Вы находитесь на странице: 1из 5

World Congress on Software Engineering

The Application of Fuzzy Comprehensive Evaluation to Quantify Design Space

Yuping-She
Electronic and information engineering department
Putian university
Putian, China
Sheyp2004@126.com

Abstract—During the conceptual phase of a building design, partially knowledge about, and sometimes not aware of, the
major design space decisions that have the greatest influence design-relevant construction input [2]. In addition, the
on buildability of a project are taken. Quality function decision-making process at early design phases tends to be ill
deployment (QFD) is an integrated decision-making structured and occurs in an unsystematic way. Quality
methodology that can assure and improve the alignment of Function Deployment (QFD) [3] is an integrated decision-
elements of design and construction processes with the making methodology that can assure and improve the
requirements customers. Taking the existence of uncertainty in alignment of elements of design and construction process
the early design into account, we present a fuzzy with the overall requirements of a construction project. QFD
comprehensive evaluation-based approach to quantify design
has the potential to aid the development of a structured and
space, and using degree of similarity (or measure of similarity)
systematic method to support the process of buildable design
and principle of choosing nearest (or principle of selecting
near) to choose the best design from the alternatives at the decision making with suitable adoption and extension to
early phase of design. facilitate its implementation. [10] presented an approach for
integrating the fuzzy set theory to House of Quality (HOQ).
Keywords-Design Space; Quantify Design Space; Quality [11] introduced the Fuzzy Quality Function Deployment
function deployment (QFD); fuzzy comprehensive evaluation; theory and the step-by-step hierarchy structure in aircraft top
degree of similarity; principle of choosing nearest decision design. Both of them make some improvement
basing on QFD.
I. INTRODUCTION Uncertainty always exists at the early phase of design,
which includes not only conceptual knowledge, quantified
Design-construction integration during the early phase of diagram, but also the client requirements and changing
a project provides the potentials for designers to give their experience etc. This paper presents an approach based on
clients better value-for-money designs. Previous research fuzzy comprehensive evaluation to quantified design space,
efforts into buildability have documented concepts; deve- which is different from QFD. Firstly, choose a appropriate
loped principles and ways to improve buildability; identified fuzzy comprehensive evaluation model; and then use the
barriers; quantified costs/benefits; and provided project level fuzzy theory to quantified the design space; thirdly, choose a
models, approaches and guide to implementation. All of best design from alternatives using degree of similarity and
these efforts focused on management systems and principle of choosing nearest. The rest of paper organized as
organizational commitment to the buildability concept and follows. Section 2 reviews the basic concept and main
proposed to use them to overcome the technical and application of fuzzy comprehensive evaluation. Section 3
contractual barriers that limit the integration of design and presents that how to apply three-stage fuzzy comprehensive
construction. evaluation for quantifying design space. Section 4 presents a
When there is no direct organizational and managerial case study and section 5 concludes.
support for buildability implementation, the integration of
design and construction heavily depends on designers’ prior II. THE BASIC CONCEPT AND MAIN APPLICATION OF
construction experience [1]. However, the designers often FUZZY COMPREHENSIVE EVALUATION
only partially knowledge about, and sometimes not aware of, People often need to judge on something influenced by
the design-relevant construction input [2]. In addition, the several uncertain factors in actual application. Such as, to
decision-making process at early design phases tends to be ill evaluate the design quality of a project, including surface
structured and occurs in an unsystematic way. Quality structure, the cost and rationality; to evaluate the quality of
Function Deployment (QFD)[3] is an integrated decision- teaching, for example, judging the quality of the students’
making methodology that can assure and improve the work, classroom disciplinary and grade etc. In order to judge
alignment of elements of design and construction When those problems in reason, we usually adopt the
there is no direct organizational and managerial support for comprehensive evaluation. In reality, the judged subjects
buildability implementation, the integration of design and often control by several uncertain factors, among which
construction heavily depends on designers’ prior construc- fuzziness is the most important. In this way, it seems quite
tion experience [1]. However, the designers often only

978-0-7695-3570-8/09 $25.00 © 2009 IEEE 499


DOI 10.1109/WCSE.2009.48

Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2011 at 05:50:38 UTC from IEEE Xplore. Restrictions apply.
natural to combine the fuzzy technique with classical theory factors, that is, it is hard to faithfully reflect the position of
of comprehensive evaluation. every factor in the whole. In this case, first-stage fuzzy
Fuzzy comprehensive evaluation couldn’t denote by comprehensive evaluation is far from enough, we always
some simple value as the classical comprehensive need compounding of multi-first stage, which called multi-
evaluation, and then use the operation of weighting and stage fuzzy comprehensive evaluation. Here we discuss the
summing to gain a score, and complete with ordering by main model of the second-stage fuzzy comprehensive
choosing the best. Fuzzy comprehensive evaluation firstly evaluation adopted by this paper:
must set up a factors set F = { f1 , ..., fn } and a comment First-stage fuzzy comprehensive evaluation:
set C = {c ,..., cn } , and then use expert evaluation methods or Bi = Ai * Ri ( i = 1, ..., n ) ;
1 Second-stage fuzzy comprehensive evaluation:
others to build evaluation matrix R = ( rij ) n×m ∈( F × C ) and make ~
R = ( B , ... B n ) ;
T
1
comprehensive evalua-tion by appropriate blurring operator.
Assume that we are given n stones, c1 , ..., cn , whose
A. basic factors of fuzzy comprehensive evaluation [3] [4] ~
[5] weights a1 , a 2 , ..., an respectively, which serve as A , and then
The points to be discussed in fuzzy comprehensive ~ ~ ~ ~

evaluation can be brought under three main factors: according to , we get B as the result of the whole
B = A* R
evaluation.
a) a factors set F = { f , ..., fn } , refers to the set of the
1 In this model, we also need to introduce two concepts:
factors influencing the evaluated subject; degree of similarity and principle of choosing nearest.
b) a comment set C = {c1 ,..., cn} , refers to the set of Degree of similarity is a measurement of the closeness
1

between two F sets.


comments; Definition 1.[3][4][5] Degree of similarity.
c) an evaluation of single factor, viz. To obtain the Assume that A, B, C ∈ f (U ) , if and only if the
( r , r , ..., r )
mapping N :(U ) × (U ) → [0,1] meets the following conditions:
fuzzy set i1 i2 im on F by judging on the single
a) N ( A, B ) = N ( B , A);
f ( i = 1 , ..., n )
factor i , so it is a fuzzy mapping from to . F C b) N ( A, A) = 1, N (U , ∅ ) = 0;

f : F →(C ), fi 6 (r , r ,..., rim ) And the fuzzy mapping f can c) If A ⊆ B ⊆ C, so N ( A, C ) ≤ N ( A, B ) ∧ N ( B , C ) , and


i1 i2
determine a fuzzy relation R ∈ , called evaluation matrix N ( A, B )
f n× m hence, is the degree of similarity between F sets A
⎛r r12 ... r1m ⎞ and B , N is the measure function of similarity.
⎜ 11 ⎟
⎜r
⎜ 21 r22 ... r2 m ⎟⎟
, which consists of the fuzzy set judging The above definition is only of principle. In reality, we
R = ⎜
⎜ ... ... ... ... ⎟
⎟ may choose the specific one, here we choose Hamming
⎜ ⎟

⎝rn1
rn 2 ... r ⎠⎟
nm
degree of similarity.
from all the single factors. Definition 2.[3][4][5] Hamming degree of similarity.
On account of inequality of factors, we must weight all 1 n
If U = {u1 , u 2 , ..., u n }, N ( A, B ) = 1 − ∑ A( u i ) − B ( ui ) and here
the factors and express the weighing of all factors by the set n i =1

A = ( a1 , a 2 , ..., an ) in F , which combines with evaluation


is Hamming degree of similarity.
N ( A, B )

matrix R to produce comprehensive evaluation of all factors. The principle of this model is as follows:
So we can obtain fuzzy comprehensive evaluation model I: Consult with the experienced expert to give a group of
weights, which is called the alternatives of the weights, and
A * R = B = ( b , b , ..., bn ) ,
1 2 then choose the perfect one from the alternatives in term of
since, principle of choosing nearest.
A = ( a1 , a2 , ..., a n ) , ∑ a = 1, a ≥ 0 The course of evaluation is as follows:
n
i i
i =1
Firstly, we assume that J = { A , ..., A s } is the fuzzy sets
R = ( rij ) n×m , ri j ∈ [0, 1]
1
(the alternatives) on U , use a fuzzy mapping fitting reality to
n
b j = ∑ ai rij , j = 1, ..., n . work out their outputs respectively, for example, we choose
i =1
Bi = Ai D R , i = 1, ..., s ,
Note that b j is function of r1 j , r2 j , ..., rn j , namely
If there exists i , which make N ( Bi , B ) = max[ N ( B j , B )]
come
evaluation function, and here A is a normalization fuzzy set. 1≤ j ≤ s
In this model (fuzzy comprehensive evaluation model I), into existence, we consider that Ai is the best weights in J .
every factor plays a part in the evaluation result.
(In this case, we need a given evaluation result B .) If the
B. Main multi-stage fuzzy comprehensive evaluation factors in this level are also composed of sub-factors, we can
model divide them into several sub-factors respectively, and then
Many judged subjects often are influenced by many carry out the further stage evaluation accordingly, for
uncertain factors, it is difficult to determine the weights of all

500

Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2011 at 05:50:38 UTC from IEEE Xplore. Restrictions apply.
example, third-stage fuzzy comprehensive evaluation or .High . Interaction manager with
multi-stage fuzzy comprehensive evaluation. extensible data types
.Medium . Extensible interaction
III. APPLICATION OF MULTI-STAGE FUZZY COMPRE- manager
.Low Abstract device variability(F22)
HENSIVE EVALUATION TO QUANTIFY DESIGN SPACE
User interface adaptability across .Ideal device
The uncertainties exist in early phase of design, such as, devices(F13)
the participants of building designs often have conflicting .None .Parameterized device
perception design problems and their corresponding
.Local behavior changes . Device with variable
solutions. Further, the design-relevant information during the operations
early design process is not always clearly or fully stated. In .Global behavior changes .Ad-hoc device
this case, we must consider several alternatives, which are Notation for user interface
.Application semantics changes
used to present different solutions to the same problem [12]. definition(F231)
Depending on other system details, one solution may be Computer system organization(F14) . Implicit in shared user
better than others. Using alternatives allows the user to interface code
specify the design space of their application. This design .Uniprocessing .Implicit in application code
space encapsulates the different design decisions for each .Multiprocessing .External declarative notation
.Distributed processing .External procedural notation
application task that must be evaluated to identify the best Basic interface class(F15) .Internal declarative notation
design. In this paper, we will introduce a fuzzy .Menu selection .Internal procedural notation
comprehensive evaluation to quantify design space and .Form filling Basis of communication(F24)
choose the best one when it comes up several result .Command language .Events
alternatives. .Natural language .Pure state
.Direct manipulation .State with hints
A. Identify three main factors of design space for user- Application portability across user .State plus events
interface architectures in evaluation interface styles(F16)
.High Control thread mechanism(F25)
The design space is too large to discuss in this paper. So, .Medium .None
the design space reported here, together with latter .Low .Standard processes
application on it, describes architectural alternatives for user .Lightweight processes
interface software. The factors influencing the design space .Non-preemptive processes
for user-interface architectures mainly come down to .Event handlers
structural dimensions and functional dimensions [6][7][8][9], .Interrupt service routines
both structural dimensions and functional dimensions are B. Apply the third-stage fuzzy comprehensive elevation to
composed of several sub-factors respectively, and the quantify design space
respective sub-factors can also be composed of several
factors respectively (Tab.Ⅰ). In order to implement quantify, we use third-stage fuzzy
We can present this relationship by F={F1, F2}, F1={F11, comprehensive elevation to achieve it. In the following part,
F12, F13, F14, F15, F16}, F2={F21, F22, F23, F24, F25}, and F11, we will illustrate the whole process of first-stage fuzzy
F12, F13, F14, F15, F16, F21, F22, F23, F24, F25 can also be comprehensive elevation in detail for user interface
architecture. We use the user interface design space
composed of several factors as shown in Tab.Ⅰ.
mentioned in [9] to illustrate the application of the third-
In order to illustrate easily in this paper, the comment
sets in every evaluation stage can all be determine as four stage fuzzy comprehensive evaluation for it. Taking F11 ={no
grades, which are extreme need, need, intermediate between external events, process events while waiting for input,
need and need and no need. In this paper, we adopt the external events preempt users commands} as example in
model as follows: detail, we weight these three factors as A11 , and
B = A * R = ( b , b , ..., bn ) , b = ∑ a r , j = 1, ..., n , the advantage of A11 =(0.1 0.2 0.7) , here, we still need to consult some experts to
n
1 2 j i ij
i =1
taking it is that every factor plays a part in the evaluation get the evaluation matrix R11 accordingly. Provided that
result. ⎛ 0.1 0.2 0.3 0.4 ⎞ . (In this paper, we don’t want to discuss
R = ⎜ 0.0 0.1 0.2 0.7 ⎟
11 ⎜ 0.8 0.1 0.1 0.0 ⎟
⎝ ⎠
TABLE I. TABLE TYPE STYLES THE DESIGN SPACE DIMENSIONS FOR
USER INTERFACE ARCHITECTURES [9] how to get evaluation matrices, which can be obtained by
consulting experts or other methods, we just simulate
Functional Dimensions (F1) Structural Dimensions(F2)
consulting an experts to obtain them.) For example, the
External event handling(F11) Application interface abstraction
level(F21) element ‘0.1’ in the matrix R11 represents the weight of the
.No external events .Monolithic program factor ‘no external events’ for the comment ‘extreme need’
.Process events while waiting for .Abstract device by experts’ opinions. According to B1 1 = A1 1 * R1 1 , we can
input
. External events preempt users .Toolkit get B11 = (0.57 0.11 0.14 0.18) . We can also get B12, B13, B14, B15,
commands
User customizability(F12) .Interaction manager with fixed B16, B21, B22, B23, B24 and B25 in the same way (we need
data types another 10 matrices here, such as R12, R13, R14, R15, R16, R21,

501

Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2011 at 05:50:38 UTC from IEEE Xplore. Restrictions apply.
R22, R23, R24 and R25, which are shown in appendix 1. In A. Identify the criterion
addition, we must get the weights A12, A13, A14, A15, A16, A21, At first the developers evaluate the specific application
A22, A23, A24 and A25 from clients or software designers.). by their existing knowledge and experience, according to the
Then we will make an evaluation for the above outcome, weights also given by their experience
namely second-stage evaluation. The evaluation matrices in
J = { A11 , A12 , A13 , A14 , A15 , A16 , A 21 , A 22 , A 23 , A 24 , A 25 , A1 , A 2 , A } ,
this stage are R1 = ( B11 B12 B13 B14 B15 B16 )T and R2 = ( B21 B22 B23 B24 B25 )T .
which are shown as follows:
Take F1={F11, F12, F13, F14, F15, F16} as an example to A11 =(0.1 0.2 0.7) , A12 =(0.6 0.3 0.1) , A13 =(0.7 0.1 0.1 0.1) ,
illustrate the second-stage fuzzy comprehensive evaluation in
A14 =(0.2 0.4 0.4) , A15 =(0.4 0.3 0.0 0.3 0.0) , A16 =(0.6 0.3 0.1) ,
detail, we assume that the weight for the six factors
is A1 = ( 0.1 0.1 0.2 0.2 0.2 0.2) , we get B1 = (0.416 0.23 0.134 0.22) in term A 21 =(0.1 0.1 0.2 0.2 0.2 0.2) , A 22 =(0.4 0.4 0.1 0.1) ,

of B1 = A1 * R1 , we can get B2 in the same way. A 23 =(0.2 0.1 0.1 0.1 0.2 0.3) , A 24 =(0.1 0.1 0.3 0.5) ,

Finally we need to evaluate the above results again by A 25 =(0.0 0.1 0.1 0.3 0.4) , A1 =(0.1 0.1 0.2 0.2 0.3 0.1) ,
fuzzy comprehensive evaluation, namely third-stage fuzzy A 2 =(0.2 0.1 0.3 0.1 0.3) , A=(0.7 0.3) . We get the evaluation
comprehensive evaluation. The matrix in this stage
is R = ( B1 B 2 )T . Take F={F1, F2} as an example to illustrate result B = ( 0.4277 0.2459 0.14206 0.1858) by third-stage fuzzy
the third -stage fuzzy comprehensive evaluation in detail. comprehensive in section 2, and the result is thought of as
The weights for this two factors is A = ( 0.7 0.3) , according the criterion for the specific application.

to B = A * R , we get B = ( 0.4277 0.2459 0.14206 0.1858) .What is said B. Two alternatives from different clients
above is a complete course of design space for user interface Different clients may have different opinion on the
architecture. specific application, which results in different alternatives in
design phase. To illustrate the problem effectively, we take
C. Choose a best alternative design for user interface two different alternatives for example. These two alternatives
architecture by degree of similarity and principle of are J1 and J2. The weights for the two alternatives are
choosing nearest 1 1 1 1 1 1 1 1 1 1 1 1 1 1
J1 = { A11 , A12 , A13 , A14 , A15 , A16 , A21 , A22 , A23 , A24 , A25 , A1 , A2 , A } and
The designer or clients often can not be consistent with 2 2 2 2 2 2 2 2 2 2 2 2 2 2
J 2 = { A11 , A12 , A13 , A14 , A15 , A16 , A21 , A22 , A23 , A24 , A25 , A1 , A2 , A }
the design alternatives for specific application in the early
respectively, which are given as follows:
phase of design. Having the method in section 2, along with 1 1 1
the methods--degree of similarity and principle of choosing A11 =(0.1 0.1 0.8) , A12 =(0.5 0.3 0.2) , A13 =(0.6 0.2 0.1 0.1) ,
nearest in this section, we can solve the inconsistent problem 1
A14 =(0.1 0.6 0.3) , 1
A15 =(0.3 0.3 0.1 0.3 0.0) , 1
A16 =(0.7 0.2 0.1) ,
in the design. When it comes into different design 1 1 1
alternatives, the developer can evaluate these alternatives by A21=(0.1 0.0 0.1 0.3 0.2 0.3) , A22 =(0.5 0.3 0.2 0.0) , A23 =(0.1 0.1 0.2 0.2 0.1 0.3) ,
the third-stage fuzzy comprehensive evaluation in section 2 1
A24 =(0.1 0.0 0.4 0.5) , 1
A25 =(0.1 0.0 0.2 0.1 0.4) , 1
A1 =(0.2 0.1 0.1 0.3 0.2 0.1) ,
and then according to the existing experience to choose the 1 2
best design alternative. A2 =(0.1 0.1 0.4 0.2 0.2) , 2
A11 =(0.3 0.5 0.2) , A12 =(0.1 0.3 0.6) ,
Given that J = { A1 1 , A1 2 , A1 3 , A1 4 , A1 5 , A1 6 , A 2 1 , A 2 2 , 2
A13 =(0.4 0.4 0.0 0.2) , 2
A14 =(0.7 0.2 0.1) , 2
A15=(0.2 0.2 0.3 0.1 0.2) ,
A23 , A24 , A25 , A1, A2 , A} and J is a fuzzy set in the design space U for 2 2
A16 =(0.2 0.6 0.2) , A21 =(0.3 0.2 0.1 0.1 0.1 0.2) ,
user interface architecture, namely, design alternative. If 2 2 2
there are several design alternatives in U , that A 22 =(0.1 0.2 0.5 0.2) , A23 =(0.0 0.3 0.2 0.3 0.1 0.1) , A24 =(0.6 0.3 0.1 0.0) ,
is S = { J 1 , J 2 , ..., J n } , we first adopt the third –stage fuzzy 2 2 2
A25 =(0.4 0.3 0.2 0.1 0.0) , A1 =(0.3 0.2 0.1 0.1 0.1 0.2) , A2 =(0.4 0.3 0.0 0.2 0.1) ,
comprehensive evaluation to evaluate every alternative and A
1
= ( 0 .6 0 .4 ) , A 2 = ( 0 .2 0 .8 ) .
obtain B J , BJ , ..., BJn . And then choose the best one by degree
1 2 Evaluating the two alternatives respectively, we obtain
of similarity and principle of choosing nearest. If there exists BJ = ( 0.4454 0.2848 0.1454 0.1244) for J1 and B J = ( 0.213 0.1892 0.2798 0.318 )
1 2
i make be possible, we then consider J i
N ( B J , B ) = max[ N ( B J , B )]
i i
for J2. Apparently, the two results are different from each
1≤ j ≤ n
other, it was difficult for us to tell which result is better or
is the best design alternative in S . more feasible for design. In this paper we adopt degree of
IV. CASE STUDY similarity and principle of choosing nearest to select a best
one.
In this section, we will study a case how to quantify
design space by fuzzy comprehensive evaluation. There are C. Using degree of similarity and principle of choosing
two alternatives according to different clients for a specific nearest to select the best from two alternatives
application of user interface architecture. In actual we only By virtue of definition 1 in section 2.2, we calculate the
can choose one of them for design. We can achieve it by the degree of similarity of two results respectively.
methods in section 2 and 3.

502

Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2011 at 05:50:38 UTC from IEEE Xplore. Restrictions apply.
1 [5] He Xin-gui, Fuzzy theories and fuzzy techniques in knowledge
N (BJ , B) = 1 − ( 0 .4 4 5 4 − 0 .4 2 7 7 + 0 .2 8 4 9 − 0 .2 4 5 9 + 0 .1 4 5 4 − 0 .1 4 0 6 processing (2nd ed), National Denfence Industry Press, Jul 1998 (in
1 4
+ 0 .1 2 4 4 − 0 .1 8 5 8 ) = 0 .9 6 9 2 7 5 chinese)
[6] Studying Software Architecture Through Design Spaces and Rules,
1
N ( B J , B ) = 1 − ( 0.213 − 0.4277 + 0.1892 − 0.2459 + 0.2798 − 0.1406 Technical Report CMU/SEI-90-TR-18, Carnegie Mellon Univ., 1990
2 4 [7] Thomas G. Lane, A design space and design rules for user interface
+ 0.318 − 0.1858 ) = 0.8643
software architecture. Technical Report CMU/SEI-90-TR-22 ESD-
Because of N ( BJ , B ) > N ( BJ , B ) , according to principle of 90-TR-223, Carnegie Mellon University, Software Engineering
1 2 Institute, November 1990.
choosing nearest, we can conclude that BJ is more similar [8] http://www.cs.cmu.edu/afs/cs/project/tinkerarch/www/html/1997/lect
1
ures/17.DsgnGuid/quick_index.html
to B . So BJ is better than BJ and we choose BJ as design [9] Mary Shaw, and David Garlan. Software Architecture: Perspectives
1 2 1
alternative for next phase. on an Emerging Discipline. Tsinghua university Press, Prentice Hall,
1998
V. CONCLUSION [10] Yi Qing Yang, Shou Qing Wang, Mohammad Dulaimi, and Sui
Pheng Low, “A fuzzy quality function deployment system for
Taking the uncertainties in early design and the conflict buildable design decision-makings”, Automation in Construction,
between designer and clients into consideration, we present a Dec 2003, 381– 393
fuzzy comprehensive evaluation for quantifying design [11] Xie Jian-xi, Song Bi-feng, and Liu Dong-xia, Application of fuzzy
space. The approach bases on fuzzy mathematics and adopts QFD to Aircraft Top Hierachy Design, International Journal of Plant
Engineering and Management, Vol.9 No.4 December 2004.
fuzzy comprehensive evaluation to judge the extent to which
the alternative from clients meets specified requirements, and [12] S. Mohanty, V. K. Prasanna, S. Neema, and J. Davis, “Rapid Design
Space Exploration of Heterogeneous Embedded Systems using
then select the best one from several alternatives. The Symbolic Search and Multi-Granular Simulation,” Language
approach also bases on the experience of the designers, so it Compilers and Tools for Embedded System, 2002.
do not only avoid the infeasible of design alternative
resulting from clients’ ignorance, but also meets the APPENDIX1
requirement in certain degree. It facilitates the communica-
tion between the designers and clients in early phase of ⎛ 0.1 0.2 0.3 0.4 ⎞
, ⎛ 0.7 0.2 0.1 0.0 ⎞
,
⎛ 0.0 0.0 0.1 0.9 ⎞
,
R11 = ⎜ 0.0 0.1 0.2 0.7 ⎟ R12 = ⎜ 0.2
0.5 0.3 0.1 0.1 ⎟
R13 =
⎜ 0.4 0.3 0.2 0.1 ⎟
design, but it still has something to complete. For example, ⎝ 0.8 0.1 0.1 0.0 ⎠ ⎝ 0.5 0.2 0.1 ⎠ ⎜ 0.6
⎜ 0.5 0.2 0.2 0.0 ⎟

⎝ 0.4 0.1 0.0 ⎠
the evaluation matrices in this paper is simulated by
consulting experts, maybe other methods also can obtain ⎛ 0.9 0.1 0.0 0.0 ⎞
⎛ 0.0 0.1 0.5 0.4 ⎞ ⎜ 0.6 0.3 0.1 0.0 ⎟ ⎛ 0.6 0.4 0.0 0.0 ⎞
them, such as the analytic hierarchy process (AHP). It will be R14 = ⎜⎜ 0.2 0.5 0.2 0.1 ⎟ , = ⎜ 0.4 0.1 ⎟
, R16 = ⎜⎜ 0.2 0.3 0.4 0.1 ⎟ ,
⎟ R15 0.3 0.2 ⎟
our further task in next phase for quantifying design space ⎝ 0.4 0.5 0.1 0.0 ⎠ ⎜⎜ 0.2 0.2 0.2 0.4 ⎟ ⎝ 0.0 0.1 0.2 0.7 ⎠

study. ⎝ 0.3 0.2 0.4 0.1 ⎠

⎛ 0.0 0.2 0.3 0.5 ⎞


⎛ 0.7 0.2 0.1 0.0 ⎞
REFERENCES ⎜ 0.1 0.4 ⎟
⎜ 0.3
0.2
0.2
0.3
0.3 0.2 ⎟ , R22 =
⎜ 0.5 0.3 0.1 0.1 ⎟ ,
⎛ 0.5
⎜ 0.6
0.3 0.1 0.1 ⎞
0.1 ⎟ ,
⎜ 0.1 0.4 ⎟
0.2 0.1
R21 = ⎜ 0.7 0.0 ⎟
[1] G.F. Salazar and D.C. Brown, “The effects of construction knowledge ⎜ 0.4 0.2 0.2 0.2 ⎟ ⎜ 0.1 0.1 0.4
⎟ R23 =
0.2 0.1
⎜ 0.4 0.1 ⎟
on the automation of preliminary building design”, Proceedings 5th ⎜ 0.5
⎜ 0.6 0.2 0.2 0.1 ⎟

⎝ 0.2 0.4 0.3 ⎠
⎜ 0.4
0.5
0.3
0.0
0.2 0.1 ⎟
⎜ 0.6 ⎟
International Symposium on Robotics in Construction ,Tokyo, Japan, ⎝ 0.2 0.1 0.1 ⎠ ⎝ 0.3 0.1 0.0 ⎠

June 6 – 8, 1988. ⎛ 0.0 0.1 0.7 0.2 ⎞ T


[2] M.Fischer, “Characteristics of design-relevant constructability
⎛ 0.0 0.0 0.3 ⎜ 0.5 0.3 0.1 0.1 ⎟ R1 = ( B11 B12 B13 B14 B15 B16 ) ,
0.7 ⎞
R24 =
⎜ 0.0 0.2 0.4 0.4 ⎟ , ,
⎜ 0.1 0.5 0.3 0.1 ⎟ R = ( B B B B B ) T
knowledge”, Journal of Construction Engineering and Management ⎜ 0.4 0.3 R25 =
0.2 0.1 ⎟
⎜ 0.1 0.4 0.4 0.1 ⎟ 2 21 T22 23 24 25
⎜ 0.5 ⎟
ASCE 123 (3), 1997, 253– 260. ⎝ 0.4 0.1 ⎜ 0.6 0.3 0.1 0.0 ⎟ R = ( B1 B2 )
0.0 ⎠
⎜ 0.5 0.3 0.1 0.1 ⎟
[3] Liu Pu-yan and Wu Meng-da, fuzzy theory and its applications, ⎝ ⎠
National University of Defence Technology Press, Nov 1998 (in Note that: The other three evaluation matrices in our third-stage fuzzy
chinese) comprehensive evaluation compute by using the results of the first-stage
fuzzy comprehensive evaluation (R1 and R2 are shown as above) and the
[4] Yang Lun-biao and Gao Ying-yi, fuzzy mathematics theory and its
second-stage fuzzy comprehensive evaluation(R is shown as above)
applications, South China University of Technology Press, Mar
respectively.
2001(in chinese)

503

Authorized licensed use limited to: IEEE Xplore. Downloaded on April 08,2011 at 05:50:38 UTC from IEEE Xplore. Restrictions apply.

Вам также может понравиться