy 2011).
Two primary and distinct strategies are available to address the interruption effects
on scheduling, namely, proactive scheduling and reactive scheduling (Akturk and
Gorgulu 1999). The aim of proactive scheduling is to build robust scheduling that is
insensitive to interruptions and protected as much as possible against interruption
during scheduling execution (Lou et al. 2012).
Contrary to proactive scheduling strategies, in reactive scheduling, an operational
solution is suggested to compensate for the interruption effects when an unanticipated
event occurs. Rescheduling is a key link in the reactive scheduling strategy whose goal is
to minimise or eliminate the destructive effects caused by abnormal events while
maintaining the stability of the initial schedule (Paprocka and Skołud 2017).
Rescheduling strategies can be divided into two types: the invariable-order rescheduling
strategy and the variable-order rescheduling strategy. The former refers to maintaining
the original scheduling sequence of all operations or only changing the directly or
indirectly affected operations and includes the right-shift rescheduling (RSR) strategy
and the affected operation rescheduling (AOR) strategy, and the latter refers to the
rescheduling of all unprocessed operations after the interruption, i.e. the total resche-
duling (TR) strategy (Fahmy, Balakrishnan, and Elmekkawy 2009).
ENTERPRISE INFORMATION SYSTEMS 5

Both classical criteria such as the makespan and tardiness and performance
measures such as robustness and stability should be considered in predictive-
reactive scheduling (S. S. Liu and Shih 2009) (Nouiri et al. 2017). Generally, the
robustness and stability indexes cannot reach their optimums at the same time, so
the problem of the AOR rescheduling strategy implementation is a dynamic FJSP
problem involving multi-objective optimisation (S. Liu, Liu, and Zhou 2018). The
ultimate goal in solving multi-objective optimisation problems is to balance the
optimisation objectives.
There are many limitations in solving discrete variable optimisation problems such
as job shop scheduling: the large quantity of calculations, difficulty in unifying the
dimensions of the objective function, large numbers of experiments needed to
obtain the weight values, and independent information of the solution process
(Shukla 2007).
Considering these limitations of the traditional multi-objective optimisation methods,
the multi-objective game optimisation method has emerged in recent years. It does not
require that the weight of each objective be determined artificially and can shorten the
optimisation time and improve the convergence speed and computational efficiency (Ji,
Li, and Qu 2018).
In recent years, the game theory method has been used more and more in solving
discrete multi-objective optimisation problems, such as multi-objective optimisation in
multi-parameter engineering design. However, in the field of shop floor scheduling, its
use is still very limited.

2.3. Sub-game Nash equilibrium and backward induction


A general technique to solving a sequential game with perfect information Nash
equilibrium is backward induction, i.e. incorporating the solution of later stages in
previous ones (Heloulou, Radjef, and Kechadi 2017a).
Backward induction is an important method to solve dynamic game, but it has
obvious limitations, which is also a hot topic in the academic circles, such as centipede
game paradox (Ehtamo 2003). Some experimental results indicate that from the applica-
tion scope, it should be avoided to use backward induction in a longer stage of dynamic
game (Heloulou, Radjef, and Kechadi 2017b). Some studies have also pointed out that
the effectiveness of backward induction is affected by a variety of irrational factors. In
short, as long as the analysis of the problem meets its conditions and requirements, the
reverse induction method is still an effective way to analyse the dynamic game (Mak,
Rapoport, and Seale 2014).

3. Perception and fusion of shop floor production information based on


internet of things
Considering the disturbance events occurring in the production process (represented
by machine breakdown), this section designs a framework of dynamic scheduling
and execution method based on the IoT. This method includes perception and fusion
of shop floor production information and adaptive dynamic scheduling and control
6 S. TIAN ET AL.

of shop floor. The transparency of production process can be maximised by building


an RFID-based integrated manufacturing environment in the shop floor.

3.1. The configuration of the internet of manufacturing thing environment


With the help of IoT technology, we can build an internet of Manufacturing Things
environment (IoMT) environment for shop floor. The shop floor system under IoMT
environment can be divided into three basic levels: application layer, network transport
layer, and perceptive and execution layer, as shown in Figure 1. Autonomous analysis of
system status and real-time response to dynamic events are the main characteristics of
shop floor production under IoMT environment. IoMT environment is used to capture
real-time information of shop floor production.
The perceptive and execution layer belongs to the bottom of the whole shop floor
system. The commonly used hardware devices are all kinds of sensors and RFID devices.
RFID is the key technology in IoMT environment. It establishes a unique identity for each
job in the system mainly through electronic tags. It can also contain the production and
processing information of each job. The manufacturing equipment acquires the specific
information of jobs in real time through the RFID reader.
In network transport layer, communication is the basis of information exchange in
IoMT environment, and also an important means to realise dynamic scheduling and
execution. Wireless network communication is the mainstream communication mode of
the IoT in the future.
The main function of application layer is to provide users of the system (including
people, organisations or other systems) with the use of interfaces, in order to achieve
the intelligent operation of the whole shop floor system. This paper focuses on the
implementation and application of the interface of shop floor real-time scheduling and
control.

Figure 1. The framework of IoMT.


ENTERPRISE INFORMATION SYSTEMS 7

3.2. Production process control and information management system of


manufacturing shop floor based on Internet of Things
The function design of production process control and information management system
based on Internet of Things adopts service-oriented mode. There are two main services:
public service and manufacturing process service.
Among them, public services include system permission configuration, RFID reader
configuration, RFID tag management and manufacturing network routing configuration.
They are mainly used to initialise RFID devices, network connection and user access.
Through the configuration of public services, a universal hardware and software envir-
onment is provided for the control and management of the IoT enabled manufacturing
process.
Manufacturing process service includes three key parts: process management service,
process control service and data interface service.
The process management service manages data acquisition of manufacturing process
based on RFID technology, including RFID equipment management, production factor
status management, on line inspection and testing of work-in-process (WIP) and man-
ufacturing task management.
Process control service provides visual supervising of manufacturing process,
optimal scheduling of production factors and big data functions of manufacturing
process. Through the visual supervising of manufacturing process, shop floor can be
controlled more intuitively and real-time, and shop floor scheduling system can
obtain production status that keeps synchronisation with field manufacturing envir-
onment. On the basis of real-time supervising of manufacturing process, real-time
optimal scheduling of shop floor production factors is carried out to adjust the
deviation between actual production status and production operation plan. In the
dynamic shop floor environment, big data analysis and decision-making are carried
out for manufacturing process, including rescheduling drive, rescheduling strategy
and rescheduling algorithm.
Data interface service mainly aims at data interaction and information integration
among subsystems, including information system interfaces such as RFID terminal inter-
face, scheduling system and shop floor control execution system, and other manufactur-
ing network interfaces.
In order to realise the effective perception and fusion of the information of hetero-
geneous and big data in the environment of IoT enabled manufacturing, the standar-
dised and reasonable design of the system database is the foundation. According to the
characteristics of the manufacturing process, the database design needs to meet the
following requirements: due to the frequent acquisition of information in the manufac-
turing process based on RFID, data redundancy should be minimised; there is a large
number of real-time data interaction between client and server to facilitate access; the
system should be scalable and relatively adaptable; data safety should be ensured.
Considering the data requirement, dynamic scheduling requirement, processing
requirement, security and integrity of the IoT enabled manufacturing system,
a database for the production process control and information management system is
designed. Figure 2 shows up the Entity-Relationship (E-R) model figure, and construct
the relation table structure.
8 S. TIAN ET AL.

3.3. Internet of Things shop floor behaviour model


Most of the existing shop floor control systems are centralised control systems, which
can only determine a fixed process path statically and cannot meet the needs of

Figure 2. Database E-R model of production process control and information management system.
ENTERPRISE INFORMATION SYSTEMS 9

scheduling and control in dynamic shop floor environment. In IoMT environment, the
jobs need dynamic negotiation and interaction with transporting equipment, processing
equipment, etc. This negotiation and interaction can self-adaptively complete the trans-
porting and processing. The prefabricated processing path in the RFID tag will be
updated in real time according to the operation status of the shop floor.
The shop floor production process is composed of parallel processing of multiple
operations, and each operation represents a minimum production cycle. Jobs are
transported by AGV or transfer robot to the cache location of the machine arranged
by the initial scheduling process. According to the initial scheduling scheme, the cache
requests for machining process by the corresponding machine.
The transfer robot reads the RFID tag information and transfers the job to the buffer of the
machine selected in the scheduling scheme based on the scheduling information stored in the
tag. When the job arrives at the buffer of the machine selected by the operation to be
processed, the transfer robot sends the processing status query message to the corresponding
machine, and the machine feeds back its own processing status information to the transfer
robot. According to the feedback information, the transfer robot arranges the feeding robot to
finish waiting or feeding, processing and cutting. In this process, according to the machine
processing status and the label information of each job in the buffer, the feeding robot
chooses the job with the earliest planned start processing time to clamp on the machine for
processing, and puts the job back into the cache after the processing is completed. Figure 3 is
the negotiation of machining process.

Figure 3. Mechanism of negotiation of machining process.


10 S. TIAN ET AL.

Manufacturing systems often encounter many disturbances, and this paper focuses
on the shop floor coordination strategy when machine breakdown. When the machine
breakdown occurs, the unscheduled and affected operations need to be rescheduled to
ensure the smooth operation of the shop floor.
The objective function of the static shop floor problem mostly reflects the production
efficiency of the shop floor, while the dynamic shop floor problem is more focused on reducing
the deviation between the rescheduling scheme and initial scheduling scheme. Therefore, this
paper mainly studies the dynamic shop floor problem from the perspectives of robustness and
stability.
At a certain time after the abnormal event disturbance occurs, the transfer robot
reads the data in the RFID tag of the job and selects one of the candidate machines to
process. The basis of selection is the result of game equilibrium between robustness and
stability of shop floor under current state. Detailed game-based rescheduling methods
are given in Sections 4 and 5.
After updating the label scheduling information, the manufacturing processing can
continue according to the rescheduling scheme. The dynamic coordination and negotia-
tion mechanism control mechanism of manufacturing process is shown in Figure 4.

4. Rescheduling in a dynamic flexible job shop with random machine


breakdown
The rescheduling process can be regarded as a multi-objective optimisation process
with the robustness and stability as indicators. However, the robustness and stability are
usually a pair of contradictory indicators. In this section, we use the game theory-based
multi-objective optimisation method to obtain the rescheduling scheme.

Figure 4. Dynamic coordination and negotiation control mechanism for manufacturing processing.
ENTERPRISE INFORMATION SYSTEMS 11

4.1 Problem descriptions


The static shop floor scheduling problem or FJSP is specifically described as the job
task set Job= {J1, J2, . . ., Jm} to be processed on the machine set M= {M 1, M 2, . . ., M n}.
Each job Ji consists of a sequence of ordered operations, and each operation Oij (i.e.
operation j of job i) has to be processed by one machine Mk out of a set of allowable
machines Sij. The processing time of an operation Oij on machine Mk is Pk ij. The
constraints of processing include one machine cannot handle multiple operations at
a time, and one operation cannot be processed on multiple machines at a time.
A solution to the problem consists of assigning both a machine and a starting time to
each operation to minimise the maximum completion time of the jobs (i.e. min
(Cmax)).
In the real shop floor environment, the implementation of scheduling may be inter-
rupted or delayed due to the disturbance of abnormal events. Machine breakdown is
a common abnormal event; i.e. a machine in the shop floor environment cannot work
properly within a certain period of time and needs maintenance or repair.
When the machine repair is completed, the shop floor system runs according to the
new schedule. Therefore, the mean time to repair (MTTR) and mean time between
failures (MTBF) can be used to describe the dynamic flexible job shop scheduling
problem, and it is generally believed that both MTTR and MTBF obey the gamma
distribution.
The following assumptions are made about the nature of an interruption:

(1) The end time of the interruption is known at the startup.


(2) If an operation is affected by an interruption, the processing must be restarted.
The process of the affected operation needs to start again.

The objective function of the static FJSP problem mostly reflects the production effi-
ciency of the shop floor, while the dynamic FJSP problem is more focused on reducing
the deviation between the rescheduling scheme and initial scheduling scheme.
Therefore, this paper mainly studies the dynamic FJSP problem from the perspectives
of robustness and stability.
To calculate the robustness, we use the relative robustness measurement model
proposed by Kouvelis and Yu (1997):

jMS R  MS Pj
RM ¼  100% (1)
MS P
where MS_R is the realised makespan and MS_P is the makespan of the predicted
scheduling.
To calculate the stability, we use the stability measurement model proposed by Al-
Hinai and Elmekkawy (2011).
Pn Pqi  

i¼1 j¼1 COijP  COijR
SM ¼ Pn  100% (2)
ð i¼1 Oi Þ

where COijP is the predicted completion time of operation j of job i, COijR is the realised
completion time of operation j of job i, Oi is the total number of operations with no
12 S. TIAN ET AL.

rerouting to alternate machines for job i, n is the number of jobs, and qi is the number of
operations of job i.

4.2. Game theory-based multi-objective rescheduling method


In multi-objective optimisation problems, there is often no unique global optimal
solution, but there is an optimal solution set called the Pareto optimal solution set,
also known as the non-inferior solution set.
To overcome the limitations of the traditional multi-objective optimisation methods,
the multi-objective game optimisation method has emerged in recent years. It can
improve the convergence speed and computational efficiency without determining
the weight of each objective artificially.
According to game theory, G= {S1, S2, . . ., SP; u1, u2, . . ., uP} represents a game,
where the strategic space set of P game players is {S1, S2, . . ., SP}, sgh∈Ss represents
the hth strategy of game player g, and {u1, u2, . . ., uP} is the payoffs of all game
players.
The results of scheduling can be expressed as follows: on the premise of satisfying
various constraints and process routes, determining the machine for each operation and
the start processing time of an operation on the selected machine. Therefore, the
machines can be mapped into decision variables, and the operations arranged on the
machine can be mapped into strategy sets. The objective functions of the multi-
objective optimisation problem, i.e. the robustness and stability, can be regarded as
the payoffs of the corresponding players, and the robustness and stability are the
transient values of different strategy combinations at some decision-making time. The
game model proposed in this paper is a two-player game model, in which the robust-
ness and stability are considered as two players, and the interests of the players can be
written as (RM, SM).
In general multi-objective optimisation problems, for each optimisation objective
function, all decision variables are common, and the strategy set of each player in the
game model is independent and not disturbed by other players. Therefore, it is neces-
sary to divide and classify the decision variables of the multi-objective optimisation
problem according to the mapping requirements between the player and the strategy
set and obtain the strategy set belonging to each player. In this paper, the decision
variables are divided based on machines. Each machine is assigned to a player in turn.
S is the set of alternative strategies or behaviours of the players, and S = S1 ∪ S2, S1 ∩ S2
= Ø, where S1 is the policy set of SM and S2 the policy set of RM.
In the dynamic FJSP problem, the positions of robustness and stability are equal; i.e.
there is no objective preference. RM and SM, as the players, do not pay attention to the
overall benefits. In this case, the Nash equilibrium model can be introduced into the
multi-objective game model.
Each player takes its own payoff as the goal, carries out the single-objective optimisa-
tion in the respective strategy space, and obtains the best strategy for the other players.
Because the game model in this paper assigns the machine in turn to the game player,
the action sequence of players is successive, and the extended game model can be used
to describe it. Through a dynamic game, a sub-game perfect Nash equilibrium is
obtained. The sub-game perfect Nash equilibrium is also the optimal solution of the
ENTERPRISE INFORMATION SYSTEMS 13

multi-objective dynamic FJSP problem. The game theory-based solution process for the
dynamic FJSP is as Figure 5.

4.3. The solution of the sub-game perfect Nash equilibrium


The backward induction method is the most commonly used method to solve the sub-
game perfect Nash equilibrium. The general steps of the backward induction method for
solving finite extended games are as follows:
Step 1: Listing all the sub-games of the current game.
Step 2: Solving the lowest sub-game, deleting the non-Nash equilibrium strategy, and
solving the upper sub-game step by step until the original game.
The solution obtained by the backward induction method is a sub-game perfect Nash
equilibrium, which is a rescheduling scheme to satisfy the requirements of stability and
robustness optimisation.
To illustrate this process, an example is presented in Table 1, in which rows represent
processes and columns represent candidate machines. Figure 6(a) is a Gantt diagram of

Figure 5. The game theory-based solution process for the dynamic FJSP.
14 S. TIAN ET AL.

Table 1. Example of a three-job, three-machine FJSP.


Machines
Operations M1 M2 M3
O11 3 4 ∞
O12 2 ∞ 4
O13 3 2 2
O21 ∞ 2 4
O22 4 5 3
O23 3 4 3
O31 2 3 2
O32 ∞ 4 2
O33 ∞ 2 4

Figure 6. (a) Initial scheduling Gantt chart of a dynamic FJSP. (b) The procedure of solving the sub-
game perfect Nash equilibrium. (c) The scheduling Gantt chart of an optimal rescheduling scheme.

the initial scheduling scheme for this example. At time 3, machine M3 requires a repair of
1 unit time.
At time t = 4, machine M3 is idle. According to the process constraints, the operations
{O1, 2, O2, 2} can be selected. Because there is only one idle machine, and machine M1
selects the stability index, the operation with less stability is selected, i.e. O1, 2. At time
t = 6, machine M1, machine M2 and machine M3 are idle. Machine M1 can choose the
operations {O2, 2, O1, 3}, machine M2 can choose the operations {O1, 3, O2, 2, O3, 3}, and
machine M3 can choose the operations {O1, 3, O2, 2, O3, 3}. The procedure of backward
ENTERPRISE INFORMATION SYSTEMS 15

induction is shown in Figure 6(b). According to the game result, M1 selects O2, 2, M2
selects O3, 3, and M3 selects O1, 3. At time t = 10, machine M1, machine M2 and
machine M3 are idle. According to the process constraints, the operation {O2,3} can be
selected. Because the stability and robustness indexes of machine M3 are the optimal
values, machine M3 selects operation O2,3. The Gantt diagram of the rescheduling
scheme is shown in Figure 6(c). The results meet the requirements of practical
engineering.

5. Rolling horizon rescheduling strategy and dynamic scheduling strategy


To realise a rapid response to the complex and changing environment of the shop
floor, the most ideal strategy is to carry out rapid rescheduling at each scheduling
time, but frequent rescheduling must be avoided in the actual shop floor environ-
ment. In addition, because the solution obtained by using the game method is
a trade-off solution, with the expansion of the scale of the scheduling problem
and the increase of the disturbance frequency of abnormal events, the solution
quality and efficiency will inevitably decrease. The introduction of the rolling horizon
rescheduling strategy can greatly simplify the model of rescheduling, i.e. the initial
schedule is revised by rescheduling in a rolling horizon, rather than in the whole
horizon.

5.1. Rolling horizon rescheduling strategy


The whole dynamic scheduling optimisation process can be decomposed into several
scheduling optimisation intervals, and the scheduling scheme is revised by the rolling
horizon optimisation.
When the rolling horizon rescheduling strategy is applied to rescheduling, three process
windows are defined first: the completion window, scheduling-execution window and
waiting window. The completion window stores all the completed execution operations
in the completion scheduling period. The scheduling-execution window stores all the
operations that could not be executed in the previous period and need to be scheduled
and executed in this period. The waiting window stores all operations except for the
operations in the completion window and the scheduling-execution window.
The rescheduling process is carried out in the rolling time domain: a certain number
of operations are selected from the waiting window to the scheduling-execution win-
dow at the initial time, and processing instructions are issued to the execution system
after the rescheduling is completed; when the next rescheduling period arrives, the
operations that are executed in the scheduling-execution window are moved out to the
completion window, and then a certain number of operations are selected from the
waiting window. The move in and move out are carried out periodically in the schedul-
ing-execution window until all operations are moved to the completion window, thus
realising the rescheduling based on rolling horizon optimisation.
Corresponding to the definitions of the windows, the operations can also be divided
into four sets: the complete execution operations set, executing operations set, and
waiting for execution operations set and waiting for scheduling operations set. Figure 7
is a mapping relationship between the windows of each operation set. Among them, the
16 S. TIAN ET AL.

Figure 7. Relation chart of the operation windows and the collections of operations.

completion window stores the complete execution of the process set, the scheduling-
execution window stores the execution of the process set and the waiting for execution
of the process set, and the waiting window stores the complete execution operations.

5.2. Dynamic scheduling strategy


There are two types of rescheduling-driven strategies based on rolling horizon optimisa-
tion: the event-driven strategy and the periodic-driven strategy. The event-driven strat-
egy is that whenever an abnormal event has happened, the rescheduling is triggered.
The periodic rescheduling strategy is that rescheduling is triggered at the beginning of
every defined rescheduling period (A. Liu, Fowler, and Pfund 2016).
The periodic-driven strategy is mainly used to deal with the deviation or cumulative
deviation events encountered in the execution process of the shop floor system. The
periodic-driven strategy is simple and reliable, but it cannot handle abnormal event
interference. The event-driven strategy can deal with abnormal event interference, but it
lacks the ability to predict future and global optimisation. Rescheduling can only be
triggered at a specific time when using periodic-driven strategy, and it is difficult to
predict the impact of interference on the operation of the shop floor in the future when
using event-driven strategy.
Considering the advantages and disadvantages of these two driving strategies,
this paper adopts a rescheduling strategy based on a combination of them. If
abnormal event interference such as machine breakdown occurs, rescheduling will
be triggered. Otherwise, rescheduling will be triggered at the beginning of every
defined rescheduling period. The rescheduling period can be determined according
to the tasks' load of the shop floor. Therefore, to solve the dynamic FJSP problem
with random machine breakdown, an HGR method is proposed, which is shown in
Figure 8.
ENTERPRISE INFORMATION SYSTEMS 17

Figure 8. Method flow of a hybrid method combining the game theory method and rolling horizon
rescheduling strategy (HGR).

6. Computational and experimental studies


To test the performance of the proposed HGR method, extensive experimental evalua-
tion and comparisons with existing method s are provided using well-known flexible job
shop scheduling (FJSP) benchmark sets. The methods are coded and executed by using
MATLAB R2009a (The MathWorks Inc., www.mathworks.com) on an Intel Core™ i5 GHz
with 3.86 GB RAM. In the following, we apply the designed shop floor information
management system to a locomotive axle manufacturer enterprise to further test the
engineering application ability.

6.1. Experimental study with machine breakdown scenarios


Two sets of instances are considered in this paper:

(1) Four total FJSP benchmarks: Ex1 consisting of 4 × 5, Ex2 consisting of 10 × 7, Ex3
consisting of 10 × 10, and Ex4 consisting of 10 × 15 taken from Kacem, Hammadi,
and Borne (2002b).
18 S. TIAN ET AL.

(2) Eleven partial FJSP benchmarks: Ex5 consisting of 8 × 8 taken from Kacem,
Hammadi, and Borne (2002a) and examples MK01–MK10 with different sizes
varying between 10 × 6 and 20 × 15 as proposed by Brandimarte (1993).

In our work, we first represent a machine breakdown scenario by L (i, t, r), which
indicates that machine i at time t needs r units of time to be recovered.
Here, based on the experiment results presented in Holthaus (1999), each L (i, t, r) can
be determined. Following Holthaus (1999), in our work, we set Ag and MTTR to 0.05 and
p̅ , respectively, where p̅ denotes the mean total processing time of a job, MTTR denotes
the mean time to repair, and Ag denotes the breakdown level of the shop.
To determine the start time of breakdown t, the parameter λ of a Poisson distribution
is considered as follows:

λ ¼ modðmakespanpre ; MTBFÞ þ ðn  1Þ  MTBF; n ¼ 1 : fixðmakespanpre ; MTBFÞ (3)

where makespanpre is the makespan of predictive scheduling; fix (makespanpre, MTBF)


and mod (makespan, MTBF) are the quotient and remainder, respectively; n is the
number of breakdowns; and MTBF is the mean time between failures.
To determine the recovery duration r, the parameter of an exponential distribution is
considered to be the MTTR.
The results of the stochastic simulation, i.e. the details for scenarios of random
machine breakdown, are listed in Table 2.
For the breakdown, we choose the parameter L (i, t, r) and use these scenarios for the
experiments. For each datum of a Kacem instance, prescheduling without machine
breakdown is generated by the HDFA approach proposed by Karthikeyan, Asokan, and
Nickolas (2014). For each datum of a Brandimarte instance, prescheduling without
machine breakdown is generated by the HTGA approach proposed by Cheng, Peng,
and Lü (2016).
To measure the contribution of the proposed approach, we compare the outcome
with gap reduction (SGR) (Hasan et al. 2011), RSS (Wang, Yin, and Da Qin 2013), and all
rerouting (AAR) (Kutanoglu and Sabuncuoglu 2001). All obtained rescheduling instances
are applied to the same scenario.

Table 2. Scenarios of random machine breakdown.


Instance n× m p̅ L (I, t, r)
Ex1 4×5 2.67 (4, 9, 3)
Ex2 10 × 7 2.14 (3, 5, 3)
Ex3 10 × 10 1.40 (4, 3, 6)
Ex4 10 × 15 1.63 (5, 3, 5)
Ex5 8×8 2.85 (4, 5, 3)
MK01 10 × 6 3.11 (6, 11, 6)
MK02 10 × 6 2.62 (1, 7, 3)
MK03 15 × 8 6.62 (4, 17, 7) (1, 75, 4) (6, 130, 5)
MK04 15 × 8 4.33 (3, 20, 5) (7, 57, 4)
MK05 15 × 4 6.47 (1, 64, 12) (3, 115, 5)
MK06 10 × 15 3.01 (9, 10, 2) (3, 40, 3)
MK07 20 × 5 6.97 (1, 11, 10) (4, 78, 6)
MK08 20 × 10 11.53 (9, 2, 11) (4, 102, 11) (10, 179, 14) (7, 309, 7) (1, 451, 7) (10, 520, 15)
MK09 20 × 10 10.14 (9, 37, 11) (6, 112, 10) (9, 227, 16)
MK10 20 × 15 8.50 (4, 63, 9) (8, 138, 13)
ENTERPRISE INFORMATION SYSTEMS 19

In this research, we use two criteria to evaluate and compare the results in predictive-
reactive scheduling. These criteria include the following:

(1) Relative error

The relative error of each of the problems is as follows:


 
RE ¼ makespanrea  makespanpre =makespanpre  100% (4)

where makespanpre denotes the makespan of prescheduling, and makespanrea denotes


the makespan of rescheduling.

(2) Mean ideal distance

To evaluate the efficiency of the HGR method on the multi-objective optimisation


(also known as Pareto optimisation) problem, we use the criterion of the mean ideal
distance (MID), which presents the distance of the Pareto front solutions from an ideal
point.

The ideal point in this work is point (0, 0) on a two-dimensional graph, with the X-axis
denoting robustness and the Y-axis indicating stability. A lower value of the MID indicates
that the performance of the algorithm is improved. The MID is calculated as follows:

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
MID ¼ Robustness2 þ Stability2 (5)

where Robustness and Stability denote the values of the solution for robustness and
stability, respectively.
In our test environment, the rescheduling period is set to 15. We present the results
for the two sets of instances noted above in terms of the RE and MID, as presented in
Table 3. The first and second columns record the instance name and predictive

Table 3. Computational results for test cases using the criteria of relative error and MID.
HGR RSS ARR SGR
Instance Predictive makespan RE MID RE MID RE MID RE MID
MK01 40 2.50 5.35 10.00 11.45 2.50 6.32 5.00 6.94
MK02 26 7.69 8.81 7.69 8.02 7.69 8.69 11.54 12.51
MK03 204 5.88 10.81 6.37 7.36 5.39 9.99 6.86 11.41
MK04 60 11.67 13.14 26.67 27.68 13.33 14.73 15.00 16.33
MK05 173 4.62 7.26 12.14 12.78 3.47 6.79 6.36 8.49
MK06 61 7.02 7.22 19.30 20.04 8.77 9.36 8.77 9.05
MK07 141 9.93 14.62 19.86 20.35 9.93 15.28 10.64 17.07
MK08 523 8.60 15.93 9.37 9.80 9.37 16.35 9.37 18.19
MK09 307 3.58 8.46 5.86 5.95 4.23 9.40 4.56 8.41
MK10 213 4.23 6.68 5.16 5.89 7.51 9.74 6.10 8.55
Ex1 11 18.18 18.30 18.18 18.30 18.18 18.30 18.18 18.30
Ex2 11 18.18 18.35 36.36 36.64 36.36 36.97 18.18 18.78
Ex3 7 28.57 28.68 57.14 57.41 42.86 43.12 57.14 57.74
Ex4 12 25.00 25.63 33.33 33.51 25.00 25.80 25.00 25.96
Ex5 14 14.29 16.88 35.71 36.87 21.43 22.85 14.29 16.88
a a
Mean / 11.33 13.74 20.21 20.80 14.40 16.91 14.47 16.98
a
The optimum result.
20 S. TIAN ET AL.

makespan of each problem, respectively. Figure 9(a) shows the initiating scheduling
Gantt chart on MK04. The proposed method is shown in Figure 9(b).

6.2. Application example


Locomotive axle manufactory enterprises are typical mixed-flow production enterprises.
The main production processes of locomotive axles are listed in Table 4. The production
plan of a locomotive axle manufacturer for one week is listed in Table 5. The proposed
shop floor information management system and dynamic scheduling method based on
the Internet of Things are applied to the shop floor production. When the disturbance of
machine breakdown occurs in the shop floor, the new scheduling scheme can be
adaptively obtained, as shown in Figure 10. The e-Signage is shown in Figure 11.

Figure 9. (a) Gantt chart for initiating scheduling (i.e. instance MK04). (b) Gantt chart determined by
HGR for instance MK04 with random machine breakdown.
ENTERPRISE INFORMATION SYSTEMS 21

Table 4. Processing technology of locomotive Axle.


Number Name
30 Rough Turning Shape
40 Semi-finish Turning Shape
50 Drilling Inner Hole (Axle Center Hole)
60 Honing Inner Hole (Axle Center Hole)
65K Ultrasound Flaw Detection (Inside Hole Surface Defect Detection)
70 Saw Axle End
80 Fine Milling End Face
90 Precision Turning Frame Position, End Face and Cone Angle
100 Semi-finish Turning Shape
110 Drilling Four Screw Holes at Axis End
120 Marking
125K Ultrasonic Examination (Second Inspection of Internal Discontinuity)
130 Fine Turning Shape
140 Finishing cone angle
150 Precision Grinding of Shaft Neck and Dust-proof Seat
160 Precision Grinding of Wheel Seat and Gear Box Seat
170 Rolling Circle and Arc
180 Polishing
185K Finished Product Inspection
190K Magnetic-particle inspection

Table 5. Weekly production plan for locomotive Axles.


Product name Product Number Quantity
RD2 axle (semi-finishing) 02 3
RD3A axle (semi-finishing) 11 20
DZ760B axle (semi-finishing) 12 12
Power axle of Chengdu Line 10 (semi-finishing) 13 15
Trailer axle of Chengdu Line 10 (semi-finishing) 14 11
GE-A1 locomotive axle 06 8
GE-3 locomotive axle 06 8
GE Pakistan Locomotive Axle 06 10
RE2B Axle 03 10
BA302 axle 15 7
G-2 Australia Locomotive axle 16 3
Zhengzhou metro power axle 17 2

Figure 10. Dynamic scheduling scheme.


22 S. TIAN ET AL.

Figure 11. The e-Signage for shop floor.

6.3. Results discussion


As seen in Figure 6, the change of the operations sequence is not significant. From the
results reported in Table 4, for the 15 problems, the increase in makespan for breakdown
maintenance is not significantly different, on average, from the preventive maintenance
scenarios using HGR.
From the results reported in Table 4, two indexes of criteria (i.e. RE and MID) are
examined in each scenario. Because the average of the RE for HGR is equal to 11.33, the
HGR algorithm results in a higher quality than the SGR policy, AAR policy or RSS policy.
Additionally, the average of the MID for HGR is equal to 13.74, which means that the result
of HGR has better performance in terms of robustness and stability than the other
algorithms. The proposed method can greatly reduce the transfer and accumulation of
deviations between windows, and can also dialectically integrate local and global optima.
The experiment results show that even if machine breakdown occurs, one-day
production tasks can be finished in several hours and therefore the production efficiency
can be greatly improved. From the results of Application example, it can be concluded
that the shop floor information management system and dynamic scheduling method
can be used to solve the dynamic scheduling problems according to the real-time status
of the shop floor, the negotiation-based dynamic scheduling strategy and wireless
communication coordination mechanism.

7. Conclusion and prospects


This paper offers a production information management system based on industrial
Internet of Things (IoT) technology. The application of information management system
can reasonably assign and manage the resources of orders, materials and equipment,
which makes the manufactory enterprise have the ability of dynamic shop floor schedul-
ing. A new dynamic shop floor rescheduling method, HGR, is proposed, which combines
ENTERPRISE INFORMATION SYSTEMS 23

the game theory method with the rolling horizon rescheduling strategy. The HGR can
solve not only the predictive scheduling problem but also the real-time scheduling or
rescheduling problem. To evaluate the performance of the proposed method,
a comparative experiment with other rescheduling methods is designed, which simu-
lates a random machine breakdown disturbance to the FJSP standard example problem.
According to the analysis of the RE and MID indexes, i.e. the values of RE and MID are
11.3 and 13.74, respectively, the proposed HGR shows better performance in 15 dynamic
FJSP instances. An applied example of a locomotive axle manufacturing enterprise
shows that the IoT-enabled shop floor information management system and dynamic
scheduling method can handle the practical problems and improve production
efficiency.
In many practical applications, there are many kinds of abnormal disturbances, such
as urgent order arrival, over- or underestimation of the processing time, order cancella-
tion, changes to the due date and lagging behind or being ahead of schedule. Further
work will consider using the proposed method to study the dynamic FJSP under other
abnormal event interruptions.

Acknowledgments
The authors are grateful for the financial support provided by National Natural Science Foundation
of China (Grant No. 51475324) and National Natural Science Foundation of China (Grant No.
51605328).

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by the National Natural Science Foundation of China
[51475324,51605328].

References
Akturk, M. S., and E. Gorgulu. 1999. “Match-Up Scheduling under a Machine Breakdown.” European
Journal of Operational Research 112 (1) PO BOX 211, 1000 AE AMSTERDAM, NETHERLANDS:
ELSEVIER SCIENCE BV: 81–97. doi: 10.1016/S0377-2217(97)00396-2.
Al-Hinai, N., and T. Y. Elmekkawy. 2011. “Robust and Stable Flexible Job Shop Scheduling with
Random Machine Breakdowns Using a Hybrid Genetic Algorithm.” International Journal of
Production Economics 132 (2) PO BOX 211, 1000 AE AMSTERDAM, NETHERLANDS: ELSEVIER
SCIENCE BV: 279–281. doi: 10.1016/j.ijpe.2011.04.020.
Bi, Z. 2017. “Embracing Internet of Things (Iot) and Big Data for Industrial Informatics.” Enterprise
Information Systems 11 (7): 949–951. doi:10.1080/17517575.2016.1258734.
Bitam, S., S. Zeadally, and A. Mellouk. 2018. “Fog Computing Job Scheduling Optimization Based
on Bees Swarm.” Enterprise Information Systems 12 (4): 373–397. doi:10.1080/
17517575.2017.1304579.
Brandimarte, P. 1993. “Routing and Scheduling in a Flexible Job Shop by Tabu Search.” Annals of
Operations Research 41 (3): 157–183. doi:10.1007/BF02023073.
24 S. TIAN ET AL.

Cheng, T. C. E., B. Peng, and Z. Lü. 2016. “A Hybrid Evolutionary Algorithm to Solve the Job Shop
Scheduling Problem.” Annals of Operations Research 242 (2): 223–237. doi:10.1007/s10479-013-
1332-5.
Cochran, D. S., J. F. Arinez, M. T. Collins, and Z. Bi. 2017. “Modelling of Human–Machine Interaction
in Equipment Design of Manufacturing Cells.” Enterprise Information Systems 11 (7): 969–987.
doi:10.1080/17517575.2016.1248495.
Ehtamo, H. 2003. “Dynamic Noncooperative Game Theory.” Journal of Economic Dynamics and
Control 21 (6): 1113–1116. doi:10.1016/s0165-1889(97)00025-0.
Fahmy, S. A., S. Balakrishnan, and T. Y. Elmekkawy. 2009. “A Generic Deadlock-Free Reactive
Scheduling Approach.” International Journal of Production Research 47 (20): 5657–5676.
doi:10.1080/00207540802112652.
Glaschke, C., and N. Gronau. 2016. “New Approaches for Automated Process Model Discovery.”
Lecture Notes in Business Information Processing 257: 23–36. doi:10.1007/978-3-319-40512-4_2.
Hasan, S., M. Kamrul, R. Sarker, and D. Essam. 2011. “Genetic Algorithm for Job-Shop Scheduling
with Machine Unavailability and Breakdowns.” International Journal of Production Research 49
(16) 4 PARK SQUARE, MILTON PARK, ABINGDON OX14 4RN, OXON, ENGLAND: TAYLOR &
FRANCIS LTD: 4999–5015. doi: 10.1080/00207543.2010.495088.
Heloulou, I., M. S. Radjef, and M. T. Kechadi. 2017a. “A Multi-Act Sequential Game-Based
Multi-Objective Clustering Approach for Categorical Data.” Neurocomputing 267: 320–332.
doi:10.1016/j.neucom.2017.06.011.
Heloulou, I., M. S. Radjef, and M. T. Kechadi. 2017b. “Automatic Multi-Objective Clustering Based on
Game Theory.” Expert Systems with Applications 67: 32–48. doi:10.1016/j.eswa.2016.09.008.
Holthaus, O. 1999. “Scheduling in Job Shops with Machine Breakdowns: An Experimental Study.”
Computers and Industrial Engineering 36 (1): 137–162. doi:10.1016/S0360-8352(99)00006-6.
Ji, Y., M. Li, and S. Qu. 2018. “Multi-Objective Linear Programming Games and Applications in
Supply Chain Competition.” Future Generation Computer Systems 86: 591–597. doi:10.1016/j.
future.2018.04.041.
Jiang, Y. C., L. P. Chen, and J. Duan. 2016. “A New Practice-Driven Approach to Develop Software in
A Cyber-Physical System Environment.” Enterprise Information Systems 10 (2): 211–227.
doi:10.1080/17517575.2014.939107.
Kacem, I., S. Hammadi, and P. Borne. 2002a. “Pareto-Optimality Approach for Flexible Job-Shop
Scheduling Problems: Hybridization of Evolutionary Algorithms and Fuzzy Logic.” Mathematics
and Computers in Simulation 60 (3–5): 245–276. doi:10.1016/S0378-4754(02)00019-8.
Kacem, I., S. Hammadi, and P. Borne. 2002b. “Approach by Localization and Multiobjective
Evolutionary Optimization for Flexible Job-Shop Scheduling Problems.” IEEE Transactions on
Systems, Man and Cybernetics Part C: Applications and Reviews 32 (1): 1–13. doi:10.1109/
TSMCC.2002.1009117.
Karthikeyan, S., P. Asokan, and S. Nickolas. 2014. “A Hybrid Discrete Firefly Algorithm for
Multi-Objective Flexible Job Shop Scheduling Problem with Limited Resource Constraints.”
International Journal of Advanced Manufacturing Technology 72 (9–12): 1567–1579.
doi:10.1007/s00170-014-5753-3.
Kouvelis, P., and G. Yu. 1997. Robust Discrete Optimization and Its Applications. Boston, MA: Kluwer
Academic Publishers.
Kutanoglu, E., and I. Sabuncuoglu. 2001. “Routing-Based Reactive Scheduling Policies for Machine
Failures in Dynamic Job Shops.” International Journal of Production Research 39 (14): 3141–3158.
doi:10.1080/00207540110057909.
Liu, A., J. Fowler, and M. Pfund. 2016. “Dynamic Co-Ordinated Scheduling in the Supply Chain
considering Flexible Routes.” International Journal of Production Research 54 (1): 322–335.
doi:10.1080/00207543.2015.1115908.
Liu, S., G. Liu, and H. Zhou. 2018. “A Robust Parallel Object Tracking Method for Illumination
Variations.” Mobile Networks and Applications. doi:10.1007/s11036-018-1134-8.
Liu, S., W. Bai, G. Liu, W. Li, and H. M. Srivastava. 2018. “Parallel Fractal Compression Method for Big
Video Data.” Complexity 2018: 1–16. doi:10.1155/2018/2016976.
ENTERPRISE INFORMATION SYSTEMS 25

Liu, S. S., and K. C. Shih. 2009. “Construction Rescheduling Based on a Manufacturing Rescheduling
Framework.” Automation in Construction 18 (6) PO BOX 211, 1000 AE AMSTERDAM,
NETHERLANDS: ELSEVIER SCIENCE BV: 715–723. doi: 10.1016/j.autcon.2009.02.002.
Liu, Z., S. Q. Jiang, B. Y. Tang, J. H. Zhang, and H. Zhong. 2000. “The Study and Realization of
SCADA System in Manufacturing Enterprises.” In Proceedings of the 3rd World Congress on
Intelligent Control and Automation, 3688–3692\r3766. Vols 1–5.
Lou, P., Q. Liu, Z. Zhou, H. Wang, and S. X. Sun. 2012. “Multi-Agent-Based Proactive-Reactive
Scheduling for a Job Shop.” International Journal of Advanced Manufacturing Technology 59
(1–4) 236 GRAYS INN RD, 6TH FLOOR, LONDON WC1X 8HL, ENGLAND: SPRINGER LONDON LTD:
311–324. doi: 10.1007/s00170-011-3482-4.
Mak, V., A. Rapoport, and D. A. Seale. 2014. “Sequential Search by Groups with Rank-Dependent
Payoffs: An Experimental Study.” Organizational Behavior and Human Decision Processes 124 (2):
256–267. doi:10.1016/j.obhdp.2014.03.004.
Medhane, D. V., and A. K. Sangaiah. 2017. “Search Space-Based Multi-Objective Optimization
Evolutionary Algorithm.” Computers and Electrical Engineering 58: 126–143. doi:10.1016/j.
compeleceng.2017.01.025.
Nouiri, M., A. Bekrar, A. Jemai, D. Trentesaux, A. C. Ammari, and S. Niar. 2017. “Two Stage Particle
Swarm Optimization to Solve the Flexible Job Shop Predictive Scheduling Problem considering
Possible Machine Breakdowns.” Computers and Industrial Engineering 112: 595–606. doi:10.1016/
j.cie.2017.03.006.
Ouelhadj, D., and S. Petrovic. 2009. “A Survey of Dynamic Scheduling in Manufacturing Systems.”
Journal of Scheduling 12 (4): 417–431. doi:10.1007/s10951-008-0090-8.
Paprocka, I., and B. Skołud. 2017. “A Hybrid Multi-Objective Immune Algorithm for Predictive and
Reactive Scheduling.” Journal of Scheduling 20 (2): 165–182. doi:10.1007/s10951-016-0494-9.
Răileanu, S., F. Anton, T. Borangiu, S. Anton, and M. Nicolae. 2018. “A Cloud-Based Manufacturing
Control System with Data Integration from Multiple Autonomous Agents.” Computers in Industry
102: 50–61. doi:10.1016/j.compind.2018.08.004.
Sangaiah, A. K., O. W. Samuel, X. Li, M. Abdel-Basset, and H. Wang. 2018. “Towards an Efficient Risk
Assessment in Software Projects–Fuzzy Reinforcement Paradigm.” Computers and Electrical
Engineering 71: 833–846. doi:10.1016/j.compeleceng.2017.07.022.
Sartal, A., J. Llach, X. H. Vázquez, and R. de Castro. 2017. “How Much Does Lean Manufacturing
Need Environmental and Information Technologies?” Journal of Manufacturing Systems 45:
260–272. doi:10.1016/j.jmsy.2017.10.005.
Shalaby, M. A., T. F. Abdelmaguid, and Z. Y. Abdelrasol. 2012. “New Routing Rules for Dynamic
Flexible Job Shop Scheduling with Sequence-Dependent Setup Times.” In Proceedings of the
2012 International Conference on Industrial Engineering and Operations Management, Istanbul,
Turkey, July 3–6, 747–756. Red Hook, NY: Curran Associates, Inc.
Shukla, P. K. 2007. “On the Normal Boundary Intersection Method for Generation of Efficient
Front.” Computational Science–ICCS 2007: 310–317. doi:10.1007/978-3-540-72584-8_40.
Tao, Z., T. Xiao, and C. Hao. 2007. “Petri Net and GASA Based Approach for Dynamic JSP.” In
Proceedings of the 2007 IEEE International Conference on Mechatronics and Automation, ICMA
2007, 3888–3893. doi:10.1109/ICMA.2007.4304196.
Verdouw, C. N., R. M. Robbemond, T. Verwaart, J. Wolfert, and A. J. M. Beulens. 2015. “A Reference
Architecture for IoT-Based Logistic Information Systems in Agri-Food Supply Chains.” Enterprise
Information Systems. doi:10.1080/17517575.2015.1072643.
Vishwasrao, M. D., and A. K. Sangaiah. 2017. “ESCAPE: Effective Scalable Clustering Approach for
Parallel Execution of Continuous Position-Based Queries in Position Monitoring Applications.”
IEEE Transactions on Sustainable Computing 2 (2): 49–61. doi:10.1109/tsusc.2017.2690378.
Wang, L., L. Da Xu, Z. Bi, and Y. Xu. 2014. “Data Cleaning for RFID and WSN Integration.” IEEE
Transactions on Industrial Informatics 10 (1): 408–418. doi:10.1109/TII.2013.2250510.
Wang, Y. M., H. L. Yin, and K. Da Qin. 2013. “A Novel Genetic Algorithm for Flexible Job Shop
Scheduling Problems with Machine Disruptions.” International Journal of Advanced
Manufacturing Technology 68 (5–8): 1317–1326. doi:10.1007/s00170-013-4923-z.
26 S. TIAN ET AL.

Zhang, J., J. Yang, and Y. Zhou. 2016. “Robust Scheduling for Multi-Objective Flexible Job-Shop
Problems with Flexible Workdays.” Engineering Optimization 48 (11) PO BOX 211, 1000 AE
AMSTERDAM, NETHERLANDS: ELSEVIER SCIENCE BV: 1973–1989. doi: 10.1080/
0305215X.2016.1145216.
Zhang, Y., W. Wang, N. Wu, and C. Qian. 2016. “IoT-Enabled Real-Time Production Performance
Analysis and Exception Diagnosis Model.” IEEE Transactions on Automation Science and
Engineering 13 (3): 1318–1332. doi:10.1109/TASE.2015.2497800.
Zhong, R. Y., Q. Y. Dai, T. Qu, G. J. Hu, and G. Q. Huang. 2013. “RFID-Enabled Real-Time
Manufacturing Execution System for Mass-Customization Production.” Robotics and Computer-
Integrated Manufacturing 29 (2): 283–292. doi:10.1016/j.rcim.2012.08.001.

You might also like