IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO.

4, NOVEMBER 2007 1743

Multi-Agent Systems for Power Engineering

Applications—Part I: Concepts, Approaches,
and Technical Challenges
Stephen D. J. McArthur, Senior Member, IEEE, Euan M. Davidson, Member, IEEE,
Victoria M. Catterson, Member, IEEE, Aris L. Dimeas, Student Member, IEEE,
Nikos D. Hatziargyriou, Senior Member, IEEE, Ferdinanda Ponci, Member, IEEE, and
Toshihisa Funabashi, Senior Member, IEEE

Abstract—This is the first part of a two-part paper that has what benefits are offered by multi-agent systems? What differ-
arisen from the work of the IEEE Power Engineering Society’s entiates them from existing systems and approaches? To what
Multi-Agent Systems (MAS) Working Group. sort of problem can they be applied?
Part I of this paper examines the potential value of MAS tech-
If and when MAS technology is deemed appropriate for a
nology to the power industry. In terms of contribution, it describes
fundamental concepts and approaches within the field of multi- particular power engineering application, then other questions
agent systems that are appropriate to power engineering applica- naturally follow: how should multi-agent systems be designed?
tions. As well as presenting a comprehensive review of the mean- How should multi-agent systems be implemented? Are there
ingful power engineering applications for which MAS are being any special considerations for the application of MAS in power
investigated, it also defines the technical issues which must be ad- engineering?
dressed in order to accelerate and facilitate the uptake of the tech- The IEEE Power Engineering Society’s (PES) Intelligent
nology within the power and energy sector.
Part II of this paper explores the decisions inherent in engi- System Subcommittee (within the PSACE Committee) has
neering multi-agent systems for applications in the power and formed a Working Group to investigate these questions about
energy sector and offers guidance and recommendations on how the use of multi-agent systems. Its first remit is to define the
MAS can be designed and implemented. drivers for and benefits gained by the use of multi-agent sys-
Index Terms—Multi-agent systems. tems in the field of power engineering. As MAS are a relatively
new technology, a number of technical challenges need to
be overcome if they are to be used effectively. The Working
I. INTRODUCTION Group’s second remit is to identify and disseminate details of
those challenges. Its third and final remit is to provide technical
OR over a decade, the proposed use of multi-agent sys- leadership in terms of recommendation and guidance on the
F tems (MAS) to address challenges in power engineering
has been reported in IEEE transactions and conference papers.
appropriate use of the standards, design methodologies, and
implementation approaches which are currently available.
MAS technology is now being developed for a range of applica- This paper reports on the research of the Working Group. It
tions including diagnostics [1], condition monitoring [2], power begins by describing key concepts and approaches associated
system restoration [3], market simulation [4], [5], network con- with multi-agent systems. As a result of research and discus-
trol [5], [6], and automation [8]. Moreover, the technology is sions by the Multi-Agent Systems Working Group, definitions
maturing to the point where the first multi-agent systems are of MAS terminology and concepts have been tailored for use by
now being migrated from the laboratory to the utility, allowing the power engineering community.
industry to gain experience in the use of MAS and also to eval- The engineering drivers behind the use of MAS and the bene-
uate their effectiveness [1]. fits they may offer are presented. The recent increase in activities
Nevertheless, despite a growing awareness of the technology, in this area has led to some inappropriate uses of the technology;
some fundamental questions often arise from other researchers hence, this paper considers the principal problems which can be
and, in particular, industrial partners when discussing multi- tackled by MAS. Comparisons with existing technologies, such
agent systems and their role in power engineering. These are: as web services, grid computing, and intelligent systems tech-
niques, are drawn to illustrate how MAS differ.
Manuscript received September 26, 2006; revised May 31, 2007. Paper no.
Additionally, this part of the paper (part I) presents a com-
TPWRS-00656-2006. prehensive review of the power engineering applications for
S. D. J. McArthur, E. M. Davidson, and V. M. Catterson are with the Institute which MAS technology is being investigated, and outlines the
for Energy and Environment, University of Strathlcyde, Glasgow, U.K. (e-mail: key technical issues and research challenges which the authors
[email protected]).
A. L. Dimeas and N. D. Hatziargyriou are with the Power Division, Elec- believe need to be addressed if MAS technology is to be de-
trical and Computer Engineering Department, National Technical University of ployed within the power industry.
Athens, Athens, Greece. The uptake of multi-agent systems has increased over the last
F. Ponci is with the Electrical Systems Department, University of South Car-
olina, Columbia, SC 29208 USA.
few years, in terms of number of research projects. However,
T. Funabashi is with the Meidensha Corporation, 103-8513 Tokyo, Japan. it is essential at this stage of maturity of research into the ap-
Digital Object Identifier 10.1109/TPWRS.2007.908471 plication of MAS that appropriate standards and guidance are
0885-8950/$25.00 © 2007 IEEE
1744 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO. 4, NOVEMBER 2007

available for those developing multi-agent systems in the power designer, regardless of external changes in circumstance? Au-
engineering community; these are discussed in the companion tonomy is therefore the ability to schedule action based on en-
Part II paper. vironmental observations.
From an engineering perspective, this definition is problem-
II. CONCEPTS: TERMINOLOGY AND DEFINITIONS atic: it does not clearly distinguish agents from a number of ex-
isting software and hardware systems. Arguably, under the defi-
In order to explore the potential benefits of MAS to power en- nition above, some existing systems could be classed as agents.
gineering and the areas where their application may be justified, For example, a protection relay could be considered as an agent.
the basic concepts and approaches associated with multi-agent It is situated in its environment, i.e., the power system. It reacts
systems need to be understood. This leads us to a basic but es- to changes in it environment, i.e., changes to voltage or/and cur-
sential, and unfortunately difficult, question: what is an agent? rent. It also exhibits a degree of autonomy. Similar arguments
can be made for software systems such as Unix daemons and
A. Definition of Agency virus checkers.
The computer science community has produced myriad defi- Renaming existing systems or new systems built using ex-
nitions for what an agent is [9]–[13]. The fact that so many dif- isting technologies as “agents” offers nothing new and no con-
ferent definitions exist testifies to the difficulty in defining the crete engineering benefit. While Russell and Norvig [9] argue
notion of agency. A comparison of these definitions and discus- that “The notion of an agent is meant to be a tool for analyzing
sion of their relative merits and weaknesses, from a computer systems, not an absolute characterization that divides the world
science perspective, can be found in [14]. into agents and non-agents,” being able to distinguish agent sys-
While all the definitions referenced above differ, they all share tems from existing systems is important. There is a need to know
a basic set of concepts: the notion of an agent, its environment, how agents and multi-agent systems differ from existing sys-
and the property of autonomy. Wooldridge’s basic definition of tems and system engineering approaches. Moreover, it is the po-
an agent [13] echoes that of Russell and Norvig [9] and Maes tential advantages gained through these differences that interest
[10]. According to Wooldridge, an agent is merely “a software us as power engineers and that have motivated the exploration
(or hardware) entity that is situated in some environment and is of the application of MAS to power engineering problems.
able to autonomously react to changes in that environment.” The
environment is simply everything external to the agent. In order B. Definition of an Intelligent Agent
to be situated in an environment, at least part of the environ- In order to help differentiate MAS from existing systems, the
ment must be observable to, or alterable by, the agent. The en- authors have adopted the definition of agency as proposed by
vironment may be physical (e.g., the power system), and there- Wooldridge [13]. Wooldridge extends the concept of an agent,
fore observable through sensors, or it may be the computing en- given above, to that of an intelligent agent by extending the def-
vironment (e.g., data sources, computing resources, and other inition of autonomy to flexible autonomy. An agent which dis-
agents), observable through system calls, program invocation, plays flexible autonomy, i.e., an intelligent agent, has the fol-
and messaging. An agent may alter the environment by taking lowing three characteristics.
some action: either physically (such as closing a normally-open • Reactivity: an intelligent agent is able to react to changes
point to reconfigure a network), or otherwise (e.g., storing diag- in its environment in a timely fashion, and takes some ac-
nostic information in a database for others to access). tion based on those changes and the function it is designed
The separation of agent from environment means that agents to achieve.
are inherently distributable. Placing copies of the same agent • Pro-activeness: intelligent agents exhibit goal-directed be-
in different environments will not affect the reasoning abilities havior. Goal-directed behavior connotes that an agent will
of each agent nor the goals it was designed to achieve; rather, dynamically change its behavior in order to achieve its
the specific actions taken by each may differ due to different goals. For example, if an agent loses communication with
observations from the two environments. This means that an another agent whose services it requires to fulfill its goals,
agent can operate usefully in any environment which supports it will search for another agent that provides the same
the tasks the agent intends to perform. services. Wooldridge describes this pro-activeness as an
Under Wooldridge’s definition, an entity situated in an en- agent’s ability to “take the initiative.”
vironment is only an agent if it can act autonomously in re- • Social ability: intelligent agents are able to interact with
sponse to environmental changes. Autonomy is a somewhat elu- other intelligent agents. Social ability connotes more than
sive term, used in all definitions of agency, but rarely defined. the simple passing of data between different software and
The loosest definition of autonomy says that an agent “exercises hardware entities, something many traditional systems do.
control over its own actions” [14], meaning that it can initiate It connotes the ability to negotiate and interact in a coop-
or schedule certain actions for execution. Russell and Norvig erative manner. That ability is normally underpinned by
go further, by requiring the scheduling of actions to be in re- an agent communication language (ACL), which allows
sponse to some change in the environment, and not simply the agents to converse rather than simply pass data.
result of the agent’s in-built knowledge [9]. This requirement While an agent, in terms of our earlier definition, and many
for environmental change is in agreement with Wooldridge, and existing systems display the characteristic of reactivity, in order
makes intuitive sense; can an agent really be considered au- to be classed as an intelligent agent under Wooldridge’s defini-
tonomous if it takes action at times predefined by the agent tion, an agent must also have some form of pro-activeness and
MCARTHUR et al.: MULTI-AGENT SYSTEMS FOR POWER ENGINEERING APPLICATIONS—PART I 1745

some form of social ability. It is the goal-directed behavior of may gain a new type of sensor, and require a new data anal-
individual agents and the ability to flexibly communicate and ysis algorithm. A state-estimator system may be upgraded to
interact that set intelligent agents apart. use a faster load-flow calculation algorithm. For distribution
Not only do the characteristics of reactivity, pro-activeness, networks, a distributed network control and management
and social ability help us distinguish agents from traditional system responsible for voltage control may be extended to
hardware and software systems, it is from these characteristics, also automate restoration and the management of distributed
as shall be discussed in the following sections, that many of their generation. Importantly, a truly extensible system will allow
benefits are derived. new functionality to be added without the need to re-implement
the existing functionality.
C. Definition of a Multi-Agent System Across many applications in power engineering, there is
A multi-agent system is simply a system comprising two or also a requirement for fault tolerance and graceful degradation:
more agents or intelligent agents. It is important to recognize should part of the system fail for whatever reason, the system
that there is no overall system goal, simply the local goals of should still be able to meet its design objective or, if that is not
each separate agent. The system designer’s intentions for the possible, it should accomplish what it can without interfering
system can only be realized by including multiple intelligent with other systems.
agents, with local goals corresponding to subparts of that MAS can provide a way of building such systems. Indeed,
intention. the ability of MAS to be flexible, extensible, and fault tolerant
Depending on the definition of agency adhered to, agents in a is often part of the justification for their use. However, in order
multi-agent system may or may not have the ability to commu- for that justification to be valid, the way in which MAS pro-
nicate directly with each other. However, under Wooldridge’s vide flexibility, extensibility, and fault tolerance needs to be un-
definitions, intelligent agents must have social ability and there- derstood. The properties of agents and MAS that produce these
fore must be capable of communication with each other. qualities are examined below.
For the sake of this paper, the authors have focused on MAS 1) Benefits of Autonomy and Agent Encapsulation: An agent
where this communication is supported. This clearly differenti- encapsulates a particular task or set of functionality, in a similar
ates the type of MAS discussed in this paper from other types way to modular or object-oriented programming. This means
of systems. that the benefits of standard interfaces and information-hiding
are also available with agent programming through the use of
III. POTENTIAL BENEFITS OF MAS TECHNOLOGY AND messaging with a standard agent communication language, but
DRIVERS FOR ITS USE IN POWER ENGINEERING APPLICATIONS there is also the additional capability of autonomous action.
Recall that autonomous action means each agent is able to
To answer the question of how (and why) MAS may be ap- schedule its own activity in order to achieve its goals. In a mod-
plied in power engineering requires an understanding of the ular programming situation, external modules can call functions
basic ways MAS can be exploited. In this paper, the authors have which the module has no choice but to execute. With agent pro-
called these “approaches.” gramming, external agents can only send messages requesting
To date, MAS have a tendency to be exploited in two ways: the agent take some action: the autonomous agent can decide
as an approach to building flexible and extensible hardware/soft- whether to fulfill the request, the priority of the task, and if other
ware systems; and as a modeling approach. actions should also be scheduled. This can be useful in situations
when an agent is receiving many requests and cannot fulfill them
A. MAS as an Approach to the Construction of Robust, all within a reasonable timescale, such as with multiple requests
Flexible, and Extensible Systems for a processing-intensive task like a load-flow calculation.
There are many power engineering application areas for The autonomy of each agent and the messaging interface
which flexible and extensible solutions are beneficial. are what contribute most to flexible and extensible systems.
Flexibility connotes the ability to respond correctly to dy- Because agents are not directly linked to others, it is easy to
namic situations, and support for replication in varied situa- take one out of operation or add a new one while the others
tions (environments). This sounds very similar to autonomy and are running. Any agents interacting with the stopped one can
therefore intelligent agents should automatically be flexible; but use the standard service location facilities to locate another
if autonomy is the ability of an agent to schedule its own ac- agent that performs the same task, and by this mechanism, new
tions, flexibility relates to having a number of possible actions agents can be included within the system. The agent framework
from which to select the most appropriate. Some specific exam- provides the functionality for messaging and service location,
ples of flexible behavior would be correct handling of different meaning that new agent integration and communications are
formats of one type of data (such as temperatures in Centigrade handled without effort from the system designer.
or Fahrenheit); or the ability to construct a new plan if a par- This allows systems to be extensible: extra functionality can
ticular control action fails; or a system that can be deployed on be added simply by deploying new agents, which use service
any feeder, which senses the connection of distributed genera- location to find others to communicate with; and parts of sys-
tion and changes protection settings accordingly. tems can be upgraded by deploying a replacement agent and
Extensibility connotes the ability to easily add new func- removing the obsolete one. Flexibility also follows: the appro-
tionality to a system, augmenting or upgrading any existing priate mix of agents can be deployed to fit the details of indi-
functionality. For example, a condition monitoring system vidual situations, and flexible handling of messages between
1746 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO. 4, NOVEMBER 2007

agents allows the system to self-configure. Finally, legacy sys- 4) Fault Tolerance: Building redundancy into systems is one
tems can be incorporated within the system simply by wrapping of the standard engineering approaches to gaining fault toler-
legacy functionality in a layer of agent messaging. ance. Building redundancy into MAS simply involves providing
2) Benefits of Open MAS Architectures: An open agent archi- more than one agent with a given set of abilities. If an agent
tecture places no restrictions on the programming language or needs the services of a second agent in order to fulfill its goals,
origin of agents joining the system, and allows flexible commu- and the second agent fails, the agent can pro-actively seek an al-
nication between any agents. This is achievable through adher- ternative agent (perhaps using the Directory Facilitator) to pro-
ence to messaging standards: the separation of an agent from its vide the services it requires.
environment means that the messaging language an agent under- This redundancy may be provided by simple duplication of
stands is important for inter-agent communication, rather than each agent, possibly with distribution of duplicates across dif-
the programming language in which it was implemented. ferent computers. This would provide a tolerance to physical
An example of a set of standards for an open architecture faults, such as the loss of a network connection, or damage to
is that defined by the Foundation for Intelligent Physical a computer. Tolerance to programming-related faults would re-
Agents (FIPA)[15]. The FIPA Agent Management Reference quire a more design-intensive solution: rather than simply run-
Model covers the “framework within which FIPA agents exist,” ning two copies of a single agent, the same functionality would
defining standards for creating, locating, removing, and com- be coded differently in two agents. Various applications and op-
municating with agents. This is more generally called the agent erating environments will have differing requirements for levels
platform, and is simply one part of an agent’s environment. One of robustness and fault tolerance, and so the approach taken
requirement of an open agent architecture is that the platform must be application-specific.
places no restrictions on the creation and messaging of agents, However, the flexibility offered by an open architecture of
while a second is that some mechanism must be available for agents with good social ability easily leads to the design of a
locating particular agents or agents offering particular services fault-tolerant system.
within the platform. Under the FIPA model, this is achieved
through a separate agent called the Directory Facilitator: an B. Multi-Agent Systems as a Modeling Approach
agent which manages a searchable list of services offered by Multi-agent systems are more than a systems integration
other agents within the platform. method, they also provide a modeling approach. By offering
Early agent systems tended to be closed architectures, as one a way of viewing the world, an agent system can intuitively
set of agents would be deployed every time the system was run, represent a real-world situation of interacting entities, and give
with all communication explicitly defined by the system cre- a way of testing how complex behaviors may emerge.
ator. An example is the ARCHON system for distribution net- Natural representation of the world has previously been given
work management, originally built to integrate four legacy sys- as an advantage of object-oriented (OO) systems design, where
tems [16]. Such an architecture is said to be closed because new entities in a system are modeled as objects. This has recently
agents cannot be added to the community: even if a new agent found favor with the power engineering community in stan-
is created and run, other agents have no way of locating it and dards such as the Common Information Model (CIM) [17] and
communicating with it. A closed architecture removes the pos- IEC 61850 [18]. The main benefit of the object approach is
sibility of an extensible or flexible system, severely limiting the data-encapsulation: the particular data structures used to hold at-
benefits of using agents. tributes of an object are hidden from external objects, but are in-
How to specifically design an open agent architecture is dis- directly accessible through method calls and standard interfaces.
cussed in detail in Part II of this paper. Agent-based design adds another level of abstraction to this: not
3) Platform for Distributed Systems: An agent is distinct only are internal data structures hidden, but the “methods” (ac-
from its environment, meaning that it can be placed in different tions) an agent can perform are also hidden, yet indirectly ac-
environments and still have the same goals and abilities. How- cessible through standard messaging interfaces.
ever, the environment impacts upon which actions an agent takes This is a very natural way of modeling actors in some sys-
and in what order, as the agent autonomously schedules action tems such as markets: in a real market, actors have attributes
in response to sensor inputs and messages. (such as desired price and lowest price for a seller) and pos-
For this reason, an agent is inherently distributable, having no sible actions (e.g., start auction, accept bid) which other actors
fixed ties to its environment. In practice, distribution of agents cannot manipulate directly. Indirect access is available by, for
across a network is supported by the agent platform: the plat- example, presenting the seller with a high bid, in the hope that
form is run on every computer that will host an agent, and the it will be accepted. By modeling each market participant as a
agents are deployed within the platform as usual. To agents separate agent in a multi-agent system, it is easy to run simu-
within one platform, there is no difference between agents on lations of different market scenarios; the attributes of single or
the same computer and agents on a different computer, as the multiple market participants can be altered by changing the ini-
instances of the platform running on separate machines seam- tial conditions of one or more agents.
lessly connect and appear as a single instance. Marketplace simulation is an application in which the bene-
This means that the same set of agents can be deployed on one fits of using intelligent agents to represent autonomous actors
computer, and alternatively on multiple networked computers, are fairly clear. By modeling the behavior and communication
without modifying or changing the agent code. of individual agents, operation of the market can be studied for
MCARTHUR et al.: MULTI-AGENT SYSTEMS FOR POWER ENGINEERING APPLICATIONS—PART I 1747

emergent behavior patterns. However, many other power engi- the current standards, there is no provision for autonomy in
neering applications can usefully apply this way of viewing the web services [22]. Similarly there is no requirement for nodes
world, such as power systems operation and control. Generators in computational grids to exhibit autonomy.
have a degree of autonomy and cannot be directly affected by It is also the social ability and proactive nature of agents
external system actors, lending themselves to being represented that set them apart from grid computing and web services, so
by agents. Such an application would be using agents for both much so that MAS technology has been mooted as a mechanism
their modeling properties and also as a way of building a flex- for delivering improved web services [22] and grid computing
ible, extensible system. systems.
Through their use for systems integration or modeling, MAS Hence, applications where the use of agents is justified are
offer significantly different approaches to designing systems for normally cases where the characteristic of autonomy offers tan-
typical power and energy applications. gible benefits.
Pěchouček and Thompson provide interesting perspectives Another common question regards the difference between
on industry applications of multi-agent systems in a report from MAS and AI techniques per se, i.e., expert systems, model-
the Industry Track of the Fourth International Joint Conference based reasoning (MBR) systems, case-based reasoning systems,
on Autonomous Agents and Multi-Agent Systems (AAMAS artificial neural networks (ANNs).
2005) [19]. They indicate that most industrialists are interested This question is understandable from the perspective that the
in agents for the following applications: planning; scheduling; techniques above have been applied to similar problems (fault
resource and strategic decision making; diagnostics; control and diagnosis, condition monitoring, decision support) and that
real-time replanning; software systems integration; interoper- MAS are often seen as another AI technique. However, this
ability; knowledge integration; ontologies; and simulation and question also represents a misunderstanding, as MAS are not
modeling. Many of these underpin the applications of multi- an alternative or competitor to classical AI techniques. Indeed,
agent systems within the power industry which are discussed there are many cases in the literature where expert systems,
in this paper. ANNs, and MBR systems are used to provide agents with their
abilities to reason and achieve the goals for which they were
designed.
IV. MAS, GRID COMPUTING, WEB SERVICES, AND What MAS do provide is a framework for building hybrid
ARTIFICIAL INTELLIGENCE TECHNIQUES systems which integrate different AI techniques. Examples of
Before exploring the applications of MAS technology in where such an approach can be beneficial are fault diagnosis [1]
power engineering, it is worthwhile considering the relation- and condition monitoring [2].
ship between multi-agent systems, grid computing [20], web
services [21], and artificial intelligence techniques: what the V. BIBLIOGRAPHICAL ANALYSIS OF AGENT RESEARCH
technologies have in common and what makes them different. A bibliographical analysis of agent research was undertaken
The commonality between the first three is easiest to deal in the preparation of this paper. Its aim was to provide an
with: all three technologies offer a perspective on the prob- indication of the active areas of agent research, with respect to
lems associated with distributed computing, i.e., harnessing dis- power systems and related applications. For conferences, the
tributed hardware and software resources to complete a specific sources were restricted to the Proceedings of the Intelligent
objective or task. They all tend to support some form of mes- Systems Application to Power Systems conferences for 2001,
saging between their component parts. 2003, and 2005 [23]–[25]. This is a representative forum
How do they differ? Firstly, they differ in scope of applica- for agent-based research in the power industry. In addition,
tion. Grid computing is normally focused on harnessing hard- papers from relevant IEEE and IEE journals were sought
ware resources (computational power) to solve computationally and categorized. These included the IEEE TRANSACTIONS ON
complex problems. Web services, on the other hand, are de- POWER SYSTEMS, IEEE TRANSACTIONS ON POWER DELIVERY,
signed to offer interoperability between software systems, pro- IEEE TRANSACTIONS ON ENERGY CONVERSION, and IEEE
viding the mechanisms for the discovery of those systems and TRANSACTIONS ON EVOLUTIONARY COMPUTING. Further
their communication across a network. searches included the IEEE Power and Energy Magazine and
At first glance, web services and multi-agent systems look relevant IEE journals. All searches dated from 1998 onwards.
deceptively similar. Similar styles of interaction diagrams are These sources and timescales are representative of the body of
often used to describe web services and to describe agent inter- research undertaken in this field.
actions. The ideas of the “services” and the “brokerage of ser- Four categories of applications were discovered: monitoring
vices” are common to the technologies. However, standards for and diagnostics, distributed control, modeling and simulation,
multi-agent systems (e.g., [15]) support a richer set of interac- and protection. From the survey results in Table I, it is clear
tions, i.e., support for negotiation, than those required for the that most papers have concerned the use of agents for modeling
brokerage of services as supported by web services. So while and simulation or distributed control. This is not surprising, as
web services support the interoperability between software sys- these are two complex fields where the power industry faces real
tems, the nature of that interoperability is more limited than that challenges.
for multi-agent systems. Protection applications represent the least active area in terms
The key differentiator between multi-agent systems, grid of journal publications, with only five journal papers [26]–[30].
computing, and web services is the notion of autonomy. Under All the journals focused on monitoring and diagnostics have
1748 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO. 4, NOVEMBER 2007

TABLE I is an excellent tool for collecting and manipulating distributed

BIBLIOGRAPHIC SURVEY OF AGENT PAPERS information and knowledge.
1) Condition Monitoring: Condition monitoring of equip-
ment and plant items offers a number of challenges:
• gathering data from a variety of sensors;
• interpreting the data to extract meaningful information.
This often requires the use of multiple algorithmic and in-
telligent system-based approaches;
• combining the evidence and information from different in-
terpretation algorithms to generate an overall diagnostic
conclusion;
• delivering the diagnostic information in the correct format
arisen from the research activities at the University of Strath- to relevant engineers; and
clyde [1], [2], [31]–[34]. In terms of journal papers, there is • automatically altering power system and plant settings
a wide diversity of authors publishing work in the area of dis- based on the condition of the plant.
tributed control [3], [6], [8], [35]–[46] and modeling and simu- If we consider plant items such as transformers, there are var-
lation [47]–[62]. ious sensors which can be used to monitor them, such as UHF
monitoring of partial discharge, acoustic monitoring of partial
VI. APPLICATION OF MAS IN POWER ENGINEERING discharge, and online dissolved gas in oil measurement. Further-
As described in Section III, agent technology offers two main more, operational information about the circuit loading and fault
approaches to developing innovative applications. The four conditions from digital fault recorders can also be used to inform
broad fields of agent applications in power, identified through the diagnostic process. Agent technology allows the combina-
the bibliographical analysis, each use the property of flexible tion of data from all these sources in a flexible manner: infor-
autonomy to bring a new suite of techniques and abilities to mation is used when it is available and relevant by delegating
bear on traditional issues and problems in the industry. the task of monitoring each source to an autonomous agent.
Based on this, multi-agent systems should be considered As an example, an agent responsible for monitoring the
for applications which display one or more of the following output from UHF sensors can inform the engineer or diag-
characteristics. nostic algorithms when significant partial discharge activity
• There is a requirement for interaction between distinct con- has been detected. The autonomy of the agent allows it to
ceptual entities, such as different control subsystems and determine when such information should be communicated,
plant items, e.g., controlling a microgrid while taking ac- and to whom. The property of flexibility allows integration
count of thermal constraints, voltage control, and renew- of as much diagnostic data, information, and knowledge as is
able energy sources. currently available. New sensors and interpretation algorithms
• A very large number of entities must interact, where it can also be introduced seamlessly into the overall system, since
would be impossible to explicitly model overall system be- the open architecture allows extensibility.
havior, e.g., simulation of an energy marketplace where Using these principles, some of the authors have developed a
each individual generator, independent system operator, transformer condition monitoring multi-agent system [2].
and customer is modeled. As a further idea, condition monitoring agents could also be
• There is enough data/information available locally to un- capable of modifying the measurement set-up by, for example,
dertake an analysis/decision without the need for commu- altering the data acquisition rate. While the physical instrument
nication with a central point, e.g., substation-based diag- connection can rarely be changed, in a framework of virtual in-
nostics from transformer, switchgear, and protection anal- strumentation (e.g., LabVIEW), the monitoring agent can con-
ysis systems. trol execution of specific virtual instruments. This would bring
• New functions need to be implemented within existing advantages such as the optimization of resources like battery
plant items and control systems, e.g., extending substa- and computation power.
tion-based condition monitoring systems by adding data 2) Post-Fault Diagnosis of Power System Faults: When op-
interpretation functions. erational engineers investigate the causes and impact of power
• Over time, there is a requirement for functionality to be system faults, they employ a number of data sources. These
continually added or extended, e.g., asset management include supervisory, control and data acquisition (SCADA)
through the use of real-time condition monitoring on system data, digital fault recorder data, and traveling-wave fault
multiple plant items. locator data. In a similar fashion to the condition monitoring
The specific benefits of MAS technology for the four fields problem discussed previously, automation of the analysis of
of application are considered below. such data provides essential decision support to operational
engineers. For example, [1] reports on work with a U.K. utility
A. Monitoring and Diagnostics which experienced an influx of 15 000 SCADA alarms and
A key application area for multi-agent systems is the manage- 1695 digital fault records during a single storm. The engineers
ment and interpretation of data for a wide variety of power engi- require effective supporting analysis tools to combat such
neering monitoring and diagnostic functions. MAS technology situations.
MCARTHUR et al.: MULTI-AGENT SYSTEMS FOR POWER ENGINEERING APPLICATIONS—PART I 1749

Research into the application of intelligent systems for the data-driven simulation, and poses new problems regarding the
analysis of power systems data has been ongoing for the best dynamic real-time interaction of agents and the real world.
part of two decades and has produced a variety of tools and More recently, agent technology has been suggested for the
techniques for analyzing individual data sources. Multi-agent integration and coordination of different models and modeling
system technology can be used to integrate legacy data anal- software packages [47], [50].
ysis tools in order to enhance diagnostic support for engineers,
giving a holistic view of the performance of power systems D. Protection
based on a variety of data sources. Power system protection is an area where the analogue be-
tween agents and protective devices is being explored [26]–[30].
B. Distributed Control In all the papers above, protection relays and associated equip-
With the introduction of distributed power generation, load ment are seen as agents and their functionality augmented ac-
control, market operations, increasing complexity in the distri- cordingly. In doing so, researchers are investigating MAS tech-
bution network, and an increased number of interconnections, nology as a way of developing novel protection schemes which
the operation of a modern power system is extremely complex. are fault tolerant and self-coordinating.
Multi-agent systems provide a technology for flexibly con-
trolling the modern power system. The current approach of E. Maturity of Multi-Agent Systems in Power
using a central SCADA system and several smaller distributed While the potential application of MAS technology to power
SCADA systems is no longer sufficient for certain control engineering spans a diverse range of applications, some appli-
operations. An approach that provides intelligent, fast, and cations are more mature than others. Here, three particular ex-
adaptable local control and decision making is required. amples are highlighted to demonstrate the current maturity of
Applications currently being investigated in this field include: such systems.
• power system restoration; The first is an agent system for the control of microgrids, de-
• active distribution networks operation; veloped at the National Technical University of Athens (NTUA)
• microgrid control; and [6]. This system has progressed to a physical demonstrator,
• control of shipboard electrical systems. which has been employed successfully on a test electrical
Taking the example of active distribution networks, man- network.
agement and control of complex networks present a number Secondly, the Protection Engineering Diagnostic Agents
of challenges, not least in the scalability and flexibility of (PEDA) were developed at the University of Strathclyde for
solutions. A number of researchers are considering agent-based automating the analysis of power systems data [1]. This system
approaches as an alternative to centralized power system man- was successfully transferred from the laboratory to deployment
agement and control [6], [7]. By distributing management and at a utility, indicating that MAS technology is maturing to the
control functionality using intelligent agents, decision-making point where meaningful industrial applications are achievable.
regarding network restoration, reconfiguration, the dispatch Results of the trial and the issues surrounding the implementa-
of generation, and the management of loads can be locally tion of an industrial strength MAS are reported in [1].
managed. The third system is a commercial product: the IntelliTEAM
Local decision-making would require agents capable of a II by S&C Electric Company [65].
range of actions, such as monitoring local conditions, control-
ling switchgear and other plant, and coordinating with other VII. TECHNOLOGY CHALLENGES FOR POWER ENGINEERING
regions of the network.
While the potential benefits of agent technology have been
C. Modeling and Simulation thus far described, it is important to identify the key technical
challenges that are yet to be overcome to allow most effective
Within modern power systems, several operations are too implementation of multi-agent systems within the power engi-
complicated to model and simulate using traditional methods. neering community. These include the following.
For this reason, the use of agent systems as a modeling ap- • Platforms: a number of multi-agent system platforms
proach, introduced in Section III-B, could be beneficial to the exist. However, judicious selection is required to ensure
simulation of complex power systems, energy markets, overall long-term compatibility and the required robustness for
energy networks, and energy utilization. These applications online applications. The necessity to develop agents that
all have a common property: overall system behavior is very can interact with each other, irrespective of the platform
complex, but is generated by the interaction of simpler entities. they run on, is fundamental to the development of flexible,
This approach to modeling has been applied to energy mar- extensible, open architectures. For this reason, platform
ketplace simulation, where agents model suppliers, brokers, choice for standards-adherence is extremely important.
generators, and customers [4], [5]. Another such area is the • Toolkits: based on the increasing amount of agent research
planning of transmission [62]. A further simulation application within the power engineering community, there is the op-
uses an agent to provide simulated data to the rest of the portunity to reuse agent designs and functionality for the
multi-agent system for the purpose of “what if” scenario anal- benefit of the whole community. Therefore, there is a role
ysis—an approach used within research concerning the control for toolkits which allow the reuse of existing agent behav-
of shipboard electrical systems [63], [64]. This is similar to iors and capabilities.
1750 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO. 4, NOVEMBER 2007

• Intelligent agent design: new researchers and industrial of yet, no credible reason for using this approach is ap-
implementers need guidance on how exactly an agent parent. In [19], Pěchouček and Thompson state “People
should be designed or, at very least, knowledge of the often claim that agent mobility is inevitable and more es-
available options. A number of different concrete architec- sential than is actually the case. Often, migration of data or
tures for intelligent agents can be found in the literature: simple communication is sufficient, rather than migration
belief desire and intention (BDI) agents [13], reactive of an agent’s code.”
agents [13], agents with layered architectures [13], and Beyond technical and implementation issues described
agents implemented using model-based programming above, the lack of experience in the use of multi-agent system
[66]. Each of these implementation strategies will produce technology in industry is an obvious concern of both utilities
agents with differing degrees of reactivity, pro-activeness and manufacturers considering MAS solutions. According to
and social ability. What is not readily understood is how Wooldridge and Jennings [68], the migration of an agent system
flexible autonomy varies across these implementation from prototype to a solution that is robust and reliable enough
strategies and their suitability for different power engi- to be used in practice is a nontrivial step. This naturally leads
neering applications. to a requirement for the demonstration of MAS technology in
• Agent communication languages and ontologies: under- the industrial environment for a range of applications. Further-
pinning the social ability of agents are agent communi- more, there is also a requirement for clear communication of
cation languages. These define how agents exchange in- results from industrial trials of MAS technology, highlighting
formation, communicate, and negotiate. Within them are failures and problems as well as successes, to the wider power
protocols and content languages which allow meaningful engineering community.
messages to be composed and interpreted. International
standards are set by the Foundation for Intelligent Phys- VIII. CONCLUSION
ical Agents (FIPA) [15]. A key aspect of using agent-based This paper opened by posing two sets of questions sur-
technology is that all agents within power engineering ap- rounding multi-agent systems: broadly, “what are they?” and
plications should be able to cooperate and interoperate, “how should they be used?” In this paper (Part I of II), the
and this should be independent of the individual devel- first question has been answered, by defining the key termi-
oper. Therefore, the community must agree on the adoption nology and concepts associated with multi-agent systems,
of appropriate agent communication language standards. and identifying the important contributions that can be made
This extends to the area of ontologies [67] which define in the field of electrical power systems. Drivers and benefits
the terms and concepts which agents are able to exchange, have also been identified, and a survey of publications in IEEE
interpret, and understand. and IEE journals and relevant conferences has been used to
• Data Standards: the power engineering community has highlight the application areas for which MAS technology is
expended significant effort in defining data standards for currently being investigated. As well as the potential benefits
various application areas. One example is the Common In- of MAS technology, this part has also considered the technical
formation Model (CIM) for data exchange between Energy challenges which must be overcome through further research if
Management Systems and related applications [17]. An- MAS technology is to be successfully employed and deployed
other is the IEC 61850 Communication Networks and Sys- in the power industry.
tems in Substations standard for data exchange between Part II will tackle the second question, giving detailed tech-
intelligent electronic devices (IEDs) [18]. These standards nical recommendations of how MAS should be employed by
cannot be directly applied for agent communication, as the those building systems for power engineering applications.
conversational abilities of agents require a richer language
than a data-passing standard. However, there is potential to ACKNOWLEDGMENT
use them as a foundation for an ontology. This is explored The authors would like to thank the Multi-Agent Systems
further in Part II of this paper. Working Group members for their input, discussions, and ef-
• Security: due to the peer-to-peer nature of agent systems, forts. The discussions at meetings and panel sessions helped in
security can be a key concern. If agents are to seamlessly the creation of this paper.
join an agent community, there must be measures in place
to determine the level of trust between agents and the se- REFERENCES
curity of messaging. Agents from a rival utility may be [1] E. M. Davidson, S. D. J. McArthur, J. R. McDonald, T. Cumming, and
offered fewer services, for example, indicating the lower I. Watt, “Applying multi-agent system technology in practice: Auto-
trust placed in them. Similarly, communication between mated management and analysis of SCADA and digital fault recorder
data,” IEEE Trans. Power Syst., vol. 21, no. 2, pp. 559–567, May 2006.
two agents is open to attacks such as sender spoofing (the [2] S. D. J. McArthur, S. M. Strachan, and G. Jahn, “The design of a multi-
message purports to be from a more trusted agent) and mes- agent transformer condition monitoring system,” IEEE Trans. Power
sage modification (a message is changed while traveling Syst., vol. 19, no. 4, pp. 1845–1852, Nov. 2004.
[3] T. Nagata and H. Sasaki, “A multi-agent approach to power system
between agents, particularly in negotiation situations). restoration,” IEEE Trans. Power Syst., vol. 17, no. 2, pp. 457–462, May
• Mobility: a number of researchers are interested in mo- 2002.
bile agents, which move completely (source code and data) [4] S. E. Widergren, J. M. Roop, R. T. Guttromson, and Z. Huang, “Simu-
lating the dynamic coupling of market and physical system operations,”
from machine to machine [8]. While this has been sug- in Proc. IEEE Power Eng. Soc. General Meeting, 2004, Jun. 2004, pp.
gested within a few power engineering applications, as 748–753.
MCARTHUR et al.: MULTI-AGENT SYSTEMS FOR POWER ENGINEERING APPLICATIONS—PART I 1751

[5] D. Koesrindartoto, S. Junjie, and L. Tesfatsion, “An agent-based com- [33] E. E. Mangina, S. D. J. McArthur, J. R. McDonald, and A. Moyes,
putational laboratory for testing the economic reliability of wholesale “A multi agent system for monitoring industrial gas turbine start-up
power market designs,” in Proc. IEEE Power Eng. Soc. General sequences,” IEEE Trans. Power Syst., vol. 16, no. 3, pp. 396–401, Aug.
Meeting, 2005, Jun. 2005, pp. 931–936. 2001.
[6] A. L. Dimeas and N. D. Hatziargyriou, “Operation of a multi-agent [34] E. E. Mangina, S. D. J. McArthur, and J. R. McDonald, “Reasoning
system for microgrid control,” IEEE Trans. Power Syst., vol. 20, no. 3, with modal logic for power plant condition monitoring,” IEEE Power
pp. 1447–1455, Aug. 2005. Eng. Rev., vol. 21, no. 7, pp. 58–59, Jul. 2001.
[7] A. Korbik, S. D. J. McArthur, G. W. Ault, G. M. Burt, and J. R. Mc- [35] B. Zhao, C. X. Guo, and Y. J. Cao, “A multiagent-based particle
Donald, “Enabling active distribution networks through decentralized swarm optimization approach for optimal reactive power dispatch,”
autonomous network management,” in Proc. 18th Int. Conf. Electricity IEEE Trans. Power Syst., vol. 20, no. 2, pp. 1070–1078, May 2005.
Distribution (CIRED), Turin, June 2005. [36] M. M. Nordman and M. Lehtonen, “Distributed agent-based state esti-
[8] D. P. Buse, P. Sun, Q. H. Wu, and J. Fitch, “Agent-based substation mation for electrical distribution networks,” IEEE Trans. Power Syst.,
automation,” IEEE Power Energy Mag., vol. 1, no. 2, pp. 50–55, Mar./ vol. 20, no. 2, pp. 652–658, May 2005.
Apr. 2003. [37] M. M. Nordman and M. Lehtonen, “An agent concept for managing
[9] S. Russell and P. Norvig, Artificial Intelligence: A Modern Ap- electrical distribution networks,” IEEE Trans. Power Del., vol. 20, no.
proach. Englewood Cliffs, NJ: Prentice-Hall, 1995. 2, pp. 696–703, Apr. 2005.
[10] P. Maes, “Artificial life meets entertainment: Life-like autonomous [38] H. F. Wang, H. Li, and H. Chen, “Coordinated secondary voltage con-
agents,” Commun. ACM, vol. 38, no. 11, pp. 108–114, 1995. trol to eliminate voltage violations in power system contingencies,”
[11] L. N. Foner, “Entertaining agents: A sociological case study,” in Proc. IEEE Trans. Power Syst., vol. 18, no. 2, pp. 588–595, May 2003.
1st Int. Conf. Autonomous Agents, 1997. [39] H. Ni, G. T. Heydt, and L. Mili, “Power system stability agents using
[12] B. Hayes-Roth, “An architecture for adaptive intelligent systems,” robust wide area control,” IEEE Trans. Power Syst., vol. 17, no. 4, pp.
Artif. Intell. (Special Issue on Agents and Interactivity), vol. 72, pp. 1123–1131, Nov. 2002.
329–365, 1995. [40] J. Jung, C.-C. Liu, S. L. Tanimoto, and V. Vittal, “Adaptation in load
[13] M. Wooldridge, , G. Weiss, Ed., “Intelligent Agents,” in Multi-agent shedding under vulnerable operating conditions,” IEEE Trans. Power
Systems. Cambridge, MA: MIT Press, Apr. 1999, pp. 3–51. Syst., vol. 17, no. 4, pp. 1199–1205, Nov. 2002.
[14] S. Franklin and A. Graesser, “Is it an agent or just a program?,” Proc. [41] G. T. Heydt, C.-C. Liu, A. G. Phadke, and V. Vittal, “Solution for the
3rd Int. Workshop Agent Theories, Architectures, and Languages 1996, crisis in electric power supply,” IEEE Comput. Appl. Power, vol. 14,
New York: Springer-Verlag. no. 3, pp. 22–30, Jul. 2001.
[15] Foundation for Intelligent Physical Agents (FIPA), 2007. [Online]. [42] M. Amin, “Toward self-healing energy infrastructure systems,” IEEE
Available: http://www.fipa.org/. Comput. Appl. Power, vol. 14, no. 1, pp. 20–28, Jan. 2001.
[16] T. Wittig, N. R. Jennings, and E. M. Mandan, “ARCHON—A frame- [43] G. P. Azevedo, B. Feijo, and M. Costa, “Control centers evolve with
work for intelligent co-operations,” IEE-BCS J. Intell. Syst. Eng., vol. agent technology,” IEEE Comput. Appl. Power, vol. 13, no. 3, pp.
3, no. 3, pp. 168–179, 1994. 48–53, Jul. 2000.
[17] IEC Energy Management System Application Program Interface [44] K. Huang, S. K. Srivastava, and D. A. Cartes, “Solving the information
(EMS-API) - Part 301: Common Information Model (CIM) Base, accumulation problem in mesh structured agent system,” IEEE Trans.
2005, document IEC 61970-301.. Power Syst., vol. 22, no. 1, pp. 493–495, Feb. 2007.
[18] T. Wittig, N. R. Jennings, and E. M. Mandan, Communications Net- [45] I. S. Baxevanos and D. P. Labridis, “Implementing multiagent systems
works and Systems in Substations, document IEC 61850.. technology for power distribution network control and protection man-
[19] M. Pěchouček and S. Thompson, “Agents in Industry: The best from agement,” IEEE Trans. Power Del., vol. 22, no. 1, pp. 433–443, Jan.
the AAMAS 2005 Industry Track,” IEEE Intell. Syst., vol. 21, no. 2, 2007.
pp. 86–95, Mar./Apr. 2006. [46] M. E. Baran and I. M. El-Markabi, “A multiagent-based dispatching
[20] M. Irving, G. Taylor, and P. Hobson, “Plug in to grid computing,” IEEE scheme for distributed generators for voltage support on distribution
Intell. Syst., vol. 2, no. 2, pp. 40–44, Mar./Apr. 2004. feeders,” IEEE Trans. Power Syst., vol. 22, no. 1, pp. 52–59, Feb. 2007.
[21] Z. Jun, “Web services provide the power to integrate,” IEEE Power [47] K. Hopkinson, X. Wang, R. Giovanini, J. Thorp, K. Birman, and
Energy Mag., vol. 1, no. 6, pp. 40–49, Nov./Dec. 2003. D. Coury, “EPOCHS: A platform for agent-based electric power
[22] M. N. Huhns, “Agents as Web services,” IEEE Internet Comput., vol. and communication simulation built from commercial off-the-shelf
6, no. 4, pp. 93–95, Jul./Aug. 2002. components,” IEEE Trans. Power Syst., vol. 21, no. 2, pp. 548–558,
[23] Proc. 11th Int. Conf. Intelligent Systems Application to Power Systems, May 2006.
2001. [48] J. G. Vlachogiannis, N. D. Hatziargyiou, and K. Y. Lee, “Ant colony
[24] Proc. 12th Int. Conf. Intelligent Systems Application to Power Systems, system-based algorithm for constrained load flow problem,” IEEE
2003. Trans. Power Syst., vol. 20, no. 3, pp. 1241–1249, Aug. 2005.
[25] Proc. 13th Int. Conf. Intelligent Systems Application to Power Systems, [49] V. S. Koritarov, “Real-world market representation with agents,” IEEE
2005. Power Energy Mag., vol. 2, no. 4, pp. 39–46, Jul. 2004.
[26] S.-J. Park and J.-T. Lim, “Modelling and control of agent-based power [50] S. D. J. McArthur, E. M. Davidson, G. J. W. Dudgeon, and J. R. Mc-
protection system using supervisors,” Proc. Inst. Elect. Eng., Control Donald, “Toward a model integration methodology for advanced appli-
Theory Appl., vol. 153, no. 1, pp. 92–98, Jan. 2006. cations in power engineering,” IEEE Trans. Power Syst., vol. 18, no. 3,
[27] S. Sheng, K. K. Li, W. L. Chan, Z. Xiangjun, and D. Xianzhong, pp. 1205–1206, Aug. 2003.
“Agent-based self-healing protection system,” IEEE Trans. Power [51] M. Baran, R. Sreenath, and N. R. Mahajan, “Extending EMTDC/
Del., vol. 21, no. 2, pp. 610–618, Apr. 2006. PSCAD for simulating agent-based distributed applications,” IEEE
[28] D. V. Coury, J. S. Thorp, K. M. Hopkinson, and K. P. Birman, “An Power Eng. Rev., vol. 22, no. 12, pp. 52–54, Dec. 2002.
agent-based current differential relay for use with a utility intranet,” [52] J. M. Zolezzi and H. Rudnick, “Transmission cost allocation by coop-
IEEE Trans. Power Del., vol. 17, no. 1, pp. 47–53, Jan. 2002. erative games and coalition formation,” IEEE Trans. Power Syst., vol.
[29] C. Fukui, H. Kudo, J. Koda, K. Yabe, and Y. Tomita, “A cooperative 17, no. 4, pp. 1008–1015, Nov. 2002.
protection system with an agent model,” IEEE Trans. Power Del., vol. [53] A. L. Motto and F. D. Galiana, “Equilibrium of auction markets with
13, no. 4, pp. 1060–1066, Oct. 1998. unit commitment: The need for augmented pricing,” IEEE Trans.
[30] R. Giovanini, K. Hopkinson, D. V. Coury, and J. S. Thorp, “A primary Power Syst., vol. 17, no. 3, pp. 798–805, Aug. 2002.
and backup cooperative protection system based on wide area agents,” [54] P. Wei, Y. Yan, Y. Ni, J. Yen, and F. E. Wu, “A decentralized approach
IEEE Trans. Power Del., vol. 21, no. 3, pp. 1222–1230, Jul. 2006. for optimal wholesale cross-border trade planning using multi-agent
[31] S. D. J. McArthur, C. D. Booth, J. R. McDonald, and I. T. McFadyen, technology,” IEEE Trans. Power Syst., vol. 16, no. 4, pp. 833–838, Nov.
“An agent-based anomaly detection architecture for condition moni- 2001.
toring,” IEEE Trans. Power Syst, vol. 20, no. 4, pp. 1675–1682, Nov. [55] D. W. Bunn and F. S. Oliveira, “Agent-based simulation—An applica-
2005. tion to the new electricity trading arrangements of England and Wales,”
[32] J. A. Hossack, J. Menal, S. D. J. McArthur, and J. R. McDonald, IEEE Trans. Evol. Comput., vol. 5, no. 5, pp. 493–503, Oct. 2001.
“A multiagent architecture for protection engineering diagnostic [56] S.-J. Huang, “Enhancement of hydroelectric generation scheduling
assistance,” IEEE Trans. Power Syst., vol. 18, no. 2, pp. 639–647, May using ant colony system based optimization approaches,” IEEE Trans.
2003. Energy Convers., vol. 16, no. 3, pp. 296–301, Sep. 2001.
1752 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 22, NO. 4, NOVEMBER 2007

[57] C. S. K. Yeung, A. S. Y. Poon, and F. F. Wu, “Game theoretical multi- Victoria M. Catterson (M’03) received the B.Eng. (Hons) and Ph.D. de-
agent modelling of coalition formation for multilateral trades,” IEEE grees from the University of Strathclyde, Glasgow, U.K., in 2003 and 2007,
Trans. Power Syst., vol. 14, no. 3, pp. 929–934, Aug. 1999. respectively.
[58] V. Krishna and V. C. Ramesh, “Intelligent agents for negotiations in She is a Research Fellow within the Institute for Energy and Environment
market games. Part I. Model,” IEEE Trans. Power Syst., vol. 13, no. 3, at the University of Strathclyde. Her research interests include multi-agent sys-
pp. 1103–1108, Aug. 1998. tems, agent architectures, and condition monitoring and diagnostics.
[59] V. Krishna and V. C. Ramesh, “Intelligent agents for negotiations in
market games. Part II. Application,” IEEE Trans. Power Syst., vol. 13,
no. 3, pp. 1109–1114, Aug. 1998.
[60] T. Sueyoshi and G. R. Tadiparthi, “Agent-based approach to handle Aris L. Dimeas (S’04) was born in Athens, Greece, in 1977. He received the
business complexity in U.S. wholesale power trading,” IEEE Trans. diploma in electrical and computer engineering and the Ph.D. degree from the
Power Syst., vol. 22, no. 2, pp. 532–543, May 2007. National Technical University of Athens (NTUA), Athens, Greece, in 2001 and
[61] A. J. Bagnall and G. D. Smith, “A multiagent model of the UK market 2006, respectively.
in electricity generation,” IEEE Trans. Evol. Comput., vol. 9, no. 5, pp. He is currently a Senior Researcher at the Electric Energy Systems Labo-
522–536, Oct. 2005. ratory of NTUA. His research interests include dispersed generation, artificial
[62] J. Contreras and F. F. Wu, “Coalition formation in transmission expan- intelligence techniques in power systems, and computer applications in liberal-
sion planning,” IEEE Trans. Power Syst., vol. 14, no. 3, pp. 1144–1152, ized energy markets.
Aug. 1999. Dr. Dimeas is a member of the Technical Chamber of Greece.
[63] A. Deshmukh, F. Ponci, A. Monti, M. Riva, and L. Cristaldi, “Multi-
agent system for diagnostics, monitoring, and control of electric sys-
tems,” in Proc. 13th Int. Conf. Intelligent Systems Application to Power
Systems, 2005, pp. 201–206.
[64] K. Huang, D. A. Cartes, and S. K. Srivastava, “A multi-agent based al- Nikos D. Hatziargyriou (S’80–M’82–SM’90) was born in Athens, Greece. He
gorithm for mesh-structured shipboard power system reconfiguration,” received the diploma in electrical and mechanical engineering from the National
in Proc. 13th Int. Conf. Intelligent Systems Application to Power Sys- Technical University of Athens (NTUA), Athens, Greece, and the M.Sc. and
tems, 2005, pp. 188–193. Ph.D. degrees from the University of Manchester Institute of Science and Tech-
[65] S&C Electric Company. [Online]. Available: http://www.sandc.com/ nology, Manchester, U.K.
products/Energyline/IntelliTEAM, 2007. He is currently Executive Vice-Chairman and Deputy CEO of the Public
[66] B. C. Williams, M. D. Ingham, S. H. Chung, and P. H. Elliott, “Model- Power Corporation of Greece and Professor at the Power Division of the Elec-
based programming of intelligent embedded systems and robotic space trical and Computer Engineering Department of NTUA. His current interests
explorers,” Proc. IEEE, vol. 91, no. 1, pp. 212–237, Jan. 2003. include transmission and distribution systems, dispersed and renewable gener-
[67] T. R. Gruber, “A translation approach to portable ontology specifica- ation, and artificial intelligence techniques applied in power systems.
tions,” Knowl. Acquisition, vol. 5, no. 2, pp. 199–220, 1993. Dr. Hatziargyriou is member of the EU SmartGrids Technology Platform, of
[68] M. Wooldridge and N. R. Jennings, “Pitfalls of agent oriented develop- CIGRE SCC6, and of the Technical Chamber of Greece.
ment,” in Proc. 2nd Int. Conf. Autonomous Agents, 1998, pp. 385–391.

Ferdinanda Ponci (M’01) received the M.S. and Ph.D. degrees in electrical
engineering from Politecnico di Milano, Milan, Italy, in 1998 and 2002, respec-
Stephen D. J. McArthur (M’93–SM’07) received the B.Eng. (Hons.) and tively.
Ph.D. degrees from the University of Strathclyde, Glasgow, U.K., in 1992 and In 2003, she joined the Department of Electrical Engineering, University of
1996, respectively. South Carolina (USC), Columbia, as an Assistant Professor. She works with the
He is a Professor in the Institute for Energy and Environment, within the Power and Energy Research Group, USC, on the research and development of
Department of Electronic and Electrical Engineering. He is Director of the the electric ship: a project that is funded by the U.S. Navy. Her current research
British Energy Advanced Diagnostics Centre and Co-director of the Strathclyde interests include agent-based control, integrated environments for distributed
Centre for Applied Intelligent Systems. He currently chairs the IEEE PES measurement, and monitoring and diagnostics of electrical systems.
Multi-Agent System Working Group. His research interests include intelligent
system applications in power, condition monitoring, fault diagnosis, and
multi-agent systems.
Toshihisa Funabashi (M’90–SM’96) was born in Aichi, Japan, on March 25,
1951. He graduated in March 1975 from the Department of Electrical Engi-
neering, Nagoya University, Aichi, Japan. He received, in March 2000, a Doctor
Euan M. Davidson (M’06) received the M.Eng. degree from the University of degree from Doshisha University, Kyoto, Japan.
Strathclyde, Glasgow, U.K., in 2000. He joined Meidensha Corporation in April 1975 and has engaged in research
He is a Research Fellow at the Institute for Energy and Environment, within and development on power system transient simulation and integration of dis-
the Department of Electronic and Electrical Engineering. His research interests tributed generation. Currently, he is a Senior Engineer of the Power Systems
include: multi-agent systems; model-based reasoning; and the application of Sales and Engineering Division.
intelligent system techniques to diagnostic, power system control, operation, Dr. Funabashi is a chartered engineer in the U.K., a member of IET, and a
and management problems. member of the IEE of Japan.