A CASE-BASED ENTERPRISE INFORMATION SYSTEM

FOR THERMAL POWER PLANTS’ SAFETY ASSESSMENT

Dong-xiao Gu

1,6

, Chang-yong Liang

1,3,5

, Chun-rong Zuo

1,5,6

, Isabelle Bichindaritz

Jun Wang

and Wen-en Chen

1,5

School of Management at Hefei University of Technology, 193 Tunxi Road, Hefei, P.R. China

Institute of Technology at University of Washington, 1900 Commerce Street, Tacoma, U.S.A.

Department of Computer science at University of Illinois at Chicago, 851 S. Morgan Street, Chicago, U.S.A.

Department of Computer science at University of Wisconsin at Milwaukee, 2200 E. Kenwood Blvd, Milwaukee, U.S.A.

Engineering Research Center of Intelligent Decision-making

and Information System Technology of Ministry of Education of China, 193 Tunxi Road, Hefei, P.R. China

Key Laboratory of Process Optimization and Intelligent Decision-making of Ministry of Education of China

193 Tunxi Road, Hefei, P.R. China

Keywords: Thermal Power Plants, Enterprise Information System, Case-based Reasoning, Decision Support System,

Grey System Theory, Information Management, Safety Evaluation.

Abstract: Security assessment of Thermal Power Plants (TPP) is one of the important means to guarantee the safety of

production in thermal power production enterprises. Modern information technology may play a more

important role in TPP safety assessment. Essentially, the evaluation of power plant systems relies to a large

extent on the knowledge and length of experience of the experts. Therefore in this domain Case-Based

Reasoning (CBR) is introduced for the security assessment of TPPs since this methodology models

expertise through experience management. This paper provides a case-based approach for the management

system security assessment decision making of TPPs (MSSATPP). A case matching method named CBR-

Grey is introduced in which Delphi approach and Grey System theory are integrated. Based on this method,

we implement a prototype of enterprise assessment information system (CBRSYS-TPP) for the panel of

experts.

1 INTRODUCTION

Thermal Power Plants (TPPs) equip numerous

industrial departments and their productive process

is very complicated. In TPPs, the frequency of

accidents with serious consequences is extremely

high. When operating TPPs, the safety of people’s

lives and work conditions is a major concern. There

are numerous TPPs all over the world. Taking China

as an example, there are over 1200 coal-fired

thermal plants (Yang, Guo and Wang, 1999). As one

of the strongest nations in electric power generation,

due to various limitations and causes, China

produces its electric power mainly from coal

(Williams, 2001). In Turkey as well, 80% of the

total electricity is generated from thermal power

plants (Oktay, 2009). For the purpose of reducing

major and extraordinarily large accidents in TPPs

and ensuring the security of electric power

production, an increasing number of thermal power

enterprises in China pay more attention to the

security assessment issue.

Security assessment of TPPs mainly concerns

three different aspects: Production Equipment

Systems (PES), Working Circumstance Systems

(WCS), and Management Systems in production.

The latter is also referred to as the Management

System (MS) in current research. By the analysis

and evaluation of these three subsystems, the TPPs

establish the necessary corrective, remedial, and

preventive measure, and finally realize the aim of

controlling the accidents in advance. As one of

modern management ladders, safety assessment of

TPPs is one of powerful tools for automatically

diagnosing safety issues. However, numerous

evaluations for production safety are irregular,

unscientific, and capricious, as well as lacking

Gu D., Liang C., Zuo C., Bichindaritz I., Wang J. and Chen W..

A CASE-BASED ENTERPRISE INFORMATION SYSTEM FOR THERMAL POWER PLANTS’ SAFETY ASSESSMENT .

DOI: 10.5220/0003434600320039

In Proceedings of the 13th International Conference on Enterprise Information Systems (ICEIS-2011), pages 32-39

ISBN: 978-989-8425-54-6

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

powerful information and knowledge support. Accordingly, there is a sizable margin of error.

Along with the increasing perfection of security

assessment rules and the development of

information technologies, new techniques are being

applied to almost all aspects of power systems to

improve efficiency (Zhao, Wang, Nielsen, Li and

Hao, 2010). It is of both major significance and

profound social consequences for TPPs to make

their security assessment process progress toward

the quantification, scientization, and automatization.

MS security represents an important part of the

security issue in the production of TPPs. Numerous

facts show that a large part of safety accidents in

TPPs occurred due to the managerial inadequateness

and not for the equipment malfunctions.

First, common evaluation issues concerning the

power industry have been reported in the literature.

In view of the special importance of production

security for TPPs, it is important to study scientific

approaches that fit the characteristic features of the

production and management of TPPs for security

assessment. However, few research studies focus on

the safety assessment of TPPs in production - the

inside security itself. Most of the literature focuses

on the operational performance (Liu, Lin, Sue and

Lewis, 2010)，energetic and exergetic performance

analyses (Erdem, Ali, Burhanettin, et al, 2009), the

selection of an optimum power plant (Garg, Agrawal

and Gupta, 2007), air quality impact (Kumar, Mahur,

Sengupta, Prasad, 2005; Petkovšek, Batič and

Lasnik, 2008), and ecological efficiency (Lora,

Salomon, 2005). Second, as far as content

assessment is concerned, few studies concern safety

evaluation of management work. In terms of

evaluation approaches, few approaches are actually

able to solve the problems of providing powerful

and helpful information support for experts’ decision

making and the reuse of domain knowledge. Until

now, rare contributions have been made to the

assessment approaches for management security of

thermal power plants. As an important technology in

artificial intelligence, CBR can provide an

information support for the whole process of

MSSATPP decision making. Part of its advantage

lies in that it can capture expert knowledge, provide

methods for knowledge management, and give

suggestions for fast problem-solving. Different from

ANNs and decision trees, CBR can address the

problem of over fitting.

In the area of evaluation research, there are also

many articles concerning CBR, such as the

applications of CBR to software cost estimation

(Zhuang, Churilov, Burstein and Sikaris, 2009),

software effort estimation (Mukhopadhyay,

Vicinanaza and Prieutula, 1992), risk assessment in

audit judgment (Chang, Lai and Robert, 2006) ，

risk analysis for electronic commerce (Jung, Han

and Shu, 1999), web break sensitivity evaluation in a

paper machine (Ahola and Leiviskä, 2005), safety

risk analysis in information security systems (Bang,

Kim and Hwang, 2008), safety evaluation of process

configuration (Gu, Liang, Li, et al, 2010), and so

forth. In this article, we apply CBR to MSSATPP,

and propose a whole evaluation approach integrating

weight derivation approaches and case retrieval

algorithms for MSSATPP. The research novelty of

our work lies in that by taking the management

system of whole power systems as an example, we

integrate Grey System Theory and Delphi method

into case-based reasoning, and then apply the

optimized CBR to enterprise information system for

MSSATPP (CBRSYS-TPP).

2 BACKGROUND

Power plant safety evaluations are performed by

panels of experts through investigation, discussion,

and negotiation. This process is explained in this

section, as well as the motivations for building the

CBRSYS-TPP system.

2.1 TPP Safety Evaluation Process

Security assessment is one of the important

measures and safeguards for enforcing the electric

security basis in TPP production and for

guaranteeing safe, stable, and economical TPP

operation. As an important part of the whole security

assessment work of TPPs, MSSATPP is an all-

around examination and evaluation of the safety

management work in the production of TPPs. Two

different parts are involved in the security

assessment of TPPs: inside evaluation and outside

foreign expert evaluation, respectively. The former

is operated by a thermal power plant itself. Power

companies organize expert groups with relevant

personnel to evaluate their safety status, identify

issues, and then propose revision suggestions

according to the evaluation index, standard, or

criterion. The latter is generally organized by the

electric power company responsible for a group of

TPPs. To do so, the electric power companies

organize audits in which relevant experts complete

their evaluation work. To prepare for the actual

audits performed by the electric power companies,

most of the electric power incorporations currently

A CASE-BASED ENTERPRISE INFORMATION SYSTEM FOR THERMAL POWER PLANTS' SAFETY

ASSESSMENT

complete their internal thermal power plants safety

evaluation work through external experts’

evaluation. The complete evaluation steps are

approximately as follows:

Step1: organize an experts’ group to conduct

the assessment. The experts can come from a

technical layer, a management layer of the electric

power companies, the institutes of the electric

power, or universities or government departments

related to electric power.

Step2: determine the weights associated with

the evaluation index or the total score of each index

by DELPHI method (Kayacan, Ulutas and Kaynak,

2010);

Step3: organize the experts’ visit to the thermal

power plants and their scoring through the fact-

finding inspection;

Step4: gather the score, conduct group

discussions, and finally make decisions. Usually, the

evaluation can end in one of two ways: qualified

with minor correction and remedy or unqualified

with major correction and remedy.

One detail deserves to be paid attention to here:

the conclusion is not obtained simply by the direct

addition of the scores from the experts. The real

decision making process is that the experts’ group

draws the final conclusions through discussion and

consultation. The rule of “who gets a high score,

who passes” is not necessarily clear-cut. This

process is understandable because evaluating the

security on basis of the scores only is not reasonable.

Different thermal power plants are evaluated by

different experts’ groups, and the scoring measures

of experts may be different due to their diverse

characters, moods, and knowledge background.

Therefore, electric power enterprises come to

conclusions through comprehensive group

evaluation. In this practice, historical or antecedent

cases are very valuable for the decision making

process of these experts.

Several limitations of in the evaluation process

described above can be highlighted as follows. First,

the evaluation approach presents too much

subjectivity. It generally requires high costs, a long

time, and hard labour, but lacks efficiency. In

practice, most of this kind of evaluation work is too

time consuming with respect to the quality and

reliability of conclusions drawn. A second limitation

is the lack of knowledge and information available

to support the experts’ evaluation and decision

making process while historical data and

information could be resorted to. For the past ten

years or so, thermal power enterprises have

accumulated a decent number of SATPP evaluation

reports. An evaluation report can be regarded as a

case. The cases represent the intelligence gathering

activity of experts’ and permit to trace their wisdom

and knowledge. As an important information

resource, a large amount of cases is very valuable

for reference. Unfortunately, these MSSATPP cases

are left unused and not managed, analyzed, or

utilized. During the evaluation process of new

thermal power plants, the information resource is

hard to be utilized because these evaluation reports

have not been organized and analyzed. Some of

them have not been standardized nor made

electronically available yet. More than that, due to

the lack of support of information system in which

the cases are effectively organized and analyzed and

the knowledge extracted from the case data,

Enterprise information resource and the historical

knowledge of experts cannot be communicated to

the experts’ group during the decision making

process of MSSATPP. The third limitation is the

difficulty of self evaluation and day-to-day real-time

evaluation.

Therefore, it is vital for a group of experts to have

intelligent information and knowledge support

during decision making. Following, one important

purpose of our current research is to present a more

effective case matching method different from those

commonly used in case-based reasoning for the

safety assessment issue of thermal power plants.

Another aim of our current study is to develop a

case-based intelligent enterprise information system

based on historical knowledge to assist the panel of

experts in reaching a right decision making for

MSSATPP. In the next sections, a novel case

matching method combining Delphi method and

Grey System theory is presented.

2.2 Evaluation Indexes

On the basis of actual investigations of coal-fired

thermal power enterprises, currently, the safety

evaluation of thermal power plants mainly concerns

the following six aspects, which are generally

regarded as evaluation indexes.

In CBRSYS-TPP, the cases represent actual

historical evaluation reports which have been

structured. Not only the attributes (i.e. Goal,

ResponsSys, Supervision, BasicWork, SafeEdu, and

IntergratedM) are included as evaluation indexes,

but also other important attributes, such as the

Number of Items with Deducted Marks, the Number

of Major Problems, the Assessment Result, the

Suggested Amendment Opinions, are represented. In

Fig.1, the six indexes on the left are input variables,

ICEIS 2011 - 13th International Conference on Enterprise Information Systems

and four extra attributes on the right are the output

variables. The values of input variables are acquired

by expert group scoring. Then, the similar cases

including ten rather than six attributes are able to be

acquired by case matching.

The four extra attributes on the right in Fig.1

are extremely important and valuable. The former

three items, i.e. Number of Items with Deducted

Marks (IDM), Number of Major Problems (MP),

and Assessment Result, are influential for the

decision results of the current evaluation problem.

The last one, i.e. Suggested Amendment Opinions, is

extremely helpful as reference for the expert group

to derive their suggested corrective and remedial

measures based on the specific conditions of the

thermal power plant. Accordingly, CBRSYS-TPP is

able to be used by all the expert group members to

effectively acquire their knowledge and decision

support. The entire safety evaluation procedure of

thermal power plants will be eventually completed

with the powerful aid and support of CBRSYS-TPP.

Figure 1: Evaluation indexes and four extra output

attributes in CBRSYS-TPP.

3 RESEARCH METHODOLOGY

Our research methodology is presented in three

parts. Part one proposes the retrieval method based

on grey system theory and our improvement on it

combining Delphi approach. Part two describes two

statistics for performance evaluation of our proposed

method. Part three presents our implemented

enterprise information system and data set for as

experiments.

3.1 Decision Information Acquiring

Method

In our study, we use grey system theory combining

Delphi approach to complete the acquisition of

decision information. In CBR systems, the

information acquisition is also called case matching

or case retrieval. The most famous case matching

method is the traditional CBR retrieval algorithm

which is based on Euclidean distance. Besides, other

methods such as neural networks, genetic algorithms

and fuzzy logic are also studied in previous literature

(Aamodt and Plaza, 1994; Mántaras, McSherry,

Bridge, Leake, et al, 2005; Bichindaritz and Marling,

2006).

However, there still exists a gap between the

abilities of these techniques and the real requirement

to improve their accuracy and to provide more

detailed decision information. In this article, grey

system theory and Delphi method are integrated into

case-based reasoning technology and CBR-KNN is

introduced as a novel case matching method.

Grey System Theory was first built by Ju-Long

Deng in 1982 (Deng, 1982). All systems with

incomplete information can be regarded as grey

systems (Liu and Wang, 2008). The case retrieval

algorithm for knowledge acquisition of MSSATPP

has been based on grey relationship analysis. As one

of the system analysis techniques, grey relationship

analysis is an approach for analyzing the degree of

association among different factors. Here, we

integrated it into CBR for MSSATPP and proposed

CBR-Grey. The fundamental steps using grey

relationship analysis for case retrieval in MSSATPP

are as follows (Lu, He and Du, 2008).

Step1. Determine the evaluation index system

according to the evaluation purpose, and then collect

evaluation data.

Suppose there are m data series which form the

following matrix:

()

11 1

12 m

,,,

mmn

XX X

⎛⎞

⎜⎟

∧=

⎜⎟

⎝⎠

…





where n denotes the number of evaluation indexes,

and m is the number of historical MSSATPP cases

in the case base.

Step2. Use Delphi method and obtain all weight

values of the indexes. The Delphi method is a

systematic, interactive forecasting method which

relies on a panel of experts. This technique is based

on the principle that forecasts from a structured

group of experts are more accurate than those from

unstructured groups or individuals (Harman, 1992).

Step3. Determine the reference data series. The

reference data series should be an ideal contrast

standard. They can be composed of the optimal

value or worst-case value of the indexes as well as

other reference values that are selected according to

A CASE-BASED ENTERPRISE INFORMATION SYSTEM FOR THERMAL POWER PLANTS' SAFETY

ASSESSMENT

the evaluation purpose. In our current research, the

reference data series is the target case to be solved

and the attribute values are those of the objective

case to be solved. Let X

denote the reference data

series,

(

)

000 0

(1), (2), , ( )Xxx xm= 

Step4 Normalize the data.

Step5 Compute the absolute differences

between the corresponding elements of reference

data series and comparisons from the case base,

namely |x

- x

| , i=1, 2, …, m；k=1, 2, …, n.,

where k denotes the number of attributes, and i

denotes the number of evaluation objects.

Step6 Derive the values of

0k k

min min

−

and

0k k

max max

x−

Step7 Compute the correlation coefficient. By

Formula (1), respectively compute the correlation

coefficients between each comparative series and

reference series. In Formula (1),

denotes the

resolution ratio, and its values range from zero to

one. The smaller

is, the bigger the differences

among correlation coefficients are, and the stronger

the separating capacity is. Generally, the value of

is 0.5. i denotes the case number in the case base.

()

represents the correlation between the target

case and case i in the case base for index k.

0k 0k k

0k k 0k k

min min max max

()

max max

xx x x

xx xx

−+⋅ −

(1)

Step8. Compute correlative series. Respectively

compute the average value of the correlation

coefficients between the corresponding elements of

the reference series and every evaluation object

(comparative series). This average value, named

correlation series, can reflect the correlation

relationship between the reference series and the

comparative series denoted by i. We mark it as

follows.

()

∑

(k=1, 2… n) (2)

Step9 When the indexes have different roles

and importance in comprehensive assessment, we

can compute weighted means which can be shown

as follows.

(i) w ( )

global k i

=⋅

∑

(k=1, 2 … n) (3)

where w

denotes the weight of index k.

Step10 Derive the comprehensive assessment result

on the basis of the correlation series of all the

objects of observation:

(1)

global

S , (2)

global

S , …

(m)

global

S .

In the above descriptions, the local similarity is

represented by the grey association degree of the

characteristic attributes. The global similarity is

derived by the weighted addition of all the local

similarities. For the different importance of the

evaluation indexes of thermal power plants, the

weight can be integrated into the computing process

of a comparative environment when the local

similarities are being computed. Therefore an

improved local grey association algorithm is derived

and further expressed as follows in equation (4).

min min X(i,k) max max *X(i, k)

()

( *X(i,k)) maxmax *X(i,k)

+⋅

(4)

Where

X(i,k) () ()

wxk xk=−. The local

grey similarity of the index k between the objective

case and historical evaluation case can be defined as

follows.

() 1

()

dist

−

(5)

According to the definition of the Euclidean

distance, the global similarity between two cases can

be defined as follows.

global dist 2

(())

ζζ

∑

(6)

Thereby, the global similarity of two cases can

be derived by the following formula. The case

chosen for reuse is the one maximizing the global

similarity.

global

(7)

3.2 Performance Evaluation Statistics

In this research, to evaluate the performance more

fully, two statistics are used to evaluate the

performances of different case matching methods.

One is the accuracy, the most commonly used

index for the evaluation of performance.

Another is the F

–value

. In the fields of statistics

and information retrieval, the sensitivity and

specificity are generally used for evaluating an

algorithm (Rowe and Wright, 2001). Sensitivity and

specificity are complementary of each other. The

simple improvement in sensitivity will lead to a

decreasing specificity, and vice versa. Thereby, a

good retrieval system should demonstrate both high

ICEIS 2011 - 13th International Conference on Enterprise Information Systems

sensitivity and specificity, but in reality a retrieval

system performance tends to be a tradeoff between

them avoiding too low sensitivity or specificity. The

combined effect can be evaluated by the F–value.

3.3 Data Set

The data set for our experiments are mainly

collected from a mega electric power enterprise

group, GreatT Power Generation Group of China

(GreatT). As one of the largest power generation

corporations in Asia, she owns over one hundred

power plants, most of which are coal-fired thermal

power plants. The data set are mainly the historical

security assessment data of TPP of GreatT over the

years. Most of the data are the newest assessment

reports of SATPP occurring between 2007 and 2009.

Since these TPPs vary in their degree of

informatization and electronic data were not even

available in parts of them, the task of collecting the

data was hard. The current project team collected a

total of 120 MSSATPP records, and 106 complete

and valid cases were acquired after displaying and

analyzing. Among them, the number of positive

cases is 56, and the number of negative cases is 50.

The assessment reports from the same thermal plants

but occurring in different years will be regarded as

two different records. Taking LuoHo’ Power Plant

for example, its two reports in 2008 and 2009 are

two different records of the data set. In these data,

there are at most three data records occurring for the

same thermal plant. These three data records are for

three different years.

In this research, we conducted the experiments

by 10-fold-cross-validation. The test data are

extracted randomly. For each test, 96 cases will be

used as historical data in the case base, and the

remaining 10 cases represent the testing data (five

positive cases and five negative cases respectively).

For each experiment, the tests will be repeated ten

times. Although the data set is not very large, since

there are only six attributes in the cases, according to

the usual requirement: number of attributes / number

of data should equal 1:10~1:20, it can satisfy the

experimental requirements (6 / 106 = 0.057).

4 SYSTEM IMPLEMENTATION

AND EXPERIMENTS

We implemented a prototype of CBRSYS-TPP and

used it to complete the following experiment

regarding the performance of information

acquisition. In this section, we completed two

different experiments. The first one is to test the

accuracy, sensitivity and specificity as well as

calculate the F

macro

-Value of our proposed case

matching methods which combines Delphi method

and grey system theory. And the second one is to

test several common classification methods using

the same data set. 10-fold-cross-validation tests were

conducted. The performance of the methods is

evaluated by accuracy, F

macro

-value and several

statistics. In each 10-fold-cross-validation, the data

set was divided into ten mutually exclusive subsets

with the same distribution using Matlab R2008a.

Each fold should be used only once to test the

performance of the retrieval algorithms. The most

similar cases were generated from the remaining

nine folds.

4.1 Comparison Tests with KNN

In the first experiment, tests compare different case

matching methods: the traditional case retrieval

method and our proposed approach. By the tests, the

accuracy of CBR-Grey is 94%. The average

sensitivity, average specificity, recall and F

macro

value are 96%, 92%, 92.3%, 96%, and 94.11%

respectively. Meanwhile, the traditional KNN based

on Euclidean distance algorithms is used as the

second retrieval method to acquire similar cases. In

this experiment, the value of K selected is seven.

The accuracy of CBR-KNN is 90%. The average

sensitivity, average specificity, precision, recall and

macro

-value are 91%, 90%, 91%, 91.07%, and

90.03% respectively. The results are still acceptable.

But by comparison,

CBR-Grey has significantly

higher accuracy and better comprehensive

performance.

4.2 Comparison with Other Methods

Neural networks (especially RBF Network),

decision trees and logistic regression are also

common methods for different assessment issues,

especially binary classification evaluations (Boyen

and Wehenkel, 1999; Kim and Singh, 2005; Amjady,

2003).

In the current study, comparative experiments

were conducted between CBR-Grey and the other

two methods: RBF Network and logistic regression.

The first tool for this experiment is Weka 3.6.2 in

which RBF Network is integrated. The second tool

is SPSS15 which is the platform for logistic

regression analysis. The data set used here are still

the GreatT TPP data set.

A CASE-BASED ENTERPRISE INFORMATION SYSTEM FOR THERMAL POWER PLANTS' SAFETY

ASSESSMENT

Table 1: The comparative experimental results of four different approaches (based on Great TPP dataset)

Method Accuracy Precision Recall F-value Exp. Tool

CBR-Grey 94.00% 92.30% 96.00% 94.11% CBRsys-TPP,

Matlab R2008a

Logistic

Regression

91.50% 91.07% 92.73% 91.89% SPSS15

RBF Network 84.90 80.00% 89.30% 84.39% Weka3.6.2

#: the cut value is .500;

10-fold-cross-validation tests were conducted. The

experimental results are shown in Table1. Among

them, CBR-Grey has the best accuracy (94%) and F-

value (94.11%). Logistic regression has 91.50 % of

accuracy and 91.89% of F-value. Nevertheless,

RBF

Network only

has 84.90% of accuracy and 84.39%

of F-value. Accordingly, RBF Network is not

recommended for real applications in MSSATPP.

In our proposed approach, Delphi method is

also regarded as part of the case retrieval method.

Our experimental results highlight that, as far as

practical aspects of decision support for expert panel

members are concerned, in comparison with KNN

based on Euclidean distance algorithm, the most

popular retrieval algorithm，our proposed approach

seems to present the advantage of combining the

strength of Delphi method and grey system theory to

complement the weaknesses of traditional case

matching approaches. Meanwhile, we completed the

comparative experiments among our proposed

approach and three other common methods for

binary classification evaluation issues. The

conclusion is that CBR-Grey is the best both in

accuracy and F

macro

-value. This further illustrates the

validity and high performance of CBR applied to

MSSATPP. At the methodological level, the

potential advantage of CBR-Grey is in its ability to

acquire and reuse the historical knowledge whenever

the available information is complete or incomplete.

5 CONCLUSIONS

Our proposed method integrating grey system theory

and Delphi method into CBR methodologies and

intelligent enterprise information system may

provide intelligent decision support for MSSATPP,

and the evaluation cycles of experts may be reduced

with an improved efficiency. This paper provides a

novel and effective way for the security assessment

of thermal power plants as well as a new perspective

on the use of prototypes through case aggregation

which is one of the popular trends of CBR systems

in recent years (Nilsson and Sollenborn, 2004).

From a practical perspective, this approach can not

only provide the suggested conclusion but also a

whole set of evaluation and improvement

alternatives for both expert panel members and

TPPs.

By further trials in Luodian, one of high-power

stations in China, the practical results have verified

its availability and high performance again. The

computerized system works well in providing the

knowledge and decision making support for experts

during the process of MSSATPP. According to an

anonymous survey of 32 assessment experts, 29 of

them (90.6%) replied that they were mainly satisfied

with the effects of the CBRSYS-TPP system. All the

experts expressed that they got valuable information

support during the decision making and the

conclusions are more scientific and acceptable than

those without the support of CBRSYS-TPP. This

further inflects the application values of CBR in the

safety assessment of TPPs.

ACKNOWLEDGEMENTS

This research is partially supported by the National

Natural Science Foundation of China under Grant

No. 70771037, No. 90924021 and No. 70871032,

China National “863” Program (2006AA04A126),

the MOE Project of Key Research Institute of

Humanities and Social Science in University of

Anhui Province and Science Research

&Development Foundation of Hefei University of

Technology of China (2009HGXJ0039). Specially,

we are grateful to the anonymous reviewers whose

comments have improved this paper considerably.

REFERENCES

Aamodt, Agnar and Plaza, Enric, 1994. Case-Based

Reasoning: Foundational Issues, Methodological

Variations, and System Approaches. Artificial

Intelligence Communications, IOS Press, 7(1): 39-59.

ICEIS 2011 - 13th International Conference on Enterprise Information Systems

Ahola T. and Leiviskä, K. 2005. Case-based reasoning in

web breaks sensitivity evaluation in a paper machine,

Journal of Advanced Computational Intelligence and

Intelligence Informatics 9: 555–561.

Amjady, Nima, 2003. Dynamic voltage security

assessment by a neural network based method.

Electric Power Systems Research 66(3): 215-226.

Bang, Young-hwan, Kim, Jung-gwon, Hwang, II-sun,

2008.CBR (Case-Based Reasoning) Evaluation

Modelling for Security Risk Analysis in Information

Security System, 2008 International Conference on

Security Technology 66-70.

Bichindaritz, I., Marling, C., 2006. Case-based Reasoning

in the Health Sciences: What’s Next?, Artificial

Intelligence in Medicine, Special Issue on Case-based

Reasoning in the Health Sciences, Bichindaritz, I.

(Edt.), 36(2): 127-135.

Boyen, Xavier, Wehenkel, Louis, 1999. Automatic

induction of fuzzy decision trees and its application to

power system security assessment. Fuzzy Sets and

Systems 102(1): 3-19.

Chang, Pei-Chann, Lai, Chien-Yuan, Lai, K. Robert, 2006.

A hybrid system by evolving case-based reasoning

with genetic algorithm in wholesaler's returning book

forecasting. Decision Support Systems 42(3): 1715-

1729.

Deng, J. L., 1982. Control problems of grey systems.

Systems & Control Letters 1(5): 288-294.

Erdem, Hasan Huseyin, Ali Volkan Akkaya, Burhanettin

Cetin, et al, 2009. Comparative energetic and exergetic

performance analyses for coal-fired thermal power

plants in Turkey. International Journal of Thermal

Sciences, 48(11) : 2179-2186.

Garg, R. K. Agrawal, V. P. Gupta, V. K., 2007. Coding,

evaluation and selection of thermal power plants – A

MADM approach. International Journal of Electrical

Power & Energy Systems 29(9): 657-668.

Gu D. X., Liang C. Y., Li X. G., et al, 2010. Intelligent

Technique for Knowledge Reuse of Dental Medical

Records Based on Case-Based Reasoning. Journal of

medical systems 34(2): 213 -222.

Harman. D., 1992. Evaluation issues in Information

Retrieval. Information Processing & Management

28(4): 439-528.

Han, Jung, C., I. & Shu, B., 1999. Risk Analysis for

Electronic Commerce Using Case-Based Reasoning,.

International Journal of Intelligent Systems in

Accounting, Finance &Management 8(1): 61-73.

Kayacan, Erdal, Ulutas, Baris, Kaynak, Okyay, 2010.

Grey system theory-based models in time series

prediction. Expert Systems with Applications 37(2):

1784-1789.

Kim, Hyungchul, Singh, Chana, 2005. Power system

probabilistic security assessment using Bayes

classifier. Electric Power Systems Research 74(1):

157-165.

Kumar, Rajesh, Mahur, A. K., Sengupta, D., Prasad,

Rajendra, 2005. Radon activity and exhalation rates

measurements in fly ash from a thermal power plant.

Radiation Measurements 40(2-6): 638-641.

Liu, C. H., Lin, Sue J., Lewis, Charles, 2010. Evaluation

of thermal power plant operational performance in

Taiwan by data envelopment analysis. Energy Policy

38(2):1049-1058.

Liu, P. D., Wang, T. J., 2008. Research on Risk

Evaluation in Supply Chain Based on Grey Relational

Method. Journal of computers, 3(10): 28-35.

Lora, Electo Eduardo Silva, Salomon, Karina Ribeiro,

2005. Estimate of ecological efficiency for thermal

power plants in Brazil. Energy Conversion and

Management 46(7-8): 1293-1303.

Lu, Y., He, X., Du, J. J., 2008. Malfunction case retrieval

algorithm based on Grey System Theory. Chinese

Computer Engineering of China, 34(9):28-32.

Mántaras, Ramon López de, McSherry, David, et,al, 2005.

Retrieval, reuse, revision, and retention in case-based

reasoning. Knowledge Engineering Review.

Cambridge University Press. 20(3): 215-240.

Mukhopadhyay, T., Vicinanaza, S.S, Prieutula, M.J.,

1992. Examining the feasibility of a case-based

reasoning model for software effort estimation. MIS

Quarterly 16(2):155-72.

Nilsson M. and Sollenborn, M., 2004. Advancements and

trends in medical case-based reasoning: an overview

of systems and system development. In: V. Barr and

Z. Markov, Editors, Proceedings of the 17th

international Florida artificial intelligence research

society conference—special track on case-based

reasoning, AAAI Press, Menlo Park, CA 178–183.

Oktay, Zuhal, 2009.Investigation of coal-fired power

plants in Turkey and a case study: Can plant. Applied

Thermal Engineering 29(2-3): 550-557.

Petkovšek, Samar Al Sayegh, Batič, Franc, Lasnik, Cvetka

Ribarič, 2008. Norway spruce needles as bioindicator

of air pollution in the area of influence of the Šoštanj

Thermal Power Plant, Slovenia. Environmental

Pollution 151(2): 287-291.

Rowe and Wright, 2001.Expert Opinions in Forecasting.

Role of the Delphi Technique. In: Armstrong (Ed.):

Principles of Forecasting: A Handbook of Researchers

and Practitioners, Boston: Kluwer Academic

Publishers.

Williams, Robert H., 2001. Toward zero emissions from

coal in China.Energy for Sustainable Development

5(4): 39-65.

Yang, Y P., Guo, X Y., Wang, N L., 1999. Power

generation from pulverized coal in China. Energy, In

Press, Corrected Proof, Available online.

Zhao, Y., Wang, S X, Nielsen, Chris P., Li, XH., Hao, JM,

2010. Establishment of a database of emission factors

for atmospheric pollutants from Chinese coal-fired

power plants. Atmospheric Environment 44(12): 1515-

1523.

Zhuang, Zoe Y., Churilov, Leonid, Burstein, Frada,

Sikaris, Ken, 2009. Combining data mining and case-

based reasoning for intelligent decision support for

pathology ordering by general practitioners, European

Journal of Operational Research 195(3): 662-675...

A CASE-BASED ENTERPRISE INFORMATION SYSTEM FOR THERMAL POWER PLANTS' SAFETY

ASSESSMENT