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

net/publication/270214069

Distributed control system for process control using intelligent agents

Article  in  WSEAS Transactions on Systems · March 2006

CITATION READS

1 769

2 authors:

Qurban A Memon Habib-Ur Rehman

United Arab Emirates University American University of Sharjah
94 PUBLICATIONS   236 CITATIONS    39 PUBLICATIONS   461 CITATIONS   

SEE PROFILE SEE PROFILE

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

Eye ball tracking based communication and mobility with paralyzed people View project

Intelligent Process Control View project

All content following this page was uploaded by Qurban A Memon on 06 January 2015.

The user has requested enhancement of the downloaded file.

 Distributed Control System for Process Control using Intelligent Agents
QURBAN A. MEMON**, and HABIB-UR REHMAN*
EE Department, College of Engineering, UAE University, 17555, U.A.E
[email protected]**, [email protected]*

Abstract: This paper addresses intelligent communication among device entities to solve two issues in a
distributed industrial control environment: Network related bottlenecks; and intelligent control in the presence
of real time requirements. This work attempts to address customized design, development and implementation
aspects of the Distributed Control System (DCS) for process control in an environment of intelligent field
devices and Programmable Logic Controllers (PLC). The three key components; system architecture,
configuration, network and communication parameters are presented in detail. The proposed solution
architecture consists of two layer approach: use of decentralized intelligent agents at the local process level,
and use of real time operating system at the main controller level. The problems and challenges faced by such
a customized design of DCS are discussed and possible remedies are presented.

Keywords— Distributed control, Network control system, Profibus, Intelligent agents, Profinet

communication protocol [7]-[15] (ii) DCS system

1 Background hardware and software architecture [16]-[19].
The need for modularity, integrated diagnostic, The research and development in the area of DCS
decentralization of control, expanding physical network design, the configuration and
setups, and functionality resulted into today's DCS communication protocol is quite active because of
[1]-[15] which is most widely being used for many the latest developments in the communications
applications. Presently, the DCS concept is not field. For example in [7], the authors carried out
only applied to the process control and simulation and experimental evaluation of three
manufacturing automation but it has covered many control networks commonly used in the industry.
other areas like power system engineering [1]-[3], The control networks evaluated are: Ethernet,
nuclear engineering [4], ship engineering [5], and ControlNet, and DeviceNet. The evaluation
even a lighting system design [6]. The innovative procedure considered how each control network is
ideas are being applied in this field, for example used as a communication backbone for a network
[6] has made use of the power line as the controlled system connecting sensors, actuators,
communication medium for the DCS of the and controllers with a detailed discussion on how
lighting system and is based on a custom built medium access control (MAC) protocol works for
large scale integrated Neuron chip which each network. The various performance
incorporates both the control and communication parameters that are addressed include access delay,
features in a three microprocessors based system. transmission time, message delay, message
However, the focus of this work is the DCS collisions, message throughput, packet size,
designed for process control implemented using network utilization, and determinism boundaries.
PLC and intelligent field devices. In the DCS, Timing analysis of each control network has been
multipoint and intelligent Network Controlled presented that addresses blocking time, frame time
System (NCS) has replaced the traditional point- and propagation time. Using some case studies,
to- point architecture which had been used by the the authors conclude that with shorter or
industry for decades. The new architecture has prioritized messages, DeviceNet outperforms the
smaller volume of wiring and distributed other two networks where as ControlNet is
processing resulting into quick and easy suitable in case of time critical and non-time-
maintenance and low cost. This development critical messages. As the research carried out in
initiated research in two main directions: (i) this work is mainly comparative study based on
Network design and configuration; and simulation and experimental setup of conventional
control networks, this work does provide insight in restriction outperforms Profibus protocol under
setting performance limits on the respective bursty conditions, and if the ring is stable. This state
networks. of keeping logical ring stable in presence of
Regarding Profibus networks, a comprehensive transmission errors is difficult to achieve.
study has been carried out in [8], where authors In addition to network protocols or control
discuss how Profibus fieldbus networks can be networks with associated delay characteristics, a
used to support real time communication for study [11] has been carried out about design
industrial systems. The major contribution is that consideration of a specific network like DCS,
the real time behavior is guaranteed even in the where the impact of network architecture in
absence of synchronous bandwidth allocation. The Network Control System (NCS) class of DCS has
two proposed approaches i.e., an un-constrained been discussed. Design considerations include
low-priority traffic profile; and a constrained low- network parameters, control parameters, and
priority traffic profile, are based on respective network controlled system performance. The main
specific assumption, which when satisfied then the evaluation measures for the network quality of
real time traffic is guaranteed. The author in [9] service (QoS) are: time delay statistics, network
has performed experimental evaluation of efficiency, network utilization, and the number of
Profibus-FMS. In this work, the experimental lost or unsent messages. The authors have used a
model of a Profibus-based industrial application is simulation model and an experimental study to
developed. The experimental environment consists validate the performance characteristics shown by
of devices connected in a Profibus network and theoretical analysis. Based on the results, the
exchange data through the services provided by authors demonstrate that the performance
FMS. The model evaluates the delay characteristics are useful guidelines to choose
characteristics of a message once it passes through network and control parameters while designing
FMS service. Based on the experimental results, an NCS. As the performance characteristics is
the author suggests an optimal value of target shown to be dependent upon sampling period,
token rotation to reduce the delay of variable hence the work still lacks the criterion that helps to
messages, and an appropriate segment size to choose the best sampling period.
reduce delay in domain management service of A different approach to optimize the delay
FMS. This also concludes that Profibus-FMS characteristics by use of decreased
service is greatly affected by normal/conventional communications in distributed control systems is
values of target rotation time and segment size. studied in [12]. The authors estimate the outputs at
In case of wireless Profibus with wireless and wired each node of the network and then the expected
devices, a study has been carried out in [10], where communication frequency of the system is predicted.
the parameters of wireless Profibus protocol It is assumed that the identical estimators are
standard have been discussed. The author introduces available at each node for this function, and thus this
two approaches to wireless Profibus and compares approach helps in computing the control algorithms
them with respect to real time performance in the at each node by comparing the estimated value with
presence of transmission errors. The work the true value of the output. Thus the increased
demonstrates that the performance of real time computation is traded for decreased
Profibus protocol in IEEE 802.11 type PHY communications. This type of approach may not be
environment or in presence of bursty errors is much useful for wired nodes in the presence of high
unsatisfactory. The author suggests a specifically speed networks available today. The only viability
tailored protocol for wireless stations in a situation that seems to be of use with this kind of approach is
where integration of wired and wireless stations is with wireless nodes where transmission errors are
necessary in a common Profibus Local Area relatively high compared to wired nodes, and thus it
Network (LAN). The tailored protocol works on top may be preferable to reduce communications to
of IEEE 802.11 PHY and offers the same link layer offset the impact of errors in a situation when
interface to upper layers as does the Profibus relatively higher number of nodes start
protocol. The author has shown that round robin communicating.
protocol with three add-ons namely simple data In addition to the well known related works
relaying, simple poll relaying, and priority presented above, recently more focus has been
reported in literature with respect to distributed helps in building intelligent collaboration of agents,
intelligence where network nodes or devices and thus rules for activation and collaboration of
behave and contribute independently without agents can be easily embedded in architecture,
much dependence on centralized or built-in which carries internal and external operational tasks.
intelligence. The automatic reconfiguration The distributed real time control application
resulting in a predictable and stable behavior is development for task allocation in heterogeneous
another requirement of such systems because if systems is also reported in [15]. The authors address
any hardware needs to be added or removed the the situation where real time requirements are high
overall system should not be disturbed and the in presence of interoperability of networked devices.
Such an environment requires correct partitioning
modification process should remain transparent.
and allocation of application parts to the system
The advantages gained are reduced operational
resource as well as feasible schedule. The approach
changes that are often complex and difficult. This
used in this work considers functional block
initiated the agent based approach [13]-[14] in the allocation (FBALL) algorithm [16], defined to
area of DCS. In [13], the authors propose a general guarantee real time requirements. This work mainly
approach for dynamic and intelligent addresses problems related to real-timeliness in the
reconfiguration of real-time distributed control environment of interoperability. The authors build
system that utilizes the IEC 61499 function block an application specification that does not involve
model [16]. The approach takes advantage of network considerations or re-configurability of the
distributed artificial intelligence at the planning process itself.
and control levels to achieve significantly shorter The PLC are the most widely used processing unit
up-front commissioning time as well as and fieldbus is most widely used to interconnect the
significantly more responsive to changes. The process controllers, sensors and actuators in a DCS
authors do not address the criterion of handling system. The profibus is the standard protocol for
various delay characteristics that may occur communication which is vendor independent so
because of environmental or protocol changes in that the communication between devices from
the network. In order to simplify the re- different manufacturers can be implemented
configurability of control system, the criterion without any special interface adjustment. To
needs to be included in the overall behavior of the realize the plug and play type of operation, the
control system. Thus, the absence of network DCS vendors ought to adopt certain standards both
consideration in the overall behavior together with for PLC and the communication. The International
the use of agent based model, as in [13], may Electrotechnical Commission (IEC) and PLC
degrade the performance of the model in the manufacturing companies are actively involved in
situation of intrusive external agents. Another this development and establishment of standards
contribution using the similar approach has been [16]-[19]. These standards define basic concepts
reported in [14], where methodologies and tools for the design of modular, reusable, distributed
for intelligent agents in DCS environment are components for the DCS. Some of the most
discussed. The authors present a novel distributive popular IEC standards are IEC 61131, IEC 61499
intelligent control architecture using agents to and IEC 61804. Because of the importance of
handle combinatorial complexity in real time, as these standards a brief introduction is presented in
classical control do not adapt dynamically well to the next section.
the variability of the process. The suggested agents
are decoupled software entities to mimic human
2 Standards Development for DCS
brain i.e., programmed with set of heuristics to help
The following presents a brief overview of the
agents reason locally about their represented
standards and their application.
physical equipment; have social capabilities to help
and ask for additional capabilities when they IEC 61804: This standard addresses the need of
discover that their individual contributions can not common device description for representing the
fulfill the specific task. Because no central particular features of the filed devices used in the
government rules or controls the systems behavior, DCS system. The standard is applicable to
the agents act socially to enable formation of Function Blocks (FB) for process control and
decision making clusters. This kind of approach specifies the Electronic Device Description
Language (EDDL) thus improving the device information processing. The other approaches
interoperability. propose distributed intelligence in the form of
IEC 61131: This standard has different versions framework for network control system. Network
ranging from 61131-1 to 61131-8. The whole speeds available today are in the order of Gbps and
range covers general information just about the with emergence of PROFINET standard, most of
PLC, the equipment requirement and tests, defines the network delay parameters are of less
and specifies syntax and semantics of importance. However, the consideration of
programming languages including the function network parameters included in the framework of
block (FB) portion of standard for PLC languages, distributed intelligence is likely to make it a strong
and sets the standards for the communication. To candidate for distributed intelligent control system
conclude IEC 61131 was first attempt to provide in a situation where network nodes/sensors may be
common terminology and reference model about non-stationary, protocol invariant and
hardware (HW) and software (SW) architectures, reconfigurable.
communication, and languages for a special class The process under consideration can be visualized
of control devices i.e., the PLC. by Figure 1. In this figure, the controller is the
IEC 61499: This standard defines a generic main device for which distributed control system
architecture and presents guidelines for the use of is to be investigated. The complete system
function blocks as the main building block for involves interconnection of control devices, some
industrial process measurement and control system distributed field devices, and intelligent devices.
applications. It extends the FB language (defined All devices considered are connected to Profibus.
in IEC 61131-3) to more adequately meet the The control device at lower level of the system
requirements of the distributed control in a format may involve standard PLC for motion control of a
that is independent of implementation. The typical drive, for example. Normally, the
architecture is presented in terms of implement- intelligent devices are needed for a specific
able reference models, textual syntax and application, where some of the decisions and local
graphical representations. The standard addresses information gathering is done at that level and for
the need for modular software that can be used for onward transmission to the main controller to
distributed industrial process control by defining make final network control decision. As stated
the basic concepts for the design of modular, earlier, these devices are reprogrammable and
reusable, distributed components for industrial reconfigurable to suit changing needs of the
control systems. It also specifies the software tool process control. The interface shown in the figure
requirements. is for legacy intelligence devices which are now
part of the main distributed control system. The
3 Introduction to the Proposed Approach network considered is typical with Profibus
Although we surveyed the literature at length and compatible switches or network devices with
found other considerable approaches but the ones wired or wireless connectivity to a remote device
addressed in the previous section are the notable or a group of distributed or intelligent devices. The
among many others. The issues addressed in personal computer (PC) shown is for management
literature covered network considerations (i.e., purposes, especially for supervisory control and
protocol change or delay characteristics at various data acquisition (SCADA) system. All the field
stages of the corresponding interfaces), devices considered are IEC 61804 compliant and
interoperability of devices, customization and PLC are IEC 61131 compliant. The requirements
reconfiguration of the control system using for DCS development for this process control
distributed intelligence, and application include consideration of reconfigurability and
specification development. The approaches that intelligence in the form of framework within a
address the network environment analyzed timing Profibus compatible network. In the next section,
characteristics or protocols in order to reduce we intend to propose such a framework.
communication delays at various stages of
 Controller (Motion control through PLC)

PC (for SCADA) Drive

Profibus-enabled
Profibus-enabled
device
device

Profibus-enabled
device

Interface

Distributed field device

Distributed intelligent devices

Distributed (legacy)field devices
Figure 1: Prceess involving different PLC and independent devices for distributed network control

4 Proposed Approach 4.1 Local Process

Based on discussions in previous section, we Knowledge integration at the global level requires
discuss our approach for providing a distributive balanced information from local entities. These
control using an intelligent approach over a entities tend to be distributed throughout the
process involving many devices, for example in working environment to support overall
Figure 1. The idea is to provide an intelligent operations. The job of these entities can be done
control using a set of processes. These processes effectively by distributed agents. Agent-based
together will help minimize communications control at the local level can help to handle
between devices and the main controller; and combinatorial complexity, and provide framework
provide a degree of configurability of devices at to integrate knowledge [20]-[21]. The agents
the same time. The minimization of collaborate socially, learn and adjust their abilities
communications will help thwart any network within the constraints of the global process. Thus
bottlenecks arising because of interoperability of these agents are sophisticated software entities set
devices, network protocol change, or simple by global process for enabling trained intelligence
operational requirements at the device level. The at the local level. The functions within this process
configurability of devices will be provided by include: how agents/entities collaborate and
collecting operational parameters at the device(s) clusters are formed to accumulate intelligence and
level, and then estimating new characteristics of enable decision making; how access to another
concerned entities at the global level. Using local agent is facilitated; and the procedure for inter-
intelligence collected through interactions and cluster collaboration is established. To make sure
combined with main controller requirements at the that these agents form an intelligent system, the
global level will help in solving combinatorial framework has to conform to four requirements
complexity present at any time during the [14], that is: the system is decentralized; agents
operations. Since the focus of this work is physical follow centricity; they use common language; and
environment with its own variations and noises, it that the system is scalable. In order to highlight
imposes space limitations for advanced reasoning distributed agent mechanism, we discuss agent
mechanism for automated intelligence. Thus, two design mechanism and domain ontology in next
areas are targeted: social interactions of entities paragraphs.
with domain intelligence, and effective decision Agent design mechanism: There is already more
making set by a main process. Following is the work done on agents alone and details of their
discussion on both processes. definitions, possible roles and life cycle can be
found in [22]-[23]. In this work, we do not intend Cluster 1 -----
to recreate the work but desire to highlight their Cluster 2 .......
Agent
framework in the context of customized 1

distributed intelligent control. Agents, as discussed CD

before are active software entities that have social
capabilities to interact and can also request for CF1
additional capabilities once they discover that the Agent
task at hand can not be fulfilled with existing ones. 3

The programming of these agents is done at the Agent

5
global level where a set of heuristics is used for Agent
2
reasoning at the local level, and is stored as a
function block diagram (like an internal script).
The agents know about their equipment,
continuously monitor state, and thus can decide
whether to participate in a mission or not. Thus the
CF3
design of an agent is not only dependent upon Agent
6
operational parameters of a typical device, but is
Agent
normally made open ended to add flexibility CD: Cluster Directory 4
during operation. Once task is assigned to an CF: Cluster Facilitator
agent, it interacts with other agents using a
specified mechanism as shown in Figure 2, for Figure 2: Agents' Collaboration
example. • CFN passes the request to specified agents, and
The collaborating agents join (on their own will) thus cluster is formed.
and thus form a cluster in order to enable a The life cycle of the cluster is pre-set at the global
decision making. Within their constraints, they process during agent design, and which can be re-
accomplish their task. An agent may collaborate negotiated upon agent N’s request.
on concurrent clusters, as depicted in Figure 2. For efficient collaboration, CD must remain
These clusters may not exist in any specific updated for recording the information of its
location, but are virtual entities of a limited members, such as agent name, agent locator,
lifecycle performing a common goal. The agents service name, service type, and so on. Upon
have also limited lifecycle ranging from joining or leaving the cluster, an agent must
milliseconds to n number of seconds, and can register or cancel registration respectively through
request for an extended life cycle beyond current CF. Through a query, an agent can find out other
cluster. In addition to agents, there are other members’ services and locators. Through these
computing units that exist at the local level and steps, a trust is developed and thus members hold
which help to form a cluster. These are known as higher authority than non-members.
Cluster Directory (CD) and Cluster Facilitator Agent Cluster: Recently developed tools can be
(CF). The following steps describe the operational used to help design cluster facilitator and domain
scene of agent’s collaboration in clusters: ontology, using for example DARPA Markup
• Agent N receives a request from global process Language (DAML) [24]. The DAML extends
for a task planning. XML (Extensible Markup Language) and RDF
• It checks its internal scripts, if it can (Resource Description Framework) to include
participate then it solves the local steps. domain ontology. It provides rich set of constructs
• For external steps, it contacts cluster directory to create ontology and to markup information for
to check for other agents if they also have attaining machine readability and
external capability. understandability; it has capability of inference so
• In that case, CD provides contact details. that membership or service of cluster can be
• Upon receipt of these details, agent N creates precisely defined. In case of XML and RDF, there
CFN and passes on these details. CFN seems to be no significant literature to manage
understands coordination context. ontology, and this is why, we have chosen DAML
to build cluster ontology. Furthermore, we extend Cluster Ontology

the Foundation for Intelligent Physical Agent

Domain
(FIPA) agent management specification [25] to File access

}
Ontology
develop the agent role called CF to manage cluster
DAML File(s)
directory and cluster ontology. CF
Using assistance from DAML-based ontology, the Cluster
members of the cluster are able to form clusters Directory
and communicate with other agents, as shown in Figure 3. Linking CF with DAML
Figure 2. The interaction among domain ontology,
CD and CF can be best understood using Figure 3. 1. <cluster:CF rdf:ID="theCF">
Figure 3 shows how CF gets access to DAML files 2. <cluster:agentName>"CF"</cluster:agentName>
and facilitates the common goal of the cluster. 3. <cluster:agentDescription>
4. "DCS Cluster Facilitator"
There are tools available like Jena semantic web
5. </cluster:agentDescription>
[26] that can be used to handle the Cluster 6. <cluster:locator>
Directory built using DAML, and to develop a 7. "http://dcs.ee.uaeu.ac.ae/DCS/agent/CF"
Java class “Directory”. 8. </cluster:locator>
From discussions above, the main functions of CD 9. </cluster:CF>
can be summarized, as: 10.
11. <cluster:Cluster rdf:ID="DCSCluster">
• Add the information of an agent.
12.<cluster:clusterName>"DCS"</cluster:clusterName>
• Remove the information of an agent. 13. <cluster:clusterDescription>
• Get the list of agent names of all members. 14. "Distributed Control System"
• Get the information of individual agent by 15. </cluster:clusterDescription>
name. 16. <cluster ontology>
17. "http://dcs.ee.uaeu.ac.ae/DCS/ontology/dcs.daml"
• Get ontology used by members in the cluster.
18. </cluster:ontology>
• Add external ontology if provided by an agent. 19.
Keeping this local process in perspective and 20. <cluster:hasCF rdf:Resource="#theCF"/>
using the main functions of CD, the partial 21. <cluster:consistOf rdf:Resource="#agent1"/>
directory can be described as shown in Figure 4. 22. <cluster:consistOf rdf:Resource="#agent2"/>
The Figure 4 shows information of CF (lines 1-9) 23. </cluster:Cluster>
and members of cluster (lines 20-22), the cluster
Figure 4: DCS Cluster Directory
directory also records meta-data about cluster such
as cluster name (line 12), cluster description (lines from the Figure 5 that when distributed field
13-15), ontology used in cluster (lines 16-18), and device agent joins the cluster, it informs CF about
so on. corresponding ontology it provides (Figure 5(a)).
Thus the CF maintains local process ontology plus
An Example: Here, we consider a distributed the distributed field device ontology. This means
control environment and consider how a that ontology can be updated (Figure 5(b)), when
distributed field device agent may help achieve ever any agent joins the cluster to perform a
task of a user agent. It should be reminded here common goal. When user agent wants to perform
that all members’ domain knowledge (ontology) a task, it asks CF about domain ontology and the
may not be the same, and that some agents may agents that provide external capability. In
provide knowledge of the specific field device. In response, CF informs the user agent if ontology is
fact, all agents hold basic cluster ontology (the to be acquired (Figure 5(c)). Thus, the user agent
knowledge of the local process) provided by CF, can communicate with the distributed field device
that is, for example, user agent holds basic agent (Figure 5(d)).
knowledge of the local process but does not
understand the knowledge that a distributed field 4.2 Global Process
device holds. Through DAML-based ontology, This process handles core mechanism that glues
members can communicate with other to acquire organization’s local process to main global
requested service, as shown in Figure 5. It is clear process. This process may handle many functions
for example: how library of information collecting User
agents is to be created; what do these agents do; agent
d
how do they communicate with each other; how c
deadlocks are to be removed; definition of main DCS ontology

controller task; estimation of relationships among a DFD

agent
agents; security of agents at the process level; and CF

above all intelligent decision making once all that

information is available. Some of the functions for b ..........> File access
DFD ontology -------> Agent Communication
example library of entities, their job description, DFD : Distributed Field Device
definition of controller tasks, and domain ontology
can be defined offline before the implementation
actually starts. The dynamic components are Figure 5: Update in ontology provided by distributed field
device agent
removing of deadlocks, estimation of
characteristics and relationships; and intelligent Engineering
decision making. It seems that all of these dynamic wizard simulation

functions together may require high computations, object model

subsystem visual object/agent
but the advantages gained are many: (i) Reduced interface modeling
monitoring

communications between main controller and the

device(s) to offset any network bottlenecks and Implementation
provide simplicity to enable better interoperability modules
function block
runtime
Resource
allocation,
gurrantee
(ii) Intelligence gathering to build a degree of scheduling

reconfigurability in case estimated parameters device

exceed beyond a limit (iii) Reduced human
supervision. It can be also argued that complexity application

of this process is only a technology mismatch, and kernel and runtime system security
that if only small scale changes are to be decided
at the global level like reconfiguration of device memory

processes or security of agents, then intelligence Hardware drivers Network drivers

error
can be further distributed to the agents at local
level.
In Figure 6, we have drawn a model operating Hardware
system (OS) to guarantee real time processing of
sample functions at the global level. The operating
system is separated into three layers in order to Figure 6: Real time OS using function block and modules
provide ease and simplicity at engineering and
implementation levels. The engineering layer 5 Conclusions
provides design of various objects needed for main We have presented a customized framework to
controller and for simulation levels of provide intelligence at device process level using
reconfigurability. The monitoring of agents decentralized agents. The agents’ communication
behavior, its parameters and cluster characteristics and cluster behavior is facilitated through cluster
are done at this layer to assess the degree of facilitator, which updates its domain ontology
reconfigurability. The implementation layer takes whenever an agent joins the cluster. The agents are
care of parameters like handling device processes, reconfigurable scripts, and their domain
applications, security of agents, resource allocation parameters are controllable through global
and scheduling of processes. The hardware process. The idea behind splitting the process into
abstraction layer provides various compatible two processes is customization: to maximize
hardware and network drivers for interoperability decentralization at the local process; to handle
of devices. reconfigurability at operating system of the main
controller. There exists a possibility of introducing
intelligence at global level to automate
reconfigurability. This change in framework [9]. S. Hong, “Experimental Performance
would bring monitoring and some of the Evaluation of Profibus-FMS”, IEEE Robotics and
modeling/simulation functions at implementation Automation Magazine, pp. 64-72, December 2000.
level of the operating system. Currently, we are [10]. A. Willig, “Polling based MAC
investigating on these lines, and the findings will protocols for improving real-time performance in a
be reported in future. wireless Profibus”, IEEE Transactions on
Industrial Electronics, Vol. 50, No. 4, pp. 806-
References: 817, August 2003.
[1]. M. G. Ioannides, “Design and Implementation [11]. F. Lian, J. Moyne, D. Tilbury, “Network
of PLC based Monitoring Control System for Design Consideration for Distributed Control
Induction Motor”, IEEE Transactions on Energy Systems”, IEEE Transactions on Control Systems
Conversion, Vol. 19, No. 3, pp. 469-476, 2004. Technology, Vol. 10, No. 2, pp. 297-307, March
[2]. G. Prasad, E. Swidenbank, B. W. Hogg, “A 2002.
novel performance monitoring strategy for [12]. J. Yook, D. Tilbury, and N. Soparkar,
economical thermal power plant operation”, IEEE “Trading Computation for bandwidth: Reducing
Transactions on Energy Conversion, Volume 14, Communication in distributed control systems
No. 3, pp. 802-809 Sept. 1999. using state estimators”, IEEE Transactions on
[3]. R. E. Putman, F.C. Huff, J. K. Pal, “Optimal Control Systems Technology, Vol. 10, No. 4, pp.
reactive power control for industrial power 503-518, July 2002.
networks” IEEE Transactions on Industry [13]. Bernnan, M. Fletcher, and D. H. Norrie,
Applications, Volume 35, No. 3, pp. 506–514, “An agent-based approach to reconfiguration of
May-June 1999. the real-time distributed control systems,” IEEE
[4]. M. Buchler, W. Funk, A. Gutierrez, R. F. Transactions on Robotics and Automation, Vol.
Harr, P. E. Karchin, P. Liu, S. Nam, J. G. Shiu, S. 18, No. 4, August 2002.
F. Takach, M. Atiya, D. Padrazo, A. Arefiev, S. [14]. F. Maturana, R. Staron, K. Hall,
Barsuk, F. Khasanov, L. Laptin, V. Tchoudakov, “Methodologies and Tools for Intelligent Agents
I. Tichimirov, M. Titov, Y. Zaitsev, “Design and in Distributed Control”, IEEE Intelligent Systems,
operation of front-end electronics for the HERA-B pp. 42-49, February 2005.
muon detector”, IEEE Transactions on Nuclear [15]. A. Prayati, C. Koulamas, S. Koubias,
Science, Volume 46, No. 3, Part 1, pp. 126–132. and G. Papadopoulos, “A methodology for the
June 1999. development of distributed real-time control
[5]. H. Chang-kun, “A distributed control system applications with focus on task allocation in
of ship diesels”, Proceedings of the IEEE heterogeneous systems,” IEEE Transactions on
International Conference on Industrial Industrial Electronics, Vol. 51, No. 6, pp. 1194-
Technology, Dec. 1996. 1206, December 2004.
[6]. J. Alonso, J. Ribas, J. Coz, A. Calleja, E. [16]. Function Blocks (FB) for Process
Corominas, “Development of a Distributive Control, IEC 61804, 2000.
Control Scheme for Fluorescent Lighting based on [17]. IEC 61131
LonWorks Technology”, IEEE Transactions on [18]. IEC 61131-3
Industrial Electronic, Vol. 47, No. 6, pp. 1253- [19]. Function blocks for industrial-process
1262, December 2000. measurement and control systems, IEC 61499,
[7]. F. Lian, J. Moyne, and D. Tilbury, 2001.
“Performance evaluation of control networks”, [20]. M. Ahmed, “Meeting the Challenges of
IEEE Control Systems Magazine, pp. 66-83, Complexity”, InTech, Vol. 51, Oct. 2004, pp.20-
February 2001. 23.
[8]. E. Tovar, and F. Francisco, “Real-time field [21]. R. Staron, et al, “Use of an Agent Type
bus communications using Profibus Networks”, Library for the Design and Implementation of
IEEE Transactions on Industrial Electronics, Vol. Highly Flexible Control Systems”, Proceedings of
46, No. 6, pp. 1241-1251, December 1999. 8th World Multiconference on Systemics,
Cybernetics, and Informatics, 2004, pp. 256-261.
[22]. R. Brennan and D. Norrie, “Agents, [25]. Foundation for Intelligent Physical
holons and function blocks: Distributed intelligent Agents (December 2002), FIPA Agent
control in manufacturing”, Journal of Applied Management Specification [Online].
Systems Studies, Vol. 2, No. 1, pp. 1-19, 2001. http://www.fipa.org/specs/fipa00023/; accessed
[23]. M. Wooldridge and N. Jennings, September 2005.
“Intelligent Agents: Theory and Practice”, [26]. HP Labs (January 2003). [Online]. Jena
Knowledge Engineering Review, Vol. 10, No. 2, Semantic Web Toolkit Available:
pp.115-152, 1995. http://www.hpl.hp.com/semweb/jena.htm/;
[24]. DARPA Agent Markup Language accessed September 2005.
(August, 2000) [Online]. http://www.daml.org/;
accessed August 2005.

View publication stats