AGENT-BASED APPROACH FOR ELECTRICITY

DISTRIBUTION SYSTEMS

Kimmo Salmenjoki, Yaroslav Tsaruk

Department of Computer Science, University of Vaasa

P.O. Box 700, 65101 Vaasa, Finland

Vagan Terziyan

Department of Mathematical Information Technology, University of Jyväskylä

P.O. Box 35, 40014 Jyväskylä, Finland

Marko Viitala

ABB Oy, Distribution Automation

P.O.Box 699, 65101 Vaasa, Finland

Keywords: Power distribution system, agent, semantic web, GUN, dashboard, SmartResource.

Abstract: This paper describes how semantic web and agent technologies could be used in enhancing the electricity

distribution systems. The paper starts by a brief overview of functioning of electricity distribution systems.

The introduced approaches aim at improving functionality of electricity distribution network systems and

assisting the experts by supporting automating of routine tasks in daily operations. We focus on GUN

(Global Understanding Environment) framework proposed by IOG (Industrial Ontologies Group) for

intelligent services on industrial resources. The resources and infrastructure of the electricity power network

are distributed. The interoperability, automation and integration features of GUN allow us to joint and

arrange cooperation among heterogeneous resources in electricity network domain. The interaction and

cooperation among resources in GUN platform are realized via resource agents. Based on discussions held

with the domain experts we also decided to use agent approach for automated collection of additional

information from heterogeneous resources and integrate this information to the operator interface

(Dashboard). This context information supports expert in decision making processes.

1 INTRODUCTION

The power electricity network is a complicated

structure and control of the networks’ processes is a

highly complicated and dynamic task. The software

engineering systems that exist nowadays provide

facilities for management of the power network. In

order to manage the power network the operator

needs to use several software systems and use

appropriate system tools according to the case

specific needs. By case we mean here a practical

implementation of a power electricity network. As a

tool to enhance the functionality of the electricity

network management systems via cooperation

among case specific software we are going to use the

latest IT solutions, such as semantic web and agent

technologies.

In the last years agent technology has been an

area of active research on the wide range of

application domains. Agents have been used to solve

different problems (Jennings & Wooldridge, 1998).

Especially agent technology was used in industrial

applications, where conventional software systems

and teams of human experts are assisted and

supported by agents to cope with the demands of

continuously changing and complex industrial

environment. Agents are helping to perform the

work of the power network operators and arrange

cooperation and communication among different

software systems.

382

Salmenjoki K., Tsaruk Y., Terziyan V. and Viitala M. (2007).

AGENT-BASED APPROACH FOR ELECTRICITY DISTRIBUTION SYSTEMS.

In Proceedings of the Ninth International Conference on Enterprise Information Systems - AIDSS, pages 382-389

DOI: 10.5220/0002350003820389

 SciTePress

The overall topic of using agent technology in

electricity network domain was already analyzed by

different research groups (Hines et al., 2005;

Kezunovic & Latisko, 2005). One of the frameworks

that was proposed by Cockburn is CIDIM

(Cooperating Intelligent Systems for Distribution

Systems Management) (Cockburn et al., 1992).

Their solution proposes to use the agent layer to

arrange cooperation and sharing of information

among different electricity domain expert systems.

The usage of semantic web allows to structure the

information and to represent it in machine

understandable way. The framework proposed by

IOG is to create an environment, which could unite

different industrial resources. GUN environment

(Kaykova et al., 2005) allows linking different

resources and making them co-operate. The

producers of electrical equipment have come up with

there own frameworks. ABB proposed the AIP

framework (Garcia et al., 2003).

The usage of semantic web and agent technology

can help to enhance the functionality on such areas

as fault detection and fault location. In general the

fault detection and fault location functions are

complicated tasks. In order to get accurate result a

lot of parameters need to be taken into account. In

fault detection such information as status of each

node of the electricity network, network structure

and description of the electricity network

components should be taken into account.

The results of fault location could be more

accurate, if further case related context information

could be taken into account (Tomita et al., 1998). An

example of context information could be physical

location of the network and objects nearby (forest,

lake, factory, etc.), weather conditions and other on-

going activities in the area, where the network fault

is to be located. The location of the power network

affects to fault diagnostics. In case if electricity

network has a fault and part of the network is

situated in the area nearby forest and recently there

was a forest fire, then there is a high probability that

the fault happened in that region. So, this kind of

context information provides the possibility to

enhance fault detection and fault location. The

supporting context information, which should be

used in the fault location and fault diagnosis, is

presently already available in the web, but need to

be extracted from distributed sources and converted

into semantically rich format.

The next chapter explains the basic principles of

operation in a power network. This section provides

basic information about the electricity network

management system and its operation. In the next

section we describe the set of technologies, which

should bring the added value to existing systems.

The next chapter demonstrates multi agent approach

and GUN framework in some detail. At the end of

this paper we make conclusions about the work and

point out the issues, which need to be addressed in

the future.

2 OPERATING ELECTRICITY

DISTRIBUTION SYSTEMS

The common operations for the operator of

distribution systems are: monitoring, control, fault

analysis, restoration and maintenance. In order to

access full network’s status information in existing

systems the operator requires to use several software

applications. One of the leaders in electrical

distribution field ABB company comes up with their

own tools, which may be used in other domains as

well. The monitoring and control operations are

performed mainly using MicroSCADA Pro Control

System SYS 600, whereas fault analysis and

restoration occur via MicroSCADA Pro Distribution

Management System DMS 600. In process of work

operator may need information from additional

systems. The composition of this information gives

to the user a more clear picture about the current

system status. The information from SYS and DMS

systems can be used by external application for other

purposes like energy or asset management of power

network.

Figure 1: The system and its network operator Control

Center.

The Figure 1 shows the structure of the typical

power management system and network operator’s

interface, which includes the following technical

elements: DMS 600 network diagram, Trends lines,

Event list and Process displays from the SYS 600.

SYS and DMS systems provide on-line information

about the power network status, i.e. switching state

information and network topology colored by

MicroSCADA

SYS 600

Electricity network nodes

DMS 600

Control Center

(Operator Dashboard)

AGENT-BASED APPROACH FOR ELECTRICITY DISTRIBUTION SYSTEMS

383

voltage level specific colors (ABB, 2003; ABB,

2005). When a fault occurs in the power network,

the situation will be indicated by these systems and

the related fault analysis can be made by the

operator.

Many of the faults in the power network are

however such faults, which do not result into a

persisting fault into the primary process. These

faults are recovered via protection algorithms of

Intelligent Electrical Devices (IEDs) in the

following way. When a fault is detected by the IED,

it performs the automatic recovery (high speed

automatic reclosing) of power network according to

the parameter setting values for protection and

control algorithm. This means that the opened circuit

breaker becomes reclosed after some hundred of

milliseconds since the original fault detected by the

IED. The trial will be used to stop the lightning in

the spark gap. If the reclosing is not successful, the

IED will try the reclosing once more, now after a

few minutes (delayed automatic reclosing). If this

second reclosing is not successful, the permanent

fault situation is indicated in the network

management systems. To analyze the occurred fault

situation from the IED point of view, requires that

the operator initiates the manual operation to extract

the disturbance recordings from the IEDs or then

they are triggered automatically by the SYS 600 file

system. The analysis of disturbance recording in a

graphical tool reveals the technical details of the

primary process at the fault time. Otherwise the

updated power network status will indicate those

areas of the power network, which have been

defected by the permanent fault.

Depending on what kind of equipment there appears

in the primary process, the exact fault location in the

power network may be calculated automatically or

then additional switching operations need to be

performed by the operator. The results of these

additional switching operations indicate the fault

location more accurately in the power network. This

iteration process is called fault isolation. When fault

isolation has been made it is time to apply

restoration operation towards the power network.

For this, different criteria for the decision making of

operator apply. As the above describes there are

evidently requirements to get the relevant

information out of the underlying devices and

different systems to assist the operator in performing

the correct actions in the different circumstances.

Next, these assistants of operator are briefly

described including their role in the electricity

distribution process. IEDs control and monitor the

primitive equipment of the power network, such as

circuit breakers, disconnectors, earth-switches and

measurements. Their primary task is to protect the

power network and perform the control operations

either launched from the IED panel or from the

electricity distribution management system.

Additionally these devices collect the information

from the power network related to the switching

states, measurements, faults and power quality

harmonics (Vattinen et al., 2005). These devices

operate and collect information on the high accuracy

level (in milliseconds). This information is then sent

to the electricity distribution system, such as SYS

600, based on the filtering definitions made during

the IED configuration. For the measurements it is

typical that only the values exceeding the configured

deadband value in device are sent to the system.

However, the switch position indications and other

process critical state changes (typically binary) are

always sent from the device to system level together

with accurate time stamps. This is because these are

also meaningful, when the changes in the process are

later on investigated from the system history

archives.

Additionally DMS contains the accurate information

on the physical power network (location of wooden

poles, cable types) and customers supplied by

specific feeders. This information is then utilized,

when DMS participates into fault location process

and provides the suggestion of restoration operations

to the operator. This functionality is based on the

SYS and DMS data exchange, and some information

is extracted from the IEDs, too. Next, the

applicability of semantic web and agent technologies

will be investigated in electricity distribution

systems.

3 ENHANCING THE

ELECTRICITY DISTRIBUTION

SYSTEMS

In the previous chapter we’ve discussed the practical

operation of the electricity network and electricity

network management systems. The present systems

provide functionality to monitor and partially detect

and locate the faults. To address the previously

described complexity in operation, we need to

exchange information and use external sources of

information, too.

ICEIS 2007 - International Conference on Enterprise Information Systems

384

3.1 Software Components based

Approaches

A step towards integrating systems has already

previously been made by ABB company. The AIP

(Aspect Integrator Platform) framework provided by

ABB provides a mechanism for integration of

different layers of information systems and improves

interaction among them. The different hierarchical

levels of the enterprise processes and systems of AIP

can be seen in Figure 2 (Garcia et al., 2003).

Figure 2: The overall enterprise information systems with

its levels.

In our case enterprise here can be seen to be the

whole power distribution system with all its

automation applications. The first level contains

hardware devices, which need to be controlled (e.g.

feeders of the electricity network). The second level

includes hardware components, which collect data or

perform some actions on the process level hardware

(IEDs). IEDs contain embedded sensors and

actuators together with communication. The Group

control level includes components (hardware and

software), which provide interoperability for a set of

lower level hardware components. The Process

control level includes components for interaction

among the set of hardware components and allows

to monitor and to supervise the processes. The top

levels allow us to coordinate and integrate usability

of data between different technological and business

processes.

The AIP provides means for information

representation and navigation as well as interfaces to

connect to the actual processes. The main AIP

concepts that build up automation applications on

this platform are Aspect Objects and Object Types.

Aspects and Aspect Types are implemented as COM

components with the above hierarchical structure.

This architecture is however not fully applicable in

electricity distribution domain, but is extensively

used in the automation scenarios. At present the

interoperability between different system can be

provided by OPC (OLE for Process Control)

technology only. The main advantage of this

standard is that it already allows extraction of data

produced by various system devices. The OPC uses

XML/SOAP as a standard for communication, and

the other necessary open, web based approaches for

improving this hierarchical scheme will be

addressed next.

3.2 Web Technology based Approaches

As we showed in the previous section the growing

complexity of systems together with a wide array of

supporting web-based information calls for new

means of enabling information sharing between

applications. As has been recognized by the

technology approaches and solutions proposed

above, various interchangeable information

descriptions have to be used. XML technologies

have been widely accepted by the IT and software

vendors to be a unified information description

platform for any software. This has today been

realized in the two lower levels of Figure 2 only. In

communicating this information between

interchangeable components web services (with

SOAP, WSDL and UDDI) provide integration

means only requiring TCP/IP and HTTP/HTTPS

readiness in the interacting partners.

However to sufficiently manage and coordinate the

higher levels of Figure 2 we need to provide domain

specific metadata together with XML based lower

level description and web services. This is provided

by W3C’s open standards on semantic web. We only

note here that architecturally we can still work like

Figure 2 suggests, but by semantic web, web

services and their engineering standard counterparts

we improve the information exchangeability towards

intelligent industrial services. On the lower level of

Figure 2 semantic web proposes new knowledge

interchange standards like RDF(Resource

Description Framework), OWL(Ontology Web

Language) and their extensions that will be used

with domain specific ontologies alike IEC 61850 in

improving the integration and interoperability of

software service-type systems. In the GUN platform

these are extended to RSCDF, RGBDF and RPIDF

for generic industrial engineering processes and

service descriptions. These will enable the multi

agent solutions of GUN, which will be described in

section 4.2. More technical details for this will be

provided in next chapter and in reference (Terziyan,

2003).

To achieve this the existing distribution system

components need to be extended to expose their

information in a way that it becomes applicable for

these new modern technologies. Already now it can

AGENT-BASED APPROACH FOR ELECTRICITY DISTRIBUTION SYSTEMS

385

be seen that this improves the overall expandability

and ability to integrate overall system with other

third party add-ons and evolving industry standards

like IEC 61850. Hence the semantic web, web

services provide new added value and potential

benefits for the whole domain. In the next chapter

we show how agent-based systems can be readily

applied to provide the improved operating of a

power distribution network.

4 AGENT-BASED APPROACH

FOR ELECTRICITY

DISTRIBUTION DOMAIN

4.1 Agents in Multiagent Environment

We assume that the future electricity distribution

system uses the previous hierarchical architecture

together with open web service interfaces and a

structured and semantic description of information

contained in the engineering systems as discussed

above. A software agent in our case is any

component that can operate independently and

collaboratively with other components in the

electricity distribution system presented on Figure 1.

The IEDs discussed previously are not agents yet,

because they only work with binary data and do not

presently support the improved data standards like

XML or RDF with extensions. When several agents

exist in one system, they form a MAS(Multiagent

system). In general MAS can contain hundreds of

agents, but for our demonstration environment

(Tsaruk, 2006) we restrict only to three types of

agents, which allows us to enhance the fault

detection and detection mechanisms. The overall

goal of the agent approach is to provide automated

means to enhance the electricity distribution

systems. In the first stage agents should providing

the user additional, domain context aware

information. Later this approach can be developed

further to automate some of the operator processes

described in chapter 2.

The agent technology can bring added value to the

existing electricity network management systems.

The agent in the GUN platform is more generically a

software component, which has such features as:

reactivity, proactivity and communication. The

reactivity is a feature which allows software

components to react to the changes in the

environment. The proactivity feature makes it

possible to predict the future values of the

environment’s parameters. The agent needs to have

communication ability in order to share collected

experience and cooperate with other agents (Hines et

al., 2005). The agents collaborate and cooperate in

this multi agent environment. What these more

complicated features of agents are and how agents

can collaborate in MAS we will discuss more

precisely in follow up papers.

4.2 GUN Platform

The Global Understanding Environment (GUN)

(Kaykova et al., 2005) enables interfacing of

heterogeneous resources and arranging interaction

among them. With GUN the heterogeneous

resources (devices, services, systems, human experts

and documents) become web-accessible, proactive

and cooperative. The web accessibility feature

allows the resource to use functionality of other

resources available in the web and provide it’s own

service facilities. The agent supported resources are

able to analyze their state independently from other

system components or to acquire such analysis from

remote experts or web services. The semantic web

technology allows us to realize the interoperability

among heterogeneous resources. The usage of agent

technologies allows us to monitor and control the

status of each node in a highly distributed manner

(Wong & Kalam, 1997).

In Figure 3 the agent layer of GUN environment

takes care about realization of intelligent, proactive

and collaborative behaviour of the resource. The

adapter layer performs transformation from and to

resource specific format. The proactive feature

allows it to monitor and plan its behavior towards

efficient functioning. The cooperative feature allows

agents to collaborate and share experience with other

Figure 3: Layers of the GUN architecture: resource,

adapter, agent.

ICEIS 2007 - International Conference on Enterprise Information Systems

386

resources. The functionality of the environment is

realized using agent technology as described above.

The whole set of resources within GUN are

divided into following types: service consumers,

service providers and expert (in our case the

operator). All these resources can be artificial

(material or abstract) or natural (human or other).

The service consumers will be able to proactively

monitor their own state over time within the

dynamic environment; share experiences and

knowledge among other resources; to discover and

to use appropriate service providers in order to get

some additional functionality or information.

Industrial resources (e.g. devices, experts, software

components, etc.) can be linked to above semantic

web-based environment via adapters (or interfaces),

which include (if necessary) sensors with digital

output, data structuring (e.g. XML) and semantic

adapter components. A specific agent is assigned to

each resource and it is able to monitor semantically

rich data about the state of the resource.

This generic GUN framework can be used in various

domains. In SmartResource project the GUN

concept was used for industrial device maintenance

domain see (Terziyan, 2006) for more details. There

the GUN platform is used for constant monitoring

providing necessary maintenance of the industrial

resources. In case if experience and knowledge of a

specific agent is not enough the agent discovers

other agents in the MAS environment, which

represent other components and exchange

information to act in the processes and collaborate

on further operations that are needed.

Implementation of agent technologies within GUN

framework allows later highly distributed operation

and further system dynamics (like mobility, self

configuration and dynamic life cycle) of intelligent

service components. These can act between various

platforms using decentralized service discovery,

FIPA communication protocols utilization, and other

advanced MAS-type integration/composition of

distributed services.

In software architecture the distributed and service

based architectures are beginning to replace the

previously popular component-based approaches of

section 3.1. Overall, the semantic web technologies

provide mechanisms, which allow us to realize

seamless integration of different systems. This

provides enormous potential for integration of

different levels of enterprise processes. It provides

possibility to integrate applications of monitoring

and controlling network with business and planning.

Instead of “reinventing the wheel” the GUN

platform proposes to interconnect and interact with

systems that exist nowadays. This solution should be

suitable for both customers and producers of

electricity network management systems. The

customer gets benefits of integrating different

already existing systems without big investments

like ERP systems. This is extremely important since

industrial processes are running 24 hours a day and

it is often not possible to interrupt the process due to

change or upgrade of software. The producer should

have a chance to smoothly adapt software systems

according to demand of the market.

4.3 Applying the Generic GUN

Framework in the Electricity

Distribution Domain

The previous section discussed the general concepts

of GUN MAS environment. The overall setup of the

GUN architecture was done in Figure 3. In the

electricity distribution domain the consumer of the

service is a Device with an attached GUN agent

component (see Figure 3 and previous section 4.1).

It is important to note that the Device means both

the physical industrial hardware and its

complementary agent software component residing

in the GUN platform that will supplement the

enhanced system functionalities that will be realized

in the future. At present, the Device during its life

time works and produces online data. This data need

to be carefully collected and will only be used in our

case if the network has some symptoms of the fault

as was discussed in chapter 2. These symptoms

should be recognized locally by the IED when ever

possible. The request of the Expert(human operator)

should be sent for detail description of Device status

in the fault detection cases described in chapter 2.

As data to the Expert resource Device sends the

historical values of relevant Device’s parameters.

The process of interaction among Device and Expert

is combined by the Service component. This

component is a processing unit. Based on the

information about the network parameters and

previous decision made by the Expert the Service

via its agent is also gradually learning the inherent

logic of fault detection. When the Service is fully

learned it is able to perform similar, routine like

tasks as the Expert did originally. The scenario of

interaction among resources is presented in the

Figure 4.

Using this setting GUN framework can be used in

electricity distribution field to address a variety of

real world cases in the advanced control and

management of the power distribution networks.

The experience and diagnosis about fault detection

AGENT-BASED APPROACH FOR ELECTRICITY DISTRIBUTION SYSTEMS

387

could be shared among agents. In future, the

heterogeneous network hardware and software

systems allow us to demonstrate the interoperability

and added functionality among heterogeneous

electricity power network resources and systems.

The automation of the operator’s routine tasks will

also be later done via the previously described MAS

system using agent collaboration and learning

behaviors.

Figure 4: The scenario of interaction among resources

according to GUN framework.

According to the GUN platform interaction among

resources is organized via adapters and agents

realize the behavioral functionality of the resource

(proactive, cooperative resources’ behavior). In the

pilot demo environment the part of electricity

network (feeder) represents a device to be monitored

and maintained (against system failures and faults)

as was described in Figure 4. The architecture and

scenario of interaction among components

combining Figures 1 and 4 can be found in Figure 5.

The agent of the Expert resource interacts, collects

and integrates information from other agents. One

more function of the Expert agents will be

automation of operator’s routine work. In case if

Expert agent detects that the network operator does

the same operation when he gets a specified set of

events, next time operator gets the same set of

events the agent is going to support the user in

performing this operation in a semi automated

manner.

The additional case based information is provided by

Context Information Providers (CIP). The CIP

resources are provide context information about the

network (e.g. geographical location of the network,

weather conditions, activities in the area of the

electricity network, etc.). The CIP resource is a

subtype of Service resource. One of the examples of

such service could be a shared algorithm for fault

detection and location as a web service. Nowadays

this functionality is provided by the DMS 600

software system.

Figure 5: Architecture of multiagent electrical distribution controlling environment.

Device with

online data

(

)

(

)

(

)

GUN

Expert

Service

Electricity network data

exchange

ICEIS 2007 - International Conference on Enterprise Information Systems

388

5 CONCLUSIONS

In this paper we have improved electricity power

network management and control with semantic and

agent approaches. We used the general GUN

platform in establishing a future test bed for building

highly distributed and intelligent industrial control

systems for the demanding needs of the distributed

energy systems. The device agent deployed on each

node could enhance the local diagnosis done today

by the IEDs based on the shared status of the overall

electricity network(current, voltage on the power

line). The information should be provided in both

directions: import from external sources to

electricity distribution systems and export

information or functionality of electricity network

systems to systems from other domains.

Thanks to agents learning capabilities new types of

faults could be detected. The knowledge about new

type of faults and their solution could be shared

among agents because of agents’ communication

feature. The information and functionality could be

exported to the other systems using SOA (Service

Oriented Architecture). The usage of SOA allows

simplifying the integration of heterogeneous

software systems. The GUN platform used provides

a superior platform for developing next generation

engineering solutions that transform the

heterogeneous industrial resources into smart

components that enhance both the engineering and

business needs of the distributed energy domain.

ACKNOWLEDGEMENTS

The material presented in this paper was results of

SmartResource project. The authors of this paper are

grateful to all contributors of the project: TEKES,

ABB, Metso Automation, TeliaSonera. We would

like to thank for all members of Industrial Ontology

Group for their ideas, consultation and continual

support. The test bed of this project was installed in

Technobothnia research lab premises. We appreciate

the lab personnel for their assistance and support.

REFERENCES

ABB, (2003). Specification of MicroSCADA system SYS

600.

ABB, (2005). Specification of Opera Distribute

Management System DMS 600.

Cockburn, D., Varga, Z., Jennings, N. (1992).

“Cooperating intelligent systems for electricity

distribution” In Proceedings of Expert Systems 1992,

Cambridge, UK.

Coury, D., Thorp, J., Hopkinson, K., Birman, K.,

“Improving the protection of EHV teed feeders using

local agents”.

Garcia, R., Gelle, E., Strohmeier, A. (2003). “A Software

Architecture for Industrial Automation” Technical

report IC/2003/28.

Hines, P., Liao, H., Jia, D., Talukdar S. (2005).

“Autonomous Agents and Cooperation for the Control

of Cascading Failures in Electric Grids.” Proc. of the

IEEE Conf. on Networking, Sensing, and Control.

Tuscan, Mar.

Jennings, N., Wooldridge, M. (1998). “Applying Agent

Technology”, Cambridge, UK.

Kaykova, O., Khriyenko, O., Kovtun, D., Naumenko, A.,

Terziyan, V., Zharko, A. (2005). General Adaption

Framework: Enabling Interoperability for Industrial

Web Resources, In: International Journal on Semantic

Web and Information Systems, Idea Group, ISSN:

1552-6283, Vol.1, No. 3, July-September 2005, pp.

31-63.

Kaykova, O., Khriyenko, O., Kovtun, D., Naumenko, A.,

Terziyan, V., Zharko, A., (2003) Challenges of

General Adaptation Framework for Industrial

Semantic Web, In: A. Sheth and M. Lytras (eds.),

Advanced Topics in Semantic Web, Idea Group, Vol.

1, 33 pp.

Kezunovic, M., Latisko, G. (2005). “Automated

Monitoring Functions for Improved Power System

Operation and Control,” IEEE PES 2005 General

Meeting, San Francisco, California.

Terziyan, V. (2006). “Challenges of the “Global

Understanding Environment” based on Agent

Mobility” In: V. Sugumaran (ed.), Advanced Topics in

Intelligent Information Technologies, Vol. 1, Idea

Group, 20 pp. (Book chapter, submitted 15 February

2006).

Terziyan, V. (2003). Semantic Web Services for Smart

Devices in a "Global Understanding Environment", In:

R. Meersman and Z. Tari (eds.), On the Move to

Meaningful Internet Systems 2003: OTM 2003

Workshops, Lecture Notes in Computer Science, Vol.

2889.

Tomita, Y., Fukui, C., Kudo, H., Koda, J., Yabe, K.

(1998). “A Cooperative Protection System with an

Agent Model”, IEEE Transactions on Power Delivery,

Vol.13, no 4, pp. 1060-1066.

Tsaruk, Y.(2006) “Multiagent demonstration environment.

The collection and integration of case-base context

information using agent-based approach”, Online

demonstration http://ttwin.techno.uwasa.fi/Dashboard

Vaittinen, R., Ylinen, T., Kuusisto, O. (2005). "Advanced

Distribution System" 3rd edition.

Wong, S., Kalam, A. (1997). An Agent Approach to

Designing Protection Systems. Presented at IEE

Conference Publication.

AGENT-BASED APPROACH FOR ELECTRICITY DISTRIBUTION SYSTEMS

389