2118

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 46, NO. 5, SEPTEMBER/OCTOBER 2010

An Adaptive Multiagent Approach to Protection

Relay Coordination With Distributed Generators in
Industrial Power Distribution System
Hui Wan, Member, IEEE, K. K. Li, Senior Member, IEEE, and K. P. Wong, Fellow, IEEE

AbstractThis paper presents new explorations into the use

of agent technology applied to the protection coordination of
power systems. The impact of distributed generators on protection
coordination is first discussed. Then, a coordination multiagent
system is proposed with the functions of the agents described. In
the proposed system, communication will play an important role
to provide more information for the relay coordination besides the
relay settings. Communication simulation has been carried out on
the Java Agent Development Framework platform. The information communication process shows that adaptive coordination can
be achieved.
Index TermsAgent technology, distributed generation, protection coordination.

I. I NTRODUCTION

N AN INDUSTRIAL power network, each protective device is assigned a primary function to clear faults in a
specific zone and a secondary function to clear faults in the
adjacent or downstream zones to the extent within the range
of the device permits. Good practice dictates that when a fault
occurs, the area isolated by the protective device must be as
small as possible, with only the device nearest to the fault
operating. In addition, the failure possibility of a protective
device must be considered. In this situation, the next upstream
device, or device combination, must operate to provide backup
(remote) protection. When two devices operate properly in
this primary/secondary mode for any system fault, they are
said to be coordinated. Proper coordination is achieved by this
discrimination between successive devices.
Traditionally, the coordination can be achieved by topology [1][4], optimization [5][7], and intelligent methods [8].
Topological analysis is employed for setting relays in multi-

Manuscript received November 15, 2005; accepted December 31, 2009. Date
of publication July 26, 2010; date of current version September 17, 2010. Paper
ICPSD-05-37, presented at the 2005 Industry Applications Society Annual
Meeting, Hong Kong, October 26, and approved for publication in the IEEE
T RANSACTIONS ON I NDUSTRY A PPLICATIONS by the Power Systems Protection Committee of the IEEE Industry Applications Society. This work was
supported by The Hong Kong Polytechnic University, Kowloon, Hong Kong
(Account Code: RG9W).
H. Wan is with the Lane Department of Computer Science and Electrical
Engineering, West Virginia University, Morgantown, WV 26506-6109 USA
(e-mail: hwan1@mix.wvu.edu).
K. K. Li and K. P. Wong are with the Department of Electrical Engineering, The Hong Kong Polytechnic University, Kowloon, Hong Kong (e-mail:
eekkli@a.polyu.edu.hk; eekpwong@polyu.edu.hk).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TIA.2010.2059492

loop networks. Graph theoretic and functional dependency

approaches are applied to provide a solution which is the best
alternative setting, but not necessarily an optimal solution. In
optimization methods, some researchers [5], [6] used nonlinear
programming for determining the optimal time dial settings
of the relays subjected to the coordination constraints and the
limits of the relay setting; other researchers [7] applied the
linear programming technique only to minimize operating time,
while the pickup currents are selected based on experience or
by using a generalized reduced gradient technique. Intelligence
methods are also used for searching the optimization relay
setting for coordination.
The relay coordination in an interconnected power system is
a tedious and a time-consuming task. To relieve the protection
engineer form this laborious task, it is proposed to set and
coordinate relays in an online manner. In adaptive coordination,
the relays should respond to the changing system conditions
and adapt according to the new prevailing conditions. The
changing system condition could be operational or topological,
such as the distributed generator (DG) connection. Communication is a major activity in an adaptive relaying system.
Traditional protection relay coordination relies on standalone
units that use local measurements and settings as the basis for
the decision making. Communication plays a very limited role
in these legacy systems. A new era of decentralized control
and efficiency demands has led to an environment demanding
efficiency and reliability that pushes these legacy methods to
their limits. In this paper, a new protection coordination model
based on agent technology is proposed after analyzing the
impact of DG on protection coordination. A new relay coordination strategy is presented. In this paper, communication simulation of pair-to-pair relay coordination has been developed
that would facilitate information exchange for communication
between relay agents themselves, among DG and relay agents,
and between equipment and relay agents.
II. AGENT T ECHNOLOGY
An agent is a computer system that is capable of autonomous
action in this environment in order to meet its design objectives.
Autonomy means that the components in an environment function solely under their own control. Agents operate and exist in
some environment, which typically is both computational and
physical. The environment provides a computation infrastructure for such interactions to take place. The infrastructure
includes communication and interaction protocols.

0093-9994/$26.00 2010 IEEE

WAN et al.: AN ADAPTIVE MULTIAGENT APPROACH TO PROTECTION RELAY COORDINATION WITH DISTRIBUTED GENERATORS

Fig. 1.

Agent in its environment.

Fig. 2.

System configuration.

Fig. 3.

System with DG1 and DG2 connected.

A communication protocol enables agents to exchange and

understand messages. It might specify that messages for a
particular course of action be exchanged between two agents.
The behavior of a multiagent depends not just on its component
agents, but also on how they interact. In a multiagent system
of sufficient complexity, each agent would not only need to be
able to do the tasks that arise locally, but would also need to
interact effectively with other agents. The communication protocols can be regarded as the specification of these interactions.
Protocols are a nice way of enforcing modularity in the design
of a multiagent system. They help in separating the interface
between agents from their internal design.
III. I MPACT OF DG ON P ROTECTION C OORDINATION
A. Fault Level
The connection of a DG to a distribution network will
inevitably result in some local changes to the characteristics of
the network. Connecting a generator to a distribution network
has the effect of increasing the fault levels in the network close
to the point of connection. Thus, the connection of distributed
generation to the network could cause a distribution network,
which happens to be close to its fault level limit, to exceed it.
The risks when fault levels are exceeded will cause damage and
failure of the plant, with consequent risk of injury to personnel
and interruption to supplies. As for protection equipment in the
system, new fault current and setting should be calculated for
the protection coordination preparation.
B. Impact on Coordination
Figs. 17 show a main industrial distribution feeder fed
through source S and protected by relays R1, R2, and R3. The
basic relay coordination philosophy here is: for the maximum

2119

fault current in line 3 (which will be a fault at bus 3), the time
of operation of relay R2 should be made larger than the
time of operation of R3 at least by a time interval called the
coordination time interval. Similarly, relays R2 and R1 are
coordinated for the maximum fault at bus 2. The nature of
inverse relay curves is such that once coordinated for maximum
current, they will be coordinated for lower fault currents too. As
clearly shown in Fig. 2, R2 will back up R3, and R1 will back
up R2.
Depending on the placement of DG on the feeder, different
situations will occur, and they are analyzed later.
1) Two Connected DGs: If only DG1 and DG2 are connected, then the maximum and minimum currents for a fault
in Section III will change. However, R3 will never sense a
backflow for an upstream fault. This will require R3 and R2
to be coordinated at a different (usually larger) current. Since
inverse relays have sufficient tap and time settings available,
this should not pose any problem.
2) Single Connected DG: If only DG3 is connected, R2 and
R3 will sense downstream current for faults in Section III and
upstream current for faults in Section I. It is important to note
here that for any given fault downstream or upstream, these
relays will sense the same fault current. This would create a
conflict. Since only the faulted section needs to be cleared, R3
requires to operate before R2 for any fault in Section III, and
R2 requires to operate before R3 for a fault in Section I. Since
these relays sense the same current for either of these faults, it
is impossible to achieve coordination with the existing scheme.
This means that the system should be facilitated with other
kinds of protection to ensure the correct fault isolation.
3) Three Connected DGs: If DG1, DG2, and DG3 are all
connected to the feeder for a fault in Section III (or further
downstream), R3 will sense the maximum fault current, followed by R2 and R1. For a fault in Section I or any other section
upstream beyond Section I, there will be more current passing
through R2 than R3.
According to the analysis of [9], the coordination impact
under this situation can be summarized as:
If the coordination relay pair detects a different current for
a downstream or upstream fault, there is a margin available
for coordination to remain valid. If disparity in fault currents
seen by the devices is more than the margin, coordination holds.
Coordination is likely to hold if the DG fault injection is more
than the margin.
IV. M ULTIAGENT P OWER S YSTEM P ROTECTION
C OORDINATION A RCHITECTURE
In this section, the multiagent architecture for power system
protection coordination is introduced. The proposed multiagent
protection coordination system, as shown in Fig. 8, consists of
relay, DG, and equipment agents. The agents can communicate
with each other not only within the same agent society, but also
within different agent societies.
The proposed architecture uses geographically distributed
agents located in a number of intelligent electronic devices
(IEDs). An IED is a hardware environment that has the necessary computational, communication, and other I/O capabilities

2120

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 46, NO. 5, SEPTEMBER/OCTOBER 2010

Fig. 4. Downstream fault case in system with DG3 connected.

Fig. 5. Upstream fault case in system with DG3 connected.

Fig. 6. Downstream fault in system with DG1, DG2, and DG3 connected.

Fig. 9.

Structure of relay agent coordination.

Fig. 7. Upstream fault in system with DG1, DG2, and DG3 connected.

relay agent searches for relevant information by communicating

with other agents. Its purpose is to detect relay misoperations,
breaker failures, and DG connection status and perform backup
protection with a much better performance than can be expected
from traditional methods. The coordination strategy will be different under different circumstances. It should also be emphasized that, since communication is available, the relay strategy
is the main methodology for relay function and coordination.
A group of agents that achieve the same relay function can
form a small society, such as the overcurrent relay agent society
and the differential relay agent society. Each group defines its
own relay roles, and roles define the relay logic associated with
them. When an agent joins a society, it takes up one or more
relay roles and acquires the relay logic of that role.
B. DG Agent
Fig. 8. Multiagent architecture for protection coordination.

needed to support a software agent. The agent takes sensory

input, which might include local measurements of the current,
voltage, and breaker status from the system and produces output
actions, such as breaker trip signals, adjusting transformer tap
settings, and switching signals in capacitor banks.
A. Relay Agent
Each relay installed in the system will be regarded as one
relay agent. The relay agent structure is shown in Fig. 9. The

The DG agent takes every single DG as one agent. In protection coordination, the DG agent mainly communicates with
the relay agent in the distribution system to provide connection
status of its own for the relay agent to coordinate.
C. Equipment Agent
The equipment agent includes the CT agent, breaker agent,
etc. These distributed equipment collect local power system
information, operate the local power system equipment, and
communicate information with the relay agent to provide protection and coordination function.

WAN et al.: AN ADAPTIVE MULTIAGENT APPROACH TO PROTECTION RELAY COORDINATION WITH DISTRIBUTED GENERATORS

2121

Fig. 10. Communication information monitoring through the sniffer agent.

V. C OMMUNICATION S IMULATION R ESULTS

In order to test the proposed system protection coordination
model, simulation was carried out based on the Java Agent
Development Framework (JADE) [12] platform. The main
purpose of the communication simulation is to make sure that
the information, which is one of the tripping determination
parameters, is exchanged correctly between different agents.
The JADE toolkit provides a FIPA-compliant agent platform
and a package to development. It is an open-source project
distributed by Telecom Italia Labs.
The communication simulation is carried out under the two
circumstances described in Section III. The relay coordination
will focus on pair-to-pair coordination, taking R2 and R3
coordination as example. On the JADE platform, a remote monitoring agent (RMA) provides control of a platform life cycle
and all registered agents within platform, acquire information
about the platform, and execute the GUI. Through the RMA,
a sniffer agent, which is a useful debugging tool, can help
in monitoring and checking the messages exchanged among
agents. When the user decides to sniff an agent or a group of
agents, every message directed to/from that agent/agent group
is tracked and displayed in the sniffer agents GUI. The user
can view every message and save it to disk. The user can also
save all the tracked messages and reload it from a single file for
larger analysis.

A. DG1 and DG2 Are Connected

Fig. 10 shows the information track among agent communications and also one of the communication message content.
The information needed to be communicated in this condition is mainly between the DG and the relay. Once the DG
is connected to the grid, the DG agent (such as DG1) will

Fig. 11. Pair-to-pair relay coordination strategy with two DGs connected.

communicate with the relative relay agent (such as R1). As

analyzed in Section III, once a fault occurred, R2 and R3 can
be coordinated using the relay settings with DG by a traditional
method.
As described in Fig. 9, the information communication will
be used to utilize the coordination strategy to coordinate R2
and R3. The coordination logic under this circumstance can be
illustrated in Figs. 11 and 12:
B. DG1, DG2, and DG3 Are Connected
The information needed to be communicated in this condition is mainly between DGs and relays, between relays and

2122

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 46, NO. 5, SEPTEMBER/OCTOBER 2010

Fig. 12. Communication information monitoring through the sniffer agent.

CTs, and between CTs and relays themselves. Once the DG

is connected to the grid, the DG agent (such as DG1, DG2, and
DG3) will communicate with the relay agent (such as R1, R2,
and R3). Once a fault occurred, for the coordination between
R2 and R3, CT2 should communicate with R2 on the sensed
fault current, and CT3 should communicate with R2 and R3 on
the fault current it sensed. Fig. 11 shows the information track
among agent communication.
R2 and R3 can be coordinated using the communication information through a coordination strategy. Based
on the DG impact summary on coordination stated in
Section III, the coordination between R2 and R3 can be
illustrated in Fig. 13. In the strategy, time is not used
as a coordination parameter; the coordination is achieved
through the communication between primary and backup relay
agents.
Taking the downstream fault case as example, first, the
downstream fault is decided by the comparison IfR2 and IfR3 .
When a downstream fault does occur, R3 will be the primary
relay, and R2 will be the backup relay. If R3 cannot operate
properly, it will communicate immediately with R2 and request
R2 to operate. R2 will accept this request and reply to R3 that
the request has been done after R2 operates correctly. Once R2
cannot fulfill the backup function, it will start the coordination
between R2 and R1.

VI. C ONCLUSION
In this paper, a multiagent approach to power system protection coordination has been proposed. The proposed multiagent
system consists of a number of relay, DG, and equipment
agents. Coordination strategy is embedded in every relay agent
to facilitate the relay agents to be coordinated under different
system situations. In the coordination strategy, relay settings
and time will not be the only parameters that will decide the
relay coordination. Relay agents communicate themselves in
the relay society and also with DG and equipment agents in
order to obtain for a successful coordination. The validity and
effectiveness of the proposed multiagent system have been
demonstrated by applying it to an agent-based JADE platform.
The communication simulation shows that the successful information communication between agents has been achieved
indicating that the proposed multiagent system is a feasible
approach in protection coordination.
The agent-based relay coordination has the ability to selfcheck and self-correct and rapidly acts while achieving highly
selective fault region backup function when either primary
protections or circuit breakers fail. The subsequent work to
be continued will be to improve the multiagent systems performance in order for it to cope with protection coordination
in a more complex system. Future work will be directed to

WAN et al.: AN ADAPTIVE MULTIAGENT APPROACH TO PROTECTION RELAY COORDINATION WITH DISTRIBUTED GENERATORS

2123

Fig. 13. Pair-to-pair relay coordination strategy with three DGs connected.

expand this approach to handle the relay coordination within

a substation and between different substations.
ACKNOWLEDGMENT
The authors would like to thank the Department of Electrical Engineering, The Hong Kong Polytechnic University,
Kowloon, Hong Kong, for the resources and facilities support.
The authors would also like to thank The Hong Kong Polytechnic University for the provision of library support.
R EFERENCES
[1] A. H. Knable, A standardized approach to relay coordination, presented
at the IEEE Winter Power Meeting, New York, 1969.
[2] M. H. Dwaraknath and L. Nowita, An application of linear graph theory for coordination of directional overcurrent relays, in Proc. SIAM
Conf. Electr. Power ProblemsThe Mathematical Challenge, Seattle,
WA, 1980, pp. 104114.
[3] M. J. Damborg, R. Ramaswami, S. S. Venkata, and J. M. Postforoosh,
Computer aided transmission protection system design, Part I: Algorithm, IEEE Trans. Power App. Syst., vol. PAS-103, no. 1, pp. 5159,
Jan. 1984.
[4] L. Jenkins, H. Khicha, S. Shivakumar, and P. Dash, An application of
function dependencies to the topological analysis of protection schemes,
IEEE Trans. Power Del., vol. 7, no. 1, pp. 7783, Jan. 1992.
[5] A. J. Urdaneta, R. Nadira, and L. G. P. Jimnez, Optimal coordination
of directional overcurrent relays in interconnected power systems, IEEE
Trans. Power Del., vol. 3, no. 3, pp. 903911, Jul. 1988.
[6] N. A Laway and H. O. Gupta, A method for adaptive coordination of
overcurrent relays in an interconnected power system, in Proc. 5th Int.
Conf. Develop. Power Syst. Protection, 1993, pp. 240243.

[7] B. Chattopadhyay, M. S. Sachdev, and T. S. Sidhu, An on-line relay

coordination algorithm for adaptive protection using linear programming technique, IEEE Trans. Power Del., vol. 11, no. 1, pp. 165173,
Jan. 1996.
[8] C. W. So and K. K. Li, Time coordination method for power system
protection by evolutionary algorithm, IEEE Trans. Ind. Appl., vol. 36,
no. 5, pp. 12351240, Sep./Oct. 2000.
[9] A. Girgis and S. Grahma, Effect of distributed generation and protective device coordination in distribution system, in Proc. LESCOPE,
Jul. 1113, 2001, pp. 115119.
[10] C. Rehtanz, Autonomous Systems and Intelligent Agents in Power System
Control and Operation. New York: Springer-Verlag, 2003.
[11] G. Weiss, Ed., Multiagent Systems: A Modern Approach to Distributed
Artificial Intelligence. Cambridge, MA: MIT Press, 1999.
[12] Telecom Italia. (2010, Jul. 7). Java Agent DEvelopment Framework.
[Online]. Available: http://jade.tilab.com

Hui Wan (S03M07) received the M.S. degree

in electrical engineering from Southeast University,
Nanjing, China, in 2002, and the Ph.D. degree in
electrical engineering from The Hong Kong Polytechnic University, Hong Kong, in 2007.
Since 2007, she has been a Research Assistant
Professor in the Lane Department of Computer Science and Electrical Engineering, West Virginia University, Morgantown. Her research interests include
power system protection and control, distributed
generation, and application of artificial intelligence
techniques in power systems.

2124

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 46, NO. 5, SEPTEMBER/OCTOBER 2010

K. K. Li (M76SM91) received the M.Sc. degree from the University of Manchester Institute of
Science and Technology, Manchester, UK, and the
Ph.D. degree from City University, London, UK.
He is currently retired. Before his retirement, he
was an Associate Professor in the Department of
Electrical Engineering, The Hong Kong Polytechnic
University, Hong Kong. His research interests are
power system protection and applications of artificial
intelligence in power systems.

K. P. Wong (M87SM90F02) received the M.Sc.

and Ph.D. degrees from the University of Manchester
Institute of Science and Technology (UMIST),
Manchester, UK, in 1972 and 1974, respectively. He
received the Higher Doctorate D.Eng. degree from
UMIST in 2001.
He was a Guest Professor at Tsinghua University,
Beijing, China and is a Guest Professor at Southeast
University, Nanjing, China and Shandong University,
Jinan, China. He was a Professor at the University of
Western Australia, Perth, Australia. He is currently a
Chair Professor in the Department of Electrical Engineering, The Hong Kong
Polytechnic University, Hong Kong, and an Adjunct Professor in the School of
Electrical, Electronic, and Computer Engineering, The University of Western
Australia. His current research interests include evolutionary optimization in
power, power market analysis, power system planning and operation in the
deregulated environment, and power quality.
Prof. Wong is a Fellow of the Institution of Electrical Engineers, U.K., The
Hong Kong Institution of Engineers, and the Institution of Engineers Australia.