2011 IEEE International Conference on
Automation Science and Engineering
Trieste, Italy - August 24-27, 2011
ThA2.4
Modeling Complex Logistics Systems using Soft Computing
Methodology of Fuzzy Cognitive Maps
Chrysostomos D. Stylios Member, IEEE and George Georgoulas
cover all modes of transportation and related functions
involving storage and cargo handling. Their requirements
are high autonomicity, great efficiency and intelligence,
which force engineers to investigate new techniques that can
integrate and combine well-known advanced methodologies
that will be the core of these sophisticated systems.
Any effective knowledge representation is based on
advanced modeling methods. Moreover, the requirements in
modeling and adequate description of systems cannot be met
only with the existing methodologies and theories.
Therefore, it is necessary to investigate and use new
methods that will exploit human experience, will have
learning capabilities and will take into account the
imprecision and uncertainty, which characterize real world
systems [3]. The flourishing of new theories that exploit the
synergy of discipline theories: such as Fuzzy Logic, Neural
Networks, Genetic Algorithms, Probabilistic Reasoning and
Knowledge Based Systems, known as Soft Computing
Techniques and/or Computational Intelligent techniques
motivate researchers to utilize them in order to create and
develop new models and sophisticated systems based on
knowledge exploitation [4]. Such advanced representation
techniques will use effectively all the knowledge of the
complex system resources, especially the insights and
experience of front-line operators and experts, in order to
achieve continuous improvements.
In the past years, conventional methods have been used
successfully in modeling and control systems but their
contribution is limited in the representation and solution of
complex systems. In such systems, their operation,
especially in the upper level, depends on human leadership.
Generally, there is a greater demand for autonomous
systems, which may be achieved by taking advantage of
human like reasoning and description of systems. Human
reasoning for any procedure includes uncertain description
and can have subtle variations in relation to time and space;
in such situations Fuzzy Cognitive Maps (FCMs) can be a
very suitable tool.
FCM were introduced by Kosko as a synergism of Fuzzy
Logic and Neural Networks [5][6]. Kosko enhanced the
cognitive maps theory that had been used in social and
political sciences to analyze social decision-making
problems; showing a causal relationship between different
factors, where the causal relationship is expressed by either
positive or negative sign of knowledge expressions
[7].Fuzzy values are introduced in FCMs to better represent
causal reasoning [8] forming a network of interconnected
concepts that can be used to model situations by classes and
their causal links between them.
Abstract—Fuzzy Cognitive Maps (FCMs) is an abstract soft
computing modeling methodology that has been applied in
many areas quite successfully. In this paper we discuss its
modeling applicability to complex logistics systems involved in
an intermodal container terminal and the way it could
represent and handle the vast amount of information by an
abstract point of view based on a decentralized approach,
where the supervisor of the system is modeled as an FCM. We
also investigate its applicability as a metamodel of the
intermodal terminal in a simulation-optimization framework.
Experts have a key role in developing the FCM as they describe
a general operational and behavioral model of the system using
concepts for the main aspects of the system, and weighted
directed edges to represent causality. On the other hand, when
data is available, data driven approaches have also been
proposed for the development of FCM models. The FCM
representation and implementation is discussed to develop a
behavioral model of any complex system mainly based on a
hierarchical structure, as well as its use as a metamodel of the
system.
I. INTRODUCTION
M
systems such as logistics systems are
characterized by uncertainties with high degree and
great complexity as it is observed in any production,
logistics and enterprise structures [1]. A quite common
approach in modeling complex systems is their description
as a system of connected agents that exhibits an emergent
global behavior not imposed by a central controller but
resulting from the interactions between the agents [2].
Complexity and large scale characterize modern logistics
systems involved in intermodal container terminals that
usually are described as systems-of-systems, which are nonmonolithic entities characterized by geographic distribution,
operational and management independence of their
subsystems and presenting emergent behavior and
evolutionary development.
Scale and complexity of logistics systems and their
modeling requirements and organizational needs
continuously increase. In addition to this, new practices,
structures, models, techniques and methods are emerging as
complements to increased needs. Models of advanced
logistics systems in intermodal container terminals have to
ODERN
Manuscript received March 11, 2011. This work was supported by the
E.U. FP7-PEOPLE-IAPP-2009, Grant Agreement No 251589, Acronym:
SAIL.
C. D. Stylios and G. Georgoulas are with the Knowledge and Intelligent
Computing Laboratory, Dept. of Informatics and Telecommunications
Technology, Technological Educational Institute of Epirus, 47100 Artas,
Greece. (email:, stylios@teiep.gr and georgoul@gmail.com)).
978-1-4577-1732-1/11/$26.00 ©2011 IEEE
72
Ci → Cj between the concept nodes. At time t the state of the
FCM is the 1 × n concept vector At gathering the values of
concepts At =[C1, …, Cn] ∈ [0,1] or a point in the fuzzy ncube state space. The n-by-n matrix W lists the n2 rules or
pathways in the causal web, represented by the
corresponding weight. FCM dynamics depend on the
dynamics of the concept nodes and causal edges. These
adaptive feedback fuzzy systems are nonlinear function
approximations with even more complex dynamics than
feedback neural networks [6].
FCMs have been used to make decision analysis and
coordinate distributed agents [9], to develop Medical
Decision Support Systems (DSSs) [10], and to accompany
case-based reasoning approaches [11]. FCMs have also been
used as structures for automating human problem solving
skills [12] and to represent complex social systems where
relationships between social forces demand feedback [13].
They have also been used to model and support plant control
systems of a water distribution system [14] and to perform
Failure Modes and Effects Analysis in the process industry
[15]. FCMs have been proposed to model the Supervisor of
complex manufacturing systems [16], concerning
hierarchical systems, where the supervisor incorporates
knowledge [17] and is capable of learning relational
structures and evidential reasoning [18].
It is believed that FCMs can improve model
representation and development of sophisticated systems,
combining characteristics from Fuzzy Logic and Neural
Networks and can contribute from a behavioral point of
view, first to model sub-systems at a lower level and,
second, at the supervisory decision and coordination level
[19]. Moreover, new hybrid methods based on
complementary approaches for enhancing FCMs
construction abilities make FCMs potential candidates to be
used as metamodels in a simulation-optimization
framework.
In the rest of this paper, section II will briefly introduce
FCMs, presenting the main aspects of this modeling
approach. Section III discusses the main requirements for
modeling complex logistics systems and how FCMs meet
their requirements. Section IV presents how FCMs could be
used to develop DSSs for logistics systems and Section V
introduces for the first time the usage of FCMs as
metamodel of a system. Finally section VI concludes the
paper and gives future research directions.
Fig. 1. The Fuzzy Cognitive Map model
With the graphical representation it becomes clear, which
concept influences other concepts, showing the
interconnections between concepts and permitting thoughts
and suggestions in the construction of the graph, as the
adding or deleting of an interconnection or a concept. In
conclusion, FCMs are fuzzy-graph structures, which allow
systematic causal propagation, in particular forward and
backward chaining. The simplest FCMs act as asymmetrical
networks of threshold or continuous concepts and converge
to an equilibrium point or limit cycles. They have non-linear
structure of their concepts and differ from Neural Networks
in their global feedback dynamics.
Experts design and develop the fuzzy graph structure of
the system, consisting of concepts-nodes that represent the
key principles-factors-functions of the system operation and
behavior. Then, they determine the structure and the
interconnections of the network using fuzzy conditional
statements. Experts use linguistic variables in order to
describe the relationship among concepts, and then all the
variables are combined and the weights of the causal
interconnections among concepts are determined. On the
other hand, the fuzzy graph structure can emerge through a
learning process based on accumulated historical data [22].
The mathematical model of an FCM can be described by
the matrix W, which gathers the weight values of the
interconnections between the n concepts of the FCM; and
the vector A, which gathers the values of the n concepts.
The value Ai for each concept Ci is calculated by the
II. FUZZY COGNITIVE MAPS
FCMs belong to the neuro-fuzzy systems that aim at
solving real world decision-making problems, modeling and
control problems [17]. Thus, neuro-fuzzy systems with their
ability to incorporate human knowledge and to adapt their
knowledge base via new optimization techniques are likely
to play increasingly important role in the conception and
design of hybrid intelligent systems [20]. An FCM is a
conceptual network, which is in most of the cases built by
experts, using an interactive procedure of knowledge
acquisition.
FCMs are dynamical systems that have the topology of a
directed fuzzy graph (Figure 1) consisting of nodes and
edges and permitting cycles and feedback. Nodes of the
graph stand for the concepts that are used to describe the
behavior of the system and they are connected by signed and
weighted arcs representing the causal relationships that exist
among concepts (Figure 1). It must be mentioned that all the
values in the graph are fuzzy, so concepts take values in the
range between [0, 1] and the weights of the interconnections
belong in the interval [-1, 1]. The concept nodes Ci have
fuzzy nature [21]. The edges wij define rules or causal flows
following rule, presented in equation 1:
⎛
⎜ n
A = f ⎜ ∑ A tj − 1W
i
⎜ ij≠= 1j
⎝
ji
⎞
⎟
+ A it − 1 ⎟
⎟
⎠
(1)
Namely, A is the value of concept Ci at time t, Ait −1 the
i
value of concept Ci at time t-1, Atj−1 the value of concept
73
rather than a mathematically accurate one. Thus, the use of
concepts from information theory, neural networks and
fuzzy logic [25] to represent and process information in a
hybrid and hierarchical methodology could be useful [26].
In addition to this, there is a great deal of emphasis placed
on the development of integrated and holistic solutions, it
would be therefore interesting to design efficient logistics
systems considering simultaneously all integral aspects of
their operation. Especially an approach more abstract rather
mathematical one, which will require not detailed huge
amount of data or a mathematical formulation with
thousands of variables but only structured information,
could come handy in such a complex domain. Such an
approach based on the understanding of the logistics
systems, would be particularly useful when decisions have
to be made with incomplete or uncertain information; e.g.
when evaluating a business plan, or designing a system for a
long time horizon [27].
FCMs use a symbolic representation for the description
and modeling of systems. FCMs model any system from a
behavioral point of view and can utilize an abstract
methodology to describe and model the behavior of the
system. An FCM integrates the accumulated experience and
knowledge on the operation of the system, using mainly
human experts that know the operation of system and its
behavior. They represent the human accumulated knowledge
on the operation and behavior of the system, using concepts
to stand for each characteristic of the system. Experts are
actively involved in the creation of models and they interact
with the models and so their understanding for the benefits
of models will increase the quality of models, the inherent
knowledge in the model will be used more frequently and
models will be widely accepted [28].
Furthermore, complex systems operate in changing and
unknown environments. When the environment changes, the
system has to adapt and the input-output characteristics of
the system will be altered [29]. If a single model is
identified, it will have to adapt itself to the new
environment. In non-linear systems, a single model may not
be adequate to identify the changes in the process behavior
(i.e. a model may not exist in the assumed framework to
match the environment). Hence, multiple models could be
used to identify the different environments. In some
environments different models may be available whose
validity (or accuracy) depends upon the region in the state
space where the system trajectories lie. All the above
considerations suggest that multiple models may be
preferable to a single model, in many different situations
(Figure 2).
A set of separate models is used to form hybrid models.
The rationale for using multiple models is to ensure that
there is at least one model with parameters sufficiently close
to those of the unknown process. This “multiple models”
approach possesses different modeling strategies to
accommodate different operating conditions, adaptive
behavior to perform model design under uncertain or
unfamiliar situations and the capability to co-ordinate
separate models to accomplish the system task.
C j at time t-1, and the weight W ji of the interconnection
from concept
C j to concept Ci and f is a threshold
function.
The unipolar sigmoid function is the most used threshold
function, [23] where λ Z 0 determines the steepness of the
continuous function f. The sigmoid function ensures
(equation 2) that the calculated value of each concept will
belong to the interval [0, 1].
1
f (x) =
(2
1 + e− λx
)
A more general and compact mathematical model for an
FCM is presented by the following equation:
A t = f ⎛⎜ A t − 1 c W + A t − 1 ⎞⎟
⎝
⎠
(3)
Thus, equation 3 computes the new state vector At , which
results from the multiplication of the previous, at time t-1,
state vector At −1 by the weight edge matrix W . The new
state vector holds the new values of the concepts after the
interaction among concepts of the FCM. Values of concepts
are calculated and when the FCM reaches an equilibrium
point or a limit cycle, values of concepts stop to change.
An FCM can represent the human knowledge on the
operation of the system. The developed FCM is a behavioral
model of the system, which is based on the knowledge, and
experience of the expert who described the model of the
system in terms of concepts and inter-relationships among
concepts. Using the FCM model the operation of the system
can be simulated and in each step of the interaction, the new
state vector A is computed according to equation 3.
Moreover, when the relationship between the main
factors/concepts of a system cannot be exactly extracted
from the knowledge and experience of experts or could be
further improved, then data driven approaches can be
employed such as the inclusion of learning algorithms.
Then, an FCM exploiting its universal approximation
capabilities can successfully be used to model the system.
III. LOGISTIC COMPLEX SYSTEMS AND FUZZY COGNITIVE
MAPS
Logistics Systems are complex systems requiring new
models, based on the combination of knowledge based
techniques and methodologies from various areas. Usually it
is very difficult for both researchers and managers to
achieve a clear, coherent picture of how such logistics
systems work. In most cases it requires a lot of effort and
time to obtain satisfactory solutions to the technical
problems inherent in complex logistics systems.
Furthermore, we continue to witness the development of
fundamentally new approaches to the subject, mainly
because the practical world of logistics is continuously
changing over the past few decades with the explosion of
information technologies [24]. They are characterized with
high complexity and for most of them it is more practical
and convenient to construct an abstract qualitative model
74
The main FCM-DSS task is the co-coordination of the
whole system of systems. It supervises the local subsystems,
co-ordinates the sharing of the resources, it schedules,
choosing between different distribution sequences and the
right command to the right agent [34]. The role of the FCM
is to extend the range of application of a conventional
modeling approach by using a more abstract representation,
general knowledge and adaptation heuristics and to enhance
the performance of the whole system. Symbolic
representation and processing of the supervisor of a
hierarchical system using FCMs or any other similar
approach will undoubtedly play an important role in the
construction of Intelligent Systems. Using human
knowledge and experience of the system, can lead to the
creation of a higher degree of autonomous systems.
Fig. 2. An FCM aggregating multiple models
FCMs are used to aggregate the separate models and to
perform a sophisticated approach by integrating alternative
modeling techniques. An augmented FCM can accomplish
identification of the different models and cope with limited
uncertainty situations.
IV. FUZZY COGNITIVE MAP DECISION SUPPORT SYSTEM
FOR LOGISTICS SYSTEMS
DSSs are available for various stages of supply chain
management including logistics planning, production
planning, demand management and pricing decisions. DSSs
have to cope with the overwhelming flow of data,
information and knowledge. These applications involve a
large number of dynamic objects that change their state in
time and are engaged in complex spatio-temporal relations.
It is important to understand the situation in which these
objects participate, to recognize emerging trends and to
undertake protective actions that will lead to a predefined
goal situation.
Container terminals and more specifically intermodal
container terminals involve a number of complex logistics
processes such as the spatial allocation of the containers on
the terminal yard, the allocation of resources (cranes, tracks,
personnel) and the scheduling of operations in order to
maximize some performance indicators [30],[31]. Managing
such a complex systems in an integrated manner is very
demanding and usually a “hierarchical decomposition
approach” is employed. Capturing and utilizing the expert’s
knowledge, effectively and efficiently, promises to improve
modeling operational conditions [32].
For this kind of intermodal container terminals there is a
need to design and develop DSSs, which will be able to
interpret the huge amount of data arising from the
intermodal transport system and to suggest the optimal
decisions so as to support operators in performing complex
management tasks.
An FCM could be built to model the interaction of the
different concepts comprising the processes in the terminal
along with the key performance indicators (e.g. berthing
time of vessels, the resources needed (cranes, trucks, etc),
the waiting time of customer trucks, road congestion)
(Figure 3). This approach [33] could be used to develop a
DSS.
Fig. 3. A hypothetical FCM model for an abstract logistic system for a
container terminal
In the case of intermodal container terminals, each
logistics IT centre copes with a specific process taking place
within the terminal “limits”. It is well known that the
questions concerning transportation modeling and
optimization are in most cases difficult mixed integer
programming problems making it sometimes unrealistic to
solve them to optimality let alone to “attack” them in an
integrated fashion analytically. Therefore the overall
problem of optimal terminal management is broken down
into subsystems. Each one of those logistics and informatics
subsystem manages are modeled by the local model for each
logistic center with each one of them capable to “solve” the
various resource allocation and scheduling problems that
arise within its “authority”.
Since many needs in an establishment such as a container
terminal, are already satisfied by existing systems in
logistics, it is not very useful to rebuild or re-engineer the
complete IT system [35]. Therefore a hierarchical structure
can be adopted for the development of sophisticated DSSs
for intermodal container terminals in an integrated manner.
In addition, an FCM could be used as a metamodel of an
existing DES of a logistic system and more specifically of
an intermodal container terminal. Given such a metamodel,
unsatisfactory solutions/decisions can be filtered out
whereas, on the other hand, promising solutions could be
pinpointed and returned to the DES for verification and
sensitivity analysis.
75
V. INTRODUCING FUZZY COGNITIVE MAPS AS
METAMODELS
the requirements of modern complex systems and at the
same time take advantage of existing and future
technologies.
As already explained, many real world problems are too
complex to be given tractable mathematical formulations
[36]. Especially in the area of manufacturing, supply chain
management, financial management and logistics, the
systems under investigation cannot be modeled analytically
in most of the cases [37]. For such cases, a well known
approach is Discrete Event Simulation (DES) that has been
extensively used to evaluate the behavior/performance of a
system. However the use of DES comes with a cost: each
simulation cycle can last from a few seconds up to some
minutes and since a simple evaluation (run) is often
insufficient to evaluate the system, the computational burden
and the required time can be significant. Therefore a more
exploratory process is needed in the form of simulation
optimization that can provide the operator or the decision
maker with the “optimum” decision.
Since a DES run quite slowly, one of the potential
solutions is the use of a metamodel, which is a “model of the
model” and use the metamodel to search for the optimal
decision set. Since any metamodel can run much faster than
the underlying simulation model, it can be used as a
substitute/complement to the simulation optimization
problem [38]. In this way the metamodel could become the
prediction model for the simulation in the same way that the
simulation is the prediction model for the real system
[39],[40].
Metamodels can be built using “simple” linear, quadratic
and higher order polynomials, splines, spatial correlation
(kriging) models, or more recently novel advanced methods
such as neural networks, support vector regression models
etc [40].
FCMs can be used as metamodels due to their universal
approximation property, incorporating both expert
knowledge about the system as well as data driven
approaches. FCMs can also be used along with a DES if
such a model has already been developed for a particular
system. Therefore the FCM acts as a metamodel and an
appropriate optimization procedure (depending on the nature
of the problem) can be utilized to conduct a much more
effective search of the parameter/decision space.
Especially in the case of problems involving metaheuristic
optimizers [41], as it is the case in most container terminals,
an FCM metamodel could be used to filter out solutions that
are expected to be inferior than the current best solution and
also be used as a means for generating new trial solutions.
Fig. 4. FCM metamodel filter combined with a metaheuristic optimizer and
a DES
The increase in the complexity and sophistication of large
scale logistics systems requires the implementation of new
intelligent strategies. Human experts have a key role in the
supervision of any system. Capturing the heuristic
knowledge of experts, representing, modeling and exploiting
it using FCMs may provide the foundation of new directions
in modeling complex logistics systems. The main
requirements for sophisticated systems are the possession of
human-like expertise within a specific domain, adaptation
and learning to do better in changing environments. An
FCM is a symbolic representation for the modeling of
complex systems, describing different aspects in their
behavior in terms of concepts and interactions among them.
An interesting approach is the hierarchical one, where the
supervisor of systems is modeled as an FCM that follows the
principle of "decreasing precision and increasing
intelligence" [42].
Within this framework FCMs seems a very promising tool
for the development of intelligent DSSs either by
constructing an abstract model of the whole process or as a
two stage hierarchical system. The latter approach could
take advantage of the already (mature) existing technologies
that are installed in most of container terminals world-wide,
dedicated for a specific process within the terminal limits.
Furthermore FCMs can be combined with existing DES
systems offering a flexible tool to speed up the simulation
optimization procedure. To the best of our knowledge FCMs
have not yet been used within this context, even though
other members of the soft computing family (neural
networks, support vector regression models) have already
proven the potential use of such tools.
In future work we will try to develop a hierarchical
system using an FCM at the supervisor level and test it for
the case of a port in the Adriatic sea. Another candidate
direction will be to develop and train an FCM metamodel
for an already existing DES for the same port [42].
VI. CONCLUSIONS
The soft computing modeling methodology of FCMs has
been briefly reviewed and presented. It offers essential
opportunities for design and implementation of new
advanced models suitable for complex systems such as
logistics systems. Taking advantage of the characteristics
and the abilities of the FCMs, the main needs that have to be
addressed are discussed. An important question is what
methodologies might be further developed in order to meet
76
[25] L. R. Medsker, Hybrid Intelligent Systems. Kluwer Academic
Publishers, 1995
[26] T. M. Mitchell, Machine Learning. The McGraw-Hill Companies,
Inc., 1997.
[27] C. Daganzo C. Logistics systems Analysis. Springer-Verlag Berlin
Heidelberg, 2005
[28] A. Loft and M. Howarth, “Experimental design with fuzzy levels,”.
Journal of Intelligent Manufacturing, vol. 8, no 6, pp. 525-532, 1997.
[29] M. Drouin, H. Abou-Kandil and M. Mariton, Control of complex
systems: methods and technology. Plenum press, New York, 1991.
[30] A. E. Rizzoli, N. Fornara, and L. M. Gambardella, “A simulation tool
for combined rail/road transport in intermodal terminals,”
Mathematics and Computer in Simulation. vol. 59, pp. 57-71, 2002.
[31] K. G. Murty, J. Liu, Y.-W. Wan, R. Linn, “DSS (decision support
systems) for operations in a container terminal,” Decision Support
Systems vol. 39, no 3, pp.309–332, 2005.
[32] G. Vachtsevanos and S. Kim “The role of the Human in Intelligent
Control Practices,” in Proc. of 12th IEEE International Symposium on
Intelligent Control, Instabul, Turkey, pp.15-20, 1997.
[33] G. Merkuryeva, Y. Merkuryev and J. Tolujev, “Computer Simulation
and Metamodelling of Logistics Processes at a Container Terminal,”
Studies in Informatics and Control. vol. 9, no.1, pp. 53-59, 2000.
[34] P. M. Jones and C. A. Jasek, “Intelligent support for activity
management (ISAM): an architecture to support distributed
supervisory control,” IEEE Transactions on Systems, Man, and
Cybernetics, vol. 27, pp.274-288, 1997.
[35] A. G. Bruzzone and P. Giribone, “Decision-Support Systems and
Simulation for Logistics: Moving Forward for a Distributed, RealTime, Interactive Simulation Environment,” in Proc. of the IEEE
Annual Simulation Symposium, Boston, 1998.
[36] F. Glover, J. P. Kelly, and M. Laguna, “New advances for wedding
optimization and simulation,” in Proc. of the Winter Simulation
Conference, pp. 255–260, 1999.
[37] S. Olafsson and J. Kim, “Simulation optimization,” in Proc. of the
Winter Simulation Conference, pp 79- 84, 2002.
[38] J. P. C. Kleijnen, Statistical Tools for Simulation Practitioners,
Marcel Dekker, New York, 1987.
[39] R. R. Barton, and M. Meckesheimer, “Metamodelbased simulation
optimization” in Handbooks in Operations Research and Management
Science: Simulation, S. G. Henderson and B. L. Nelson eds., chapter
18, pp 535–574. Elsevier. 2006
[40] Y. F. Li, S. H. Ng, M. Xie and T. N. Goh, “A systematic comparison
of metamodeling techniques for simulation optimization in Decision
Support Systems,” Applied Soft Computing, vol. 10, pp 1257-1273,
2010.
[41] M. Laguna and R. Marti, “Neural Network prediction in a system for
optimizing simulations,” IIE Transactions, vol. 34, pp 273-282, 2002.
[42] G. Saridis, “Analytic formulation of the principle of Increasing
Precision with Decreasing Intelligence for Intelligent Machines,”
Automatica vol. 25, vo. 3, pp. 461-467, 1989.
[43] V. Boschian, M. Dotoli, M.. P. Fanti, G. Iacobellis, and G.;W.
Ukovich, “A Metamodeling Approach to the Management of
Intermodal Terminals,” IEEE Transactions on Transportation
Networks, Automation Science and Engineering, 2010
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
S. Y. Nof, F. G. Filip, A. Molina, L. Monostori and C.E. Pereiera,
“Advances in e-Manufacturing, e-logistics and e-Service Systems”. in
Proc. Of the 17th World Congress IFAC, Seoul, Korea, July 6-11,
2008, pp. 5742-5750, 2008
M. Boccara, Modeling complex systems. Springer- Verlag New York,
Inc, 2004.
L. Camarinha-Matos and H. Afsarmanesh, “A comprehensive
modeling framework for collaborative networked organizations,”,
Journal Intelligent Manufacturing, vol. 18, pp. 529-542, 2007.
A. Wierbicki, “Modeling as a way of organizing knowledge,”
European Journal of Operational Research vol. 176, pp.610-635,
2007.
B. Kosko, “Fuzzy cognitive maps,” Int. Journal of Man-Machine
Studies, 24, pp. 65 – 75, 1986.
B. Kosko, Neural Networks and Fuzzy Systems. Prentice-Hall,
Englewood Cliffs, N.J. 1992
R. Axelrod, Structure of decision: the cognitive maps of political
elites. Princeton University Press, 1976.
M. Hagiwara, “Extended Fuzzy Cognitive Maps,” in Proc. of IEEE
International Conference on Fuzzy Systems, pp. 795-801, 1992.
W. R. Zhang, S. S Chen, W. Wang and R. S. King, “A cognitive map
based approach to the coordination of distributed cooperative agents,”
IEEE Transactions Systems, Man, and Cybernetics, vol. 22, no 1, pp.
103-114, 1992.
E. Papageorgiou, C. Stylios and P. Groumpos, “An Integrated Twolevel Hierarchical System for Decision Making in Radiation Therapy
based on Fuzzy Cognitive Maps,” IEEE Transactions on Biomedical
Engineering vol. 50, no.12, pp.1326-1339, 2003.
V. Georgopoulos and C. Stylios, “Complementary case-based
reasoning and competitive Fuzzy cognitive maps for advanced
medical decisions,” Soft Computing, vol. 12, no. 2, pp.191-199, 2008.
B. J. Juliano, “Fuzzy cognitive structures for automating human
problem solving skills diagnosis,” in Proc. 9th Annual NAFIPS Conf.,
pp. 311-314, 1990.
R. Taber,“Fuzzy cognitive maps model social systems,” AI Expert,
vol. 9, pp. 8-23, 1994.
K. Goto, T. Yamaguchi and T. Takagi, “Dynamic Model for a Plant
Using Associative Memory System,” in Proc. of the International
Conference on Artificial Neural Networks (ICANN-92), pp. 15171520, 1992.
C. E. Pelaez. and J. B. Bowles, “Using Fuzzy Cognitive Maps as a
System Model for Failure Models and Effects Analysis,” Information
Sciences, vol. 88, pp. 177-199, 1996.
C. D. Stylios and P. P. Groumpos, “The Challenge of modeling
Supervisory Systems using Fuzzy Cognitive Maps,” Journal of
Intelligent Manufacturing, vol. 9, pp. 339-345, 1998.
C. Stylios and P. Groumpos, “Modeling Complex Systems Using
Fuzzy Cognitive Maps,” IEEE Transactions on Systems, Man and
Cybernetics: Part A Systems and Humans, vol. 34, no. 1, pp. 155-162,
2004.
E. Papageorgiou, C. Stylios and P. Groumpos, “Active Hebbian
Learning Algorithm to train Fuzzy Cognitive Maps,” Int. journal of
Approximate Reasoning, vol. 37, pp. 219-249, 2004.
C. D. Stylios and P. P Groumpos, “Α soft computing approach for
modeling the supervisor of manufacturing system,s” Journal of
Intelligent and Robotics Systems, vol. 26, nο 3-4, pp. 389-403, 1999.
J. S. Jang, C. T. Sun and E. Mizutani. Neuro-Fuzzy & Soft Computing.
Prentice-Hall, Upper Saddle River, 1997.
B. Kosko, Fuzzy Engineering, Prentice-Hall, New Jersey, 1997
E. Papageorgiou, K. E. Parsopoulos, C. D. Stylios, P.P. Groumpos and
M. N. Vrahatis, “Fuzzy Cognitive Maps Learning Using Particle
Swarm Optimization,” Journal of Intelligent Information vol. 25, no 1,
pp. 95-107, 2005.
C. T. Lin and C. S. G. Lee, Neural Fuzzy Systems:A Neuro-Fuzzy
Synergism to Intelligent Systems. Prentice-Hall, Upper Saddle River,
NJ., 1996
S. Graves, A. H. G.Rinnooy Kan and P. H. Zipkin, “Logistics of
production and inventory”, Handbooks in operations research and
management science, vol.4 Elsevier BV, 1993.
77