The current issue and full text archive of this journal is available at www.emeraldinsight.com/0264-4401.

htm

EC 25,1

86
Received 14 September 2006 Revised 26 September 2007 Accepted 27 September 2007

Application of geographic information system to the allocation of disaster shelters via fuzzy models
Chung-Hung Tsai
Taiwan Hospitality and Tourism College, Center of Tour Geographical Information Systems, Hualien, Taiwan

Cheng-Wu Chen
Department of Logistics Management, Shu-Te University, Yen Chau, Taiwan

Wei-Ling Chiang
Department of Civil Engineering, National Central University, Chung-li, Taiwan, and

Meng-Lung Lin
Department of Tourism, Aletheia University, Tamsui, Taiwan
Abstract
Purpose Fuzzy theory provides a rigorous, exible approach to the problem of dening and computing. Therefore, to facilitate decision making in a geographic information system (GIS), the graph layer indicator and the Takagi-Sugeno (T-S) fuzzy model must be integrated. This study aims to explain several versions of the T-S fuzzy model based on fuzzy theory and fuzzy operation. Design/methodology/approach An inference model is constructed for GIS using the T-S fuzzy model to formulate an integrated T-S decision-making (TSDMK) system. Findings The TSDMK system accommodates inexact, linguistic, vague and uncertain GIS data. The operator assigns most graph layer indicators by intuition. Practical implications Simulation results for the Hualien main station show that the proposed TSDMK system is an effective approach for GIS decision making. Originality/value This investigation assesses applications of fuzzy logic for decision making in a GIS based on TSDMK graphs focusing on model-based systems. Keywords Geographic information systems, Fuzzy control, Decision making Paper type Research paper

Engineering Computations: International Journal for Computer-Aided Engineering and Software Vol. 25 No. 1, 2008 pp. 86-100 q Emerald Group Publishing Limited 0264-4401 DOI 10.1108/02644400810841431

The authors would like to thank the National Science Council of the Republic of China, Taiwan for nancially supporting this research under Contract No. NSC 96-2628-E-366-004-MY2, NSC 96-2415-H-156-003-MY2, and NSC 95-2415-H-156-007. The authors also wish to thank Dr Pai-Hui Hsu at the National Science and Technology Center, Taiwan, for helpful discussion on the GIS and appreciate the kind assistance of Professor Roger Owen, Editor of Engineering Computations, and the constructive suggestions from anonymous reviewers for improving this paper.

1. Introduction Geographic information systems (GISs) digitally record geographical and topographical features and store this geospatial data in indexed and overlapping layers. This digital information is then converted into specic data formats and stored in computers to create a database of geospatial information. Based on various demands, the GIS system can support real-time queries and update stored information. Moreover, the functions and application elds of GIS comprise classication, calculation, measurement, overlay analysis, network analysis, digital terrain analysis, proximity analysis, buffer analysis, seek analysis, and view-shed analysis. Therefore, GIS has become one of the most useful tools for decision support. Decision support combined with GIS has become an issue important to public policy worldwide. Increasingly, governments and extensive research are concerned about the safety in disasters and saving resources. To achieve these goals, numerous successful decision-making systems in GIS have been developed (for example, see Tang et al., 2002; Kumar et al., 2005; She et al., 1999; Seker et al., 2003; Gunes and Kovel, 2000 and the references therein). However, GIS data with uncertainty have to be dealt with carefully as this data is of primary importance to reliable decision-making problems. Regan (2002) advocated that uncertainty has two broad categories: epistemic uncertainty and linguistic uncertainty. The former is uncertainty about a determinate fact, whereas and the latter is uncertainty associated with ambiguity and vagueness inherent in language. Additionally, although the spatial analytical capacity of GIS is strong, most data and knowledge concerning GIS decision making contain uncertainties that comprise both categories. How to deal with data under these situations by integrated models comprise an open eld. Hence, this study proposes the integrated Takagi-Sugeno decision-making (TSDMK) system, which combines Takagi-Sugeno (T-S) fuzzy models into GIS spatial analysis to solve uncertain problems. Uncertainty in data requires appropriate conceptual and analytical tools for successful decisions. Zadeh in 1965, in a pioneering work, simulated human thought processes and judgment in analyzing complex systems and decisions. Fuzzy logic was introduced as an extension of classical set theory. Since, Zadeh (1965) rst proposed a linguistic approach, many successful works in industrial and academic elds have been seeking (Chang and Zadeh, 1972; Zadeh, 1973; Kickert and Mamdani, 1978; Braae and Rutherford, 1979). However, all of these studies neglected adequate mathematical theory and systematic design. Takagi and Sugeno (1985) proposed a new concept, a fuzzy inference system, called the T-S fuzzy model. This fuzzy model combines the flexibility of fuzzy logic theory and rigorous mathematical analytical tools into a unied framework. Since, there have been some successful applications of the T-S fuzzy (Cococcioni et al., 2006; Hsieh et al., 2006; Zhang et al., 2006). Many applications focus on controller design in nonlinear systems (e.g. Hsiao et al., 2005; Chen, 2006; Chen et al., 2006 and the references therein). In this type fuzzy model, local dynamics in different state space regions are represented by a set of linear sub-models. The overall model of the system is then a fuzzy blending of these linear sub-models. Therefore, an inference model using T-S fuzzy scheme is constructed for GIS to formulate an integrated TSDMK system.

Allocation of disaster shelters 87

EC 25,1

88

Previous studies of disaster management strictly focused their efforts on disaster relief and prevention, but neglected to study emergency contingency measures for rapid evacuation of the general population to disaster shelters. However, after the Chi-chi Earthquake in 1999, related studies have attracted considerable attention. Once disaster strikes, the planning of emergency measures can be classied as evacuation and division of work. First, the proper evacuation site must be selected. Second, workers must be allocated to run rescue/relief missions. Due to the enormity and complexity of components involved in planning of emergency measures, this study focuses solely on the analytical solution for the rst stage of emergency response, evacuation. The remainder of this paper is organized as follows. First, the planning for disaster shelters and the theoretical basis of disaster prevention strategies are reviewed. Fuzzy set theory and the T-S type fuzzy model for dealing with GIS data are briey presented. Consequently, the formulation of TSDMK system is presented and why a graphic layer showing GIS needs the decision-making TSDMK system is discussed. Finally, an example of multi-objective decision making for tourism disaster prevention in Hualien is utilized to demonstrate the proposed approach, and conclusions are drawn. 2. The planning of disaster shelter Urban disaster shelters are the only places where the citizens can get reprieve once disaster strikes. However, in Taiwan, where the frequency of natural hazards is relatively high, urban shelters have recently gained public attention. In old urban areas and those with concentrated populations, a lack of disaster shelters can increase the difculty in applying disaster emergency measures. This is particularly obvious in areas around the train station (city centre) where losses of life and property can be considerable when earthquakes, oods or res hit the city. Other problems include the fact that natural disasters cause mass panic and the most citizens will not be able to reach disasters shelter in time to avoid the forces of nature. Particular attention needs to be paid to planning disaster shelters that ensure that people can quickly reach shelters. Current earthquake disaster shelters can be divided into three categories: (1) Emergency disaster shelters. When an earthquake strikes, the public is supposed to leave their homes and congregate in the nearest open areas within half-a-day. In immediately reacting to an earthquake, the public should follow emergency procedures and head towards safe areas, which are designated as emergency disaster shelters. These areas include roads, open spaces such as parks or schools, which often have basic features such as trees that prevent dispersion of res after an earthquake event. At the same time, the public will determine whether their homes are damaged and whether they should continue to stay at a shelter, return home or move to another temporary shelter. (2) Neighborhood disaster shelters. Neighborhood disaster shelters are temporary housing for an indeterminate length of time from half-a-day to two weeks. These shelters are used when homes are destroyed or damaged in the rst or second shockwave after an earthquake. Neighborhood disaster shelters include public schools, assembly halls, community activity centers, and temples.

These shelters are characterized by the presence of emergency facilities such as tents, which are provided by the government or private donors. (3) Urban shelters. These long-term shelters are primarily used for a periods longer than two weeks. When earthquake survivors have no home to return to, the government builds temporary shelters for those in needs. These shelters would be located in city parks, large public schools, sport facilities (baseball elds) or army camps. Water and electricity will be provided for everyday use, and all means necessary shall be undertaken to guarantee that the survivors can lead a normal life until they can locate permanent housing. According to red department statistics, Hualiens 13 townships have 179 planned shelters (Table I). Most of the shelters are in activity centers. However, due to the lack of space, shelter capacity is limited; therefore, a pressing need exists to restructure these shelters to satisfy the needs of citizens. This case study investigates how these shelters should be distributed to achieve maximum effect and fairness under limited resources. 3. Fuzzy set theory The traditional characteristic mapping (henceforth called crisp set) takes only two values: 1 if an element belongs to the set; and 0 if it does not. A fuzzy set uses an idea of a membership, rather than being crisp, dened by grade or degree. The normalized fuzzy set A is a set having a membership function which assumes a value between [0, 1]. A is usually identied with these membership functions. If x is a possible value of X, then A(x) means its membership degree in the fuzzy set A. Therefore, the value 1 indicates complete membership, while the value 0 indicates complete lack of membership. Intervening values implies intervening grades of membership. That is to say, a crisp set would have membership values limited strictly to the set {0, 1}, with no intervening values. A denition of fuzzy set is described below: ~ on R is said to be a fuzzy Denition 1. Let R be a real number set. A fuzzy set A number if the following conditions are satised:
Township Shioulin Shincheng Hualien Jian Shoufeng Fuli Fuli Juoshi Yuli Fengbin Rueisuei Guangfu Wanrung Planned disaster shelters 11 14 10 21 16 13 14 10 21 8 13 14 14

Allocation of disaster shelters 89

Table I. Currently planned disaster shelters in Hualien county

EC 25,1

x0 [ R, such that mA ~ x0 1, and membership function mA ~ x is piecewise continuous; and  ;a [ 0; 1; Aa ; {xmA ~ x $ a; x [ R} is a convex set on R,

90

~ and Aa is a crisp set. The convex set means that where x0 is the mean value of A ;x [ x1 ; x2 ; f x $ min f x1 ; f x2 : Fuzzy set theory is involved with a basic concept in mathematics. It also uses various mathematical disciplines, such as logic, algebra topology, data analysis, etc. and has the application on approximate reasoning, control paradigms, and modeling elds, etc. Similarly, relationships among fuzzy sets are dened by logical operators, much in the same way of relationships among crisp sets in classic set theory. As with traditional crisp sets, fuzzy sets obey several important laws dened below: A < A A; A < B B < A; A < B < C A < B < C ; A > A A idempotency A>BB>A commutativity associativity

A > B > C A > B > C

A < B > C A < B > A < C ; distributivity

A > B < C A > B < A > C

where < and > denote union and intersection, respectively. The union of two sets is dened to include all the members of each set and its membership function is dened as:

mA < B mA ^ mB:
Here, the symbol ^ indicates maximization. By contrast, the intersection operator is dened to contains only those members which belong to both sets and its membership function is described as:

mA > B mA _ mB
where the symbol _ denotes minimization. Fuzzy set theory provides a complete, pseudo-Boolean logic that can be used to create a mathematical framework for decision making in GIS with poorly dened information. However, fuzziness is typically mistaken for probability. Therefore, distinguish between these concepts is necessary.

4. Probabilistic and fuzzy uncertainty Probability theory is based on classical set theory and two-valued logic, such as yes-or-no or happen-or-not-happen statements. Probability theory is concerned with the development of a mathematical framework that treats random phenomena. These phenomena occur in many investigations in which repeated experiments are conducted under the same conditions yield different outcomes (events) that are not deterministically predictable; the information required for a precise prediction is lacking either due to incompleteness or uncertainty. Rather, the outcomes exhibit certain statistical regularity that, while individual outcomes are not predictable, the relative

frequency of events occurring can be approximately predicted. Therefore, probabilistic uncertainty pertains to events that have a well-dened, unambiguous meaning. Because decision-making statements in GIS cannot be well dened, the GIS will yield an unsatisfactory evaluation for the indicator of graph layer showing. Fuzzy uncertainty, conversely, pertains to events that have ill-dened, vague meanings. Fuzzy set theory is based on multiple-valued logic that can evaluate intermediate values between happening and not-happening. Therefore, we propose the following framework for rule-based fuzzy models for decision making in GIS. 5. Rule-based fuzzy model In rule-based fuzzy models, knowledge is represented by if-then rules. These rules have two parts. One is an antecedent part stating conditions on the input variables and the other is a consequent part that formulates the corresponding values of output variables. The case of a single output variable is commonly considered in real applications. The rule-based fuzzy models most frequently used include the Mamdani, relational, and T-S types. In T-S type models, the consequent part expresses a (non-)linear relationship between input and output variables, whereas in Mamdani (1977) type models, both antecedent and consequent parts consist of fuzzy statements regarding the variables involved. Figure 1 shows the structure of a typical fuzzy model. The fuzzication step calculates their membership grades in particular fuzzy sets C L and C H of the inputs and the dened membership function miC (Figure 2). The inference step determines, from input membership grades mC , the resulting membership function mres y of the model output by i i 1; 2; . . . ; m rules bases. A rule base is described as follows (Hsieh et al., 2006): R1 : IF x1 is C L AND x2 is C H THEN y f 1 x:
Input xi

Allocation of disaster shelters 91

Fuzzification Elements: membership functions of the inputs xi

Inference Elements: rule base, inference mechanism and membership functions of the output y

Defuzzification Elements: defuzzification mechanism

Figure 1. Structure of a fuzzy model

EC 25,1

92

From the resulting membership function mres y, a crisp, numerical output value y * will be computed in the defuzzication step. The value y * is the effect of the numeric input values xi. In this study, only the T-S-type models are discussed, because they show effectiveness in decision making of construction management and the T-S-type model is believed to be of utility in GIS for the indicator of graph layer showing. The fuzzy statements used have a linguistic nature, for example, expressing that a certain variable (also called linguistic variables) takes a high or low value. First of all, the T-S fuzzy model can have the form of a set of rules as well as membership functions represented below: R1 : IF z is F 1 THEN r f 1 z; R2 : IF z is F 2 THEN r f 2 z; . . . Rm : IF z is F m THEN r f m z: Then the output can be obtained on the basis of the grade of activation of the particular conclusions f i ; i 1; 2; . . .m, which is determined by following formula:
m X m F i z f i z

i1 m X m F i z i1

where Fi is the fuzzy set; the function fi(z) can be nonlinear. In the subsequent paragraphs, the mathematical verication and application of GIS to the allocation of disaster shelters is proposed and this study deals with this GIS application using fuzzy rule-based modeling. 6. Fuzzy models to assess decision making in GIS Although the spatial analytical capacity of GIS is strong, it still is insufcient for uncertain and deterministic materials when dealing with a massive decision problem.

mC 1

Figure 2. Fuzzy sets to represent low and high variables

0 0

CL

CH

Hence, this study combined GIS with T-S fuzzy models to solve an uncertain problem for effective decision making. Unlike traditional modeling, fuzzy rule-based modeling is essentially a multimodel approach in which individual rules are combined to describe the global behavior of the system (Chiang et al., 2004). Therefore, fuzzy modeling is employed to simplify the GIS decision-making problem. The GIS knowledge database of the TSDMK system is constructed by several sets of if-then (also called antecedent-consequent) rules in T-S fuzzy models. Fuzzy knowledge comprises the database for searching the graphic layer of the GIS. The variables in fuzzy rules are formulated in linguistic variables. By a fuzzy inference machine, the weights are determined to show a suitable graph in the GIS. Furthermore, the information format in the GIS is recorded to represent a pixel, and the description of a pixel results in a fuzzy rule. If there are m*n pixels in a spatial region, there are m*n fuzzy rules before removing repeats or resemblance sets of rules. Each rule contains one linguistic variable of the learning resource and learning goal. These data of learning sources and learning goals represent the same location but have a different graphic layer pixel. The combination of learning sources and learning goals becomes the fuzzy knowledge map for searching a solution of GIS graphic layers. The weighted value can be determined by the knowledge-based operation in T-S fuzzy models. A decision-maker can analyze the suitability for each space and any fuzzy degree. The function of computed weighting in T-S fuzzy models can be coded to be part of a system model. The combination of the T-S fuzzy model and GIS would be effective for a GIS database and ability for special decision making. 7. Mathematical verication and application to the allocation of disaster shelters This section gives the geography of the Hualien train station area and presents a case study the allocation of disaster shelters in the surrounding area. Based on statistical interpretations and the combined T-S fuzzy and GIS analysis, both the optimal serviced shelters and the minimum distance from the station can be determined. 7.1 Study scope and analysis of the locale The Hualien train station is located in the city center in the Guo-lien neighborhood (East Longitude 121859 0 23 00 North Latitude 23899 0 20 00 ). The two-oor station building is built of reinforced concrete. The station front has an area of 7,380 ft 2, whereas the rear area is 4,860 ft 2. The Bei-huei rail track and Hua-dong rail track intersect at this station. According to the Ministry of Transportation in 2005, there are on average 6,000 people who board trains daily at the Hualien train station. The Pacic Ocean is to the east of the station and the Shio-lin township is located to the west. Shioulin and Jian townships are located to the north and south, respectively. The station is situated close to the Mei-Luen Fault. According to the disaster potential analysis by Hualiens disaster management agency, the station is a high-risk area. Figure 3 shows the general layout of the area. Figure 4 shows the triangular fuzzy set that represents the fuzzy number with three 2 crisp parameters. Hence, the expression Ai ai ; c i ; ci stands for a triangular fuzzy number hereafter. Table II shows the distance between the Hualien train station and every disaster shelter and the data are model values, implying that shelter numbers are

Allocation of disaster shelters 93

EC 25,1
Siou-lin Township

Sin-cheng Township

Meilun Fault

94
Hua-lien City Hualien train station

Pacific Ocean Ji-an Township

Figure 3. The location of Hualien train station

mA 1

Ai

Figure 4. Triangular fuzzy set

ci 0 ai

c+ i x

2 all triangular fuzzy numbers and c i 10 percent and ci 10 percent are the left and right spreads, respectively. This case study analysis the area surrounding the Hualien train-station using the TSDMK system with the GIS, and performs a network analysis of evacuation routes using speeds of 40 and 60 km/h. Table III and Figures 5 and 6 show the optimal disaster shelters and the driving time between shelters and Hualien train station. Based on these numerical results, this study readily obtains the shortest distance to each disaster shelter from Hualien train station and the best disaster shelter under uncertain factors.

8. Results, discussion and conclusion Hualien has always been a tourist hotspot. In recent years, increased natural disasters have severely affected tourism and economic growth; therefore, nding a balance

Serial numbers Planned disaster shelters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 De Xing Sports Stadium Hualien City Government Assembly Hall National Hualien Vocational High School of Agriculture Mei Lun Junior High School Chung Zhen Primary School Chung Hua Primary School Si Wei Senior High School Zhu Quan Community Centre Guo Fu Community Centre Zhu Nong Community Centre Min Xiao Community Centre Shin Yi Primary School Ming Li Primary School Ming Yi Primary School Ming Lian Primary School Ming Chi Primary School Chung Xiao Primary School Experimental Primary School of National Hualien University of Education Tzu Chi Primary School National Police Agency Fourth Police Section Railway Police Ofces Basement Ministry of Transportation and Communications Taiwan railway Administration Maintenance department Parts Hualien Factorys basement Zi Qiang Junior High School Gang Tian Temple Zhu Qiang Primary School Hua Gang Junior High School Guo Feng Junior High School

Distance between shelter and Hualien train station (km) 1.5 1.8 1.6 3.9 2.5 1.2 1.3 2.1 3.6 2.8 2.6 2.3 1.5 2.2 1.5 2.5 2.6 4.2 1.4 0.3 0.5 3.7 1.7 3.5 1.6 1.8

Allocation of disaster shelters 95

Table II. The distance between Hualien train station and every disaster shelter

between disaster mitigation and tourism development has become a key question for the county government. The Hualien train station is a location where tourists congregate. This study attempts to resolve the issue of crowd dispersal and identify a rapid course to shelters once disaster strikes. The TSDMK system effectively provided a spatial analysis of possible shelter targets and pinpointed the three most suitable shelters Si Wei Senior High School, Tzu Chi Primary School, and Chung Hua Primary School. Moreover, the variables used in determining the most suitable shelter locations include distance from shelters and local population. However, these variables are conditioned by local topography and affect the time required for disaster relief. Additionally, the number of tourists is not taken into account; therefore, the management of disaster relief requires additional manpower and resources to build an optimum-sightseeing environment in Taiwan. Moreover, this work conrms that the proposed methodology was adequate for integrating the GIS database, graph layer indicator and T-S fuzzy model.

EC 25,1

96

Serial numbers 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

Planned disaster shelters De Xing Sports Stadium Hualien City Government Assembly Hall National Hualien Vocational High School of Agriculture Mei Lun Junior High School Chung Zhen Primary School Chung Hua Primary School Si Wei Senior High School Zhu Quan Community Centre Guo Fu Community Centre Zhu Nong Community Centre Min Xiao Community Centre Shin Yi Primary School Ming Li Primary School Ming Yi Primary School Ming Lian Primary School Ming Chi Primary School Chung Xiao Primary School Experimental Primary School of National Hualien University of Education Tzu Chi Primary School National Police Agency Fourth Police Section Railway Police Ofces Basement Ministry of Transportation and Communications Taiwan railway Administration Maintenance department Parts Hualien Factorys basement Zi Qiang Junior High School Gang Tian Temple Zhu Qiang Primary School Hua Gang Junior High School Guo Feng Junior High School

Township time required to arrival at shelter from Hualien train station in the different driving speed 40 km 60 km 4 3 3 8 6 3 4 4 8 6 6 5 4 5 3 5 5 9 1 1 1 8 4 8 3 4 2 2 1 6 4 1 2 3 5 3 3 2 2 3 2 3 3 5 1 1 1 6 2 6 2 2

Table III. Time required to arrival at shelter from Hualien train station

Unlike traditional modeling, fuzzy rule-based modeling is a multimodel approach that combines individual rules to describe the global behavior of a system. Therefore, fuzzy modeling was employed to simplify the decision-making problem in the GIS. This investigation is an indicator evaluation that provides information in a graph layer that can be employed by GIS experts to make strategic decisions. Additionally, the TSDMK system may prove helpful for future scenarios, addressing changes in linguistic variables caused by operators. However, the TSDMK system is still in exploitation. Hybrid techniques will be adopted in future to alleviate problems such as input variable selection, and optimization of decision selection and membership functions within fuzzy rule-based models.

Network Analysis of Hua-lien Railstation Disaster Mitigation Program (40Km)

309000 310000 311000 312000 313000 314000

Allocation of disaster shelters

E
2658000

N
2658000

W S

97

2657000

2656000

2654000

2655000

2653000

2652000

Legend
2651000

2650000

2650000

Driving time (Minute) 1 2 Hua-lien 3 train station 4 Disaster shelters 5 for evacuation 6 Fire staion division 7 8 Police station 9 Hospital 10 314000

2651000

2652000 309000 310000 311000 312000 313000

2653000

2654000

2655000

2656000

2657000

Note: Reproduced from the only available original

Figure 5. Network analysis diagram for Hualien train stations disaster shelters by speeds of 40

EC 25,1
Network Analysis of Hua-lien Railstation Disaster Mitigation Program (60Km)
309000 310000 311000 312000 313000 314000

98
2658000

N W S
2657000 2657000

2656000

2654000

2655000

2653000

2652000

Legend
2651000

2650000

Figure 6. Network analysis diagram for Hualien train stations disaster shelters by speeds of 60

309000

310000

311000

312000

313000

314000

2650000

Driving time (Minute) 1 2 Hua-lien 3 train station 4 Disaster shelters 5 for evacuation 6 Fire staion division 7 8 Police station 9 Hospital 10

2651000

2652000

2653000

2654000

2655000

2656000

2658000

References Braae, M. and Rutherford, D.A. (1979), Theoretical and linguistic aspects of the fuzzy logic controller, Automatic, Vol. 15, pp. 553-77. Chang, S.S.L. and Zadeh, L.A. (1972), On fuzzy mapping and control, IEEE Transactions on Systems, Man, and Cybernetics, Vol. 2, pp. 30-4. Chen, C.W. (2006), Stability conditions of fuzzy systems and its application to structural and mechanical systems, Advances in Engineering Software, Vol. 37 No. 9, pp. 624-9. Chen, C.W., Chiang, W.L., Tsai, C.H., Chen, C.Y. and Wang, M.H.L. (2006), Fuzzy Lyapunov method for stability conditions of nonlinear systems, International Journal on Articial Intelligence Tools, Vol. 15 No. 2, pp. 163-71. Chiang, W.L., Chen, C.W. and Hsiao, F.H. (2004), Stability analysis of nonlinear interconnected systems via T-S fuzzy models, International Journal of Computational Intelligence and Applications, Vol. 4 No. 1, pp. 41-55. Cococcioni, M., Guasqui, P., Lazzerini, B. and Marcelloni, F. (2006), Identication of Takagi-Sugeno fuzzy systems based on multi-objective genetic algorithms, Lecture Note in Articial Intelligence, Vol. 3849, pp. 172-7. Gunes, A.E. and Kovel, J.P. (2000), Using GIS in emergency management operations, Journal of Urban Planning and Development, Vol. 126 No. 3, pp. 136-49. Hsiao, F.H., Chen, C.W. and Liang, Y.W. et al., (2005), T-S fuzzy controllers for nonlinear interconnected systems with multiple time delays, IEEE Transactions on Circuits & Systems-I, Vol. 52 No. 9, pp. 1883-93. Hsieh, T.Y., Wang, M.H.L. and Chen, C.W. et al., (2006), A new viewpoint of s-curve regression model and its application to construction management, International Journal on Articial Intelligence Tools, Vol. 15 No. 2, pp. 131-42. Kickert, W.J.M. and Mamdani, E.H. (1978), Analysis of a fuzzy logic controller, Fuzzy Sets and Systems, Vol. 1, pp. 29-44. Kumar, P., Singh, V. and Reddy, D. (2005), Advanced traveler information system for Hyderabad city, IEEE Transactions on Intelligent Transportation Systems, Vol. 6 No. 1, pp. 26-37. Mamdani, E.H. (1977), Application of fuzzy logic to approximate reasoning using linguistic systems, Fuzzy Sets and Systems, Vol. 26 No. 1, pp. 1182-91. Regan, H.M. (2002), A taxonomy and treatment of uncertainty for ecology and conservation biology, Ecological Applications, Vol. 12 No. 2, pp. 618-28. Seker, D.Z., Kabdasli, S. and Rudvan, B. (2003), Risk assessment of a dam-break using GIS technology, Water Science and Technology, Vol. 48 No. 10, pp. 89-95. She, T.H., Aouad, G. and Sarshar, M. (1999), Geographic information system (GIS)-based bridge management system, Computer-aided Civil and Infrastructure Engineering, Vol. 14 No. 6, pp. 417-27. Takagi, T. and Sugeno, M. (1985), Fuzzy identication of systems and its applications to modeling and control, IEEE Transactions on Systems, Man, and Cybernetics, Vol. 15, pp. 116-32. Tang, A.P., Xie, L.L. and Tao, X.X. (2002), Application of GIS to build earthquake emergency response system for urban area, Journal of Harbin Institute of Technology, Vol. 9 No. 1, pp. 38-42. Zadeh, L.A. (1965), Fuzzy sets, Information and Control, Vol. 8, pp. 338-53.

Allocation of disaster shelters 99

EC 25,1

Zadeh, L. (1973), Outline of a new approach to the analysis of complex systems and decision processes, IEEE Transactions on Systems, Man, and Cybernetics, Vol. 3, pp. 28-44. Zhang, Z.Y., Zhou, H.L., Liu, S.D. and Harrington, P.B. (2006), An application of Takagi-Sugeno fuzzy system to the classication of cancer patients based on elemental contents in serum samples, Chemometrics and Intelligent Laboratory Systems, Vol. 82, pp. 294-9. Corresponding author Cheng-Wu Chen can be contacted at: cwchen@mail.stu.edu.tw

100

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com Or visit our web site for further details: www.emeraldinsight.com/reprints