Vous êtes sur la page 1sur 13

Theory and Methodology

Evaluating attack helicopters by AHP based on linguistic variable


weight
Ching-Hsue Cheng
a,
*
, Kuo-Lung Yang
b
, Chia-Lung Hwang
b
a
Department of Mathematics, Chinese Military Academy, Fengshan, Kaohsiung 830, Taiwan, ROC
b
Graduate School of Resources Management, National Defense Management College, Chung-Ho, Taipei, Taiwan, ROC
Received 13 January 1997; accepted 21 October 1997
Abstract
In this paper, we propose a new method for evaluating weapon systems by analytical hierarchy process (AHP) based
on linguistic variable weight. Many researchers used fuzzy arithmetic operations with weight and attribute of computing
performance score to solve fuzzy MADM problems, such as S.M. Chen [Evaluating weapon systems using fuzzy ar-
ithmetic operations, Fuzzy sets and systems 77 (1996) 265276]. Chen used a large number of fuzzy arithmetic opera-
tions, which not only causes information (data) loss or more fuzziness, but also adds to the diculty and accuracy of
decision-making. Therefore, we use linguistic variable weight method to solve the above problems, and to avoid the
mistake of decision-making. Our method possesses intuition, in accord with human rethinking-model, and is close to
humanized uncertainty of language expression. We use many experts' viewpoints to build the membership function in
order to calculate the performance score, and identify expectation of the decision-maker to avoid the constraint of system
alternatives and subjective judgements of the decision-maker. In this paper, we also use linguistic variable method to
revise the method of C.H. Cheng [in: Proceedings of the Fourth National Conference on Defense Management, pp.
11131124] and R.R. Yager [Fuzzy Sets and Systems 1 (1978) 8795]. Firstly, we measure the importance of relative
weight, and use it to determine the centralization or dilation power of linguistic hedge. When the power of linguistic
hedge is determined, we can use it to calculate the results of decision-making. Finally, we construct a practical example
for evaluating attack helicopters to illustrate our proposed method. 1999 Elsevier Science B.V. All rights reserved.
Keywords: Military application; Fuzzy number; Analytic hierarchy process; Linguistic variable; Linguistic hedge; Fuzzy
multiple attribute decision making
1. Introduction
Evaluating weapon systems is a complex system
of interacting elements. For example, a good weapon
system requests good weapon performance and
minimal cost; the performance and cost depend on
improvement of science and technology and eco-
nomic resources; technology depends on ideas and
resources; ideas depend on politics for their ac-
ceptance and support; and so on. In such an in-
tricate network of factors, rst causes and then
nal eects cannot be identied easily. These
European Journal of Operational Research 116 (1999) 423435
*
Corresponding author. E-mail: chcheng@cc.cma.edu.tw.
0377-2217/99/$ see front matter 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 7 - 2 2 1 7 ( 9 8 ) 0 0 1 5 6 - 8
factors directly depend on the expectations of the
decision-maker, and on additional complex sub-
factors, etc. In the complex system, our minds
have not yet evolved to the point where we can
clearly see these ultimate relationships and readily
resolve important issues.
Since 1977, Saaty [1,2] proposed analytic hier-
archy process (AHP) as a decision aid to help solve
unstructured problems in economics, social and
management sciences. AHP has been applied in a
variety of contexts: from the simple everyday
problem of selecting a school to the complex
problems of designing alternative future outcomes
of a developing country, evaluating political can-
didacy, allocating energy resources, and so on. The
AHP enables the decision-makers to structure a
complex problem in the form of a simple hierarchy
and to evaluate a large number of quantitative and
qualitative factors in a systematic manner under
conicting multiple criteria. The application of
AHP to the complex problem usually involves four
major steps:
1. Break down the complex problem into a num-
ber of small constituent elements and then
structure the elements in a hierarchical form.
2. Make aseries of pairwise comparisons amongthe
elements according to a ratio scale 1, 3, 5, 7 and9.
3. Use the eigenvalue method to estimate the rela-
tive weights of the elements.
4. Aggregate these relative weights and synthesize
them for the nal measurement of given deci-
sion alternatives.
However, in complex systems, the experiences
and judgements of humans are represented by
linguistic and vague patterns, and it is not quan-
titatively digital (well-dened condition). There-
fore, by fuzzy set theory, we can give a much better
representation of these linguistic data and thus
further rene the evaluation methods (for some
reference papers see [37]). On the other hand, the
AHP has some shortcomings as shown below:
1. The AHP method is mainly used in nearly crisp
(non-fuzzy) decision applications.
2. The AHP method creates and deals with a very
unbalanced scale of judgement.
3. The AHP method does not take into account
the uncertainty associated with the mapping
of one's judgement to a number.
4. Ranking of the AHP method is rather impre-
cise.
5. The subjective judgement, selection and prefer-
ence of decision-makers have great inuence on
the AHP method.
Therefore, in our model of evaluating weapon
system, we propose AHP based on linguistic
variable weight method to overcome some of the
above shortcomings. Our method combines the
spirit of AHP with linguistic variable weight to
respond to attributes of weapon systems and ex-
pectation of the decision-maker.
This paper, can be summarized into four parts:
(1) In our concept, our method possesses intuition,
in accord with human rethinking-model, and is
close to humanized uncertainty of language ex-
pression. (2) In our methodology, we use many
experts' viewpoints to build the membership
function in order to calculate the performance
score, and identify expectation of the decision-
maker to avoid the constraint of system alterna-
tives and subjective judgements of the decision-
maker. (3) In our calculations, we measure the
importance of relative weight, and use it to deter-
mine the centralization or dilation power of lin-
guistic hedge. When the dilation power of linguistic
hedge is determined, we can use it to calculate the
results of decision-making. (4) We also use lin-
guistic variable method to revise the method of
Cheng [7] and Yager [8].
2. Arithmetic operations of fuzzy numbers, linguistic
variable and linguistic hedge
In this section, for the purpose of reference, we
will review some important denitions.
Denition 1 (Fuzzy set). Let X be a universe of
discourse,
~
A is a fuzzy subset of X if for all x X,
there is a number l ~
A
(x) [0Y 1[ assigned to repre-
sent the membership of x to
~
A, and l ~
A
(x) is called
the membership function of
~
A.
Denition 2 (Fuzzy number). A fuzzy number
~
A is a
normal and convex fuzzy subset of X. Here, the
``Normality'' implies that
424 C.-H. Cheng et al. / European Journal of Operational Research 116 (1999) 423435
x RY .
x
l
~
A
(x) = 1
and ``Convex'' means that
\x
1
XY x
2
XY \a [0Y 1[Y
l ~
A
(ax
1
(1 a)x
2
) P min(l ~
A
(x
1
)Y l ~
A
(x
2
))X
Denition 3 (Arithmetic operations of fuzzy num-
bers). A triangular fuzzy number
~
A [9] can be
dened by a triplet (a, b, c) shown in Fig. 1. The
membership function is dened as
l ~
A
(x) =
0Y x ` aY
x a
b a
Y a Tx TbY
c x
c b
Y b Tx TcY
0Y x b cX
V
b
b
b
b
b
b
`
b
b
b
b
b
b
X
(1)
The addition operations and multiplication
operations of the triangular fuzzy numbers are
expressed below:
Fuzzy number addition
(a
1
Y b
1
Y c
1
) (a
2
Y b
2
Y c
2
) = (a
1
a
2
Y b
1
b
2
Y c
1
c
2
)X
(2)
Fuzzy number multiplication
(a
1
Y b
1
Y c
1
) (a
2
Y b
2
Y c
2
) = (a
1
a
2
Y b
1
b
2
Y c
1
c
2
)X
(3)
Denition 4 (Linguistic variable). A variable whose
states are fuzzy numbers assigned to relevant
linguistic terms.
Denition 5. A linguistic variable [10] is character-
ized by a quintuple (xY 1(x)Y UY GY
~
M)X
1. x is the name of the value.
2. U is the universe of discourse, which is associat-
ed with the base variable u.
3. T(x) denotes the term set of x, that is, the set of
the name of linguistic value of x, with each val-
ue being a fuzzy variable denoted generically by
x and ranging over U.
4. G is a syntactic rule for generating the name X,
of values of x. A particular X, that is name gen-
erated by G, is called a term.
5. M is a semantic rule for associating with each X
its meaning,
~
M(x) which is a fuzzy subset U.
Denition 6 (Linguistic hedge). Linguistic hedges
are special linguistic terms by which other linguis-
tic terms are modied. Linguistic terms such as
very, more or less, fairly, or extremely are exam-
ples of linguistic hedges.
Fig. 1. A triangle fuzzy numbers
~
A.
C.-H. Cheng et al. / European Journal of Operational Research 116 (1999) 423435 425
Denition 7. A linguistic hedge or a modier [10] is
an operation that modies the meaning of a term
more generally, of a fuzzy set. If
~
A is a fuzzy set
then the modier m generates the (composite) term
~
B = m (
~
A)X
Mathematical models frequently used for
modiers are:
Concentration (see Fig. 2.):
l
con(
~
A)
(u) = (l ~
A
(u))
n
Y where n b 1X (4)
Dilation (see Fig. 3.):
l
dil(
~
A)
(u) = (l
~
A
(u))
1an
where n b 1X (5)
Denition 8 (The classication of linguistic hedge).
In Ref. [11], fuzzy linguistic hedges and their
approximate meanings are classied in Table 1. By
experience, we obtain the dilation power n [1Y 4[
and the centralization power n [1Y 8[. In this
paper, we determine the weight power by practical
condition and use Table 1 for evaluating the
weapon system problem.
3. A new approach for decision-making
In this section, we will point out Chen's weak-
ness in fuzzy MADM, and using the linguistic
variable of fuzzy set theory overcome Chen's
weakness (Section 3.1). Using AHP based on lin-
guistic variable weight can revise Yager's method
and Cheng's method. Hence, we explain and
present our new method for AHP based on lin-
guistic variable weight in Section 3.2.
Fig. 2. Concentration
~
A.
Fig. 3. Dilation
~
A.
426 C.-H. Cheng et al. / European Journal of Operational Research 116 (1999) 423435
3.1. The weakness of fuzzy operations
Chen [6] used a large number of fuzzy arithmetic
operations with weight and attribute of computing
performance score to solve fuzzy MADM prob-
lems, which lead to mistakes in decision-making.
That is, when we use a large number of fuzzy ar-
ithmetic operations, it might not only cause some
information loss or more fuzziness, but also add to
the diculty and accuracy of decision-making.
In Chen's example, fuzzy numbers
~
A and
~
B are
two triangular fuzzy numbers, where
~
A = (4Y 5Y 6)
and
~
B = (8Y 10Y 12).
Then based on Eqs. (2) and (3), we can get
~
A
~
B = (4Y 5Y 6) (8Y 10Y 12) = (12Y 15Y 18)Y
~
A
~
B = (4Y 5Y 6) (8Y 10Y 12) = (32Y 50Y 72)X
The results of the above fuzzy number opera-
tions are shown in Fig. 4. From Fig. 4, we can see
that as a result of addition and multiplication
operations for fuzzy numbers, two slopes of tri-
angular fuzzy number approximate horizon line,
and the base value U increase two times in the
interval of the bottom edge; it shows that the in-
formation is more fuzzy.
For weight and attribute of object, it is not
reasonable that one uses the digital to directly
represent, and this is based on the experiences of
an enormously large number of individuals; and it
cannot be characterized precisely. Due to the fact
that experiences and judgements of humans are
usually represented by words in their natural lan-
guage, specially, in complex systems, by employing
fuzzy sets, we can describe the object approxi-
mately. So, we can use the linguistic variable of the
fuzzy set theory to treat reasonable models in
complex systems, and using the linguistic variable
can eectively overcome Chen's weakness.
3.2. AHP based on linguistic variable weight method
From Section 3.1, we know that the underlying
power of fuzzy set theory is that it uses linguistic
variables, rather than quantitative variables, to
represent imprecise concepts. Linguistic variables
Fig. 4. The operation of fuzzy numbers.
Table 1
Fuzzy linguistic hedges and their approximate meanings
Hedge Meaning
Quite, rather, somewhat Dilate a fuzzy region
Generally, usually Contrast diusion
Vicinity of Approximate broadly
Neighboring, close to Approximate narrowly
Almost, denitely positively Contrast intensication
Very, extremely Intensify a fuzzy region
C.-H. Cheng et al. / European Journal of Operational Research 116 (1999) 423435 427
were developed to quantitatively analyze the
vagueness of human language. Generally, ``words''
are not as accurate as ``numbers''. Hence, lin-
guistic variables provide us with a method to
quantify those complex or ill-conditioned quanti-
tative descriptions.
In this part, we modify Yager's [8] and Cheng's
[7] methods to obtain our new method for the
AHP based on linguistic variable weights. Yager's
weight derived from Saaty's AHP model, has the
weakness of loss of information; and his compar-
ison matrix is obtained under unbalanced scales.
When we get Yager's weight by the maximal ei-
genvalue corresponding to the eigenvector method
based on the comparison matrix with unbalanced
scales, then the weight may lose objectiveness.
Cheng [7] used catastrophe series based on fuzzy
scales to derive weight, and he determined optimal
solutions by a single centralizing membership
function to get the weight value, and found that
the background of catastrophe theory is not easy
to understand. Especially, Cheng's method cannot
be eectively treated under equal weight. There-
fore, we propose the AHP based on linguistic
variable weight to modify above methods and
solve complex evaluating systems.
A human's thought process cannot directly
determine the weight value, she can describe by
language and balance its importance correspond-
ing to attribute, and the weight comes from the
language. From the denition of linguistic hedge,
take ``importance'' as a term, and calculate the
centralization or dilation power of the weight
value, and then select the minimal membership
function of attribute under its alternatives, and
rank all of the alternatives. The best alternative is
the maximal membership function.
In other words, our method possesses intuition,
in accord with human rethinking-model, and is
close to humanized uncertainty of language ex-
pression. We consult many experts' viewpoints to
structure the membership function in order to
calculate the performance score, and identify the
object and expectation of the decision-maker to
avoid the constraint of system alternatives and
subjective judgements of the decision-maker. We
measure the importance of relative weight, and use
its importance to determine the centralization or
dilation power of linguistic hedge. When the power
of linguistic hedge is determined, we can use it to
calculate the results of decision-making.
For example, weighting the age by ``old'', we
usually use ``very old, old, somewhat old'' to des-
cribe the degree of oldness. Hence, we can use the
denition of linguistic hedge to analyze the prob-
lem, take ``old'' as a term, and structure its mem-
bership as in Eq. (6).
Let w(u) =very old, y(u) =old, z(u) =somewhat
old, that is, centralizing or dilating the membership
function, y(u) =old, can be determined by linguistic
description, where w(u): centralizing weight power
is 2, y(u): weight power is 1, and z(u): dilating
weight power is
1
2
, such as in Fig. 5.
l
old
(u) =
0Y u [0Y 50[Y
(1 (
u50
5
)
2
)
1
Y u [50Y 100[X
@
y(u) = 1
u 50
5

2
4 5
1
Y
:(u) = 1 [
u 50
5

2
5
1a2
Y
w(u) = 1
u 50
5

2
4 5
2
X (6)
From Fig. 5, when the linguistic mode is cer-
tainly for the term, by centralizing the weight, and
using the operation method ``power > 1'', then the
Fig. 5. linguistic variable ``old''.
428 C.-H. Cheng et al. / European Journal of Operational Research 116 (1999) 423435
linguistic mode is moderate for the term, by di-
lating weight, and the operation method is ``pow-
er ` 1''. Secondly, if the interval [50, 100] has not
changed after the operation of the base variable,
then the degree of fuzziness also has not increased,
and this method possesses intuition, in accord with
human rethinking-model, and is close to human-
ized uncertainty of language expression.
4. The algorithm of our method and its application
for evaluating weapon systems
The evaluating model of complex weapon sys-
tem is a multiple attribute decision-making prob-
lem. Firstly, complex weapon system can best be
understood by breaking them down into their
constituent elements, structuring the elements hi-
erarchically, and then composing, or synthesizing
judgements on the relative importance of the ele-
ments at each level of the hierarchy into a set of
overall priorities.
From the above procedure, a hierarchy for
evaluating weapon systems is constructed. For the
performance data of weapon systems, we consult
many experts' viewpoints to construct the mem-
bership function in order to calculate the perfor-
mance score, or, utilize fuzzy language to set up
look-up table for language values, derive the cor-
responding mean of fuzzy numbers, and use it to
calculate the performance score. Then sum up all
sub-items' performance scores corresponding to
their criteria which is called total score, and nor-
malize the total score in every criteria. Thus, we
use AHP based on linguistic variable weight to
evaluate the best weapon system.
For easy computing, we summarize the algo-
rithm for evaluating weapon systems by AHP
based on linguistic variable weight which is listed
below:
Step 1: Construct the hierarchical organization
diagram such as in Fig. 6.
Step 2: Compute the performance score g
i
of
the sub-factor, and sum up all the scores corre-
sponding to its criteria.
Fig. 6. The structure model of evaluating three attack helicopters.
C.-H. Cheng et al. / European Journal of Operational Research 116 (1999) 423435 429
In this paper, we have two methods to obtain
the score:
1. The original data are represented by quantity,
after consulting many expert's opinions to con-
struct the membership function of the sub-item,
and use the membership function to compute g
i
values.
2. Utilize fuzzy language to construct the look-up
table for values, and derive its corresponding
value to the mean of fuzzy numbers as in Ta-
ble 2, and as shown in Fig. 7.
Step 3: Normalize all total scores for every
criteria, i.e.,
where
S
i
denotes possible alternatives; X

denotes at-
tributes with which alternative performances are
measured; x
i
denotes the performance score of
Table 2
Linguistic value look-up table
Fuzzy language The mean of fuzzy numbers
Very good 1
Good 0.75
Fair 0.5
Poor 0.25
Very poor 0
Fig. 7. Membership functions of linguistic values for criteria ratings.
430 C.-H. Cheng et al. / European Journal of Operational Research 116 (1999) 423435
alternative S
i
with respect to attribute X

and
l
i
(x

) [0Y 1[X
Step 4: Determine the power of dilation or
centralization. From Denitions 7 and 8, when the
decision-makers balance the importance of crite-
ria, and determine the power of dilation or cen-
tralization, we get Eq. (7).
Step 5: Determine the best alternative l~
D
(x
i
)
We take Eq. (8) to determine the best alternative,
which is to maximize the minimum membership
value over all the criteria. If alternative i has
the same maximum value l~
D
= (x
i
) in the deter-
mining process, we can take the sub-minimum
value l
w
i
in the criteria and by Eq. (8) determine
the best alternative. Eq. (8) is shown in the
following.
l~
D
(x
i
) = max
i
(min

l
w
i
)X (8)
5. Numerical example
In this section, we construct an example for
evaluating attack helicopters to illustrate our
proposed method. Since World War II, the ad-
vanced nations have devoted themselves to devel-
Table 3
The data of technological advances for three attack helicopters and their judgement criteria
Item S1 S2 S3 Membership function
1 Turbo-shafts (kW) 1633 2 1265 2 1285 2 l
m
=
(x 1100)a400Y 1100 Tx T1500
1Y 1500 Tx
&
2 Weight empty (kg) 7000 5092 4634 l
we
=
(8000 x)a4000Y 4000 Tx T8000
1Y x T4000
&
3 Max level speed (km/h) 300 293 282 l
m
=
(x 250)a70Y 250 Tx T320
1Y 320 Tx
&
4 Max disc loading (kg/m
2
) 49 56.69 39.80 l
md
=
(x 30)a40Y 30 Tx T70
1Y 70 Tx
&
5 Service ceiling (m) 5800 6400 4270 l
sc
=
(x 4000)a3000Y 4000 Tx T7000
1Y 7000 Tx
&
6 Maximal range standard
fuel (km)
460 482 507 l
mr
=
(x 400)a200Y 400 Tx T600
1Y 600 Tx
&
7 Endurance with maximal fuel 2h 3 h 9 min 2 h l
em
=
0X5Y 2h Tx ` 3h
1Y 3h
&
8 g-limits +3/ 0.5 +3.5/ 0.5 +2.5/ 0.5 l

gl
=
(x

2)a2Y 2 Tx

T4
1Y 4 Tx

&
9 Mission capable 90 93 92 l
mc
=
xa100Y 0 Tx T100
1Y 100 Tx
&
C.-H. Cheng et al. / European Journal of Operational Research 116 (1999) 423435 431
oping helicopters for military use. To date, the
attack helicopter equipped with velocity, re
power and exibility, is suitable for complex duty
in traditional and modern threats from air defense
weapons. An attack helicopter's repower of the
same infantry company, has special abilities, such
as: vertical hover and climb, ceiling in the air, not
subject to any terrain and climate, fast mobility,
and best protection equipment. At the same time,
the attack helicopter can support repower of
ground, anti-helicopter, and anti-tank operation,
etc. Therefore, all types of attack helicopters are
developed and produced. From changeable tactics
pattern, the attack helicopters have some im-
portant roles and properties, as listed in the
following:
1. Execute the repower support by dense contact,
tting integrated air and ground operation.
2. Generalize the monitoring of war and combat
region.
3. Be equipped for night operation and have the
ability of ying on terrain features.
4. Not subject to small base, is suitable for eld
operation.
After visiting the unit of attack helicopters and
staying for a long time, we have obtained a lot of
information by practical contact and experiences.
Therefore, we propose that the best attack heli-
copter can be evaluated by ve criteria; They are:
technological advance, logistic capability, arma-
ment, avionics, and subsisting ability. In [12], we
select S1(MI-28), (S2)AH-64, and S3(AH-1w) at-
tack helicopters as our evaluating entities. Their
criteria, sub-criteria and computational procedure
are detailed below.
1. Structure the hierarchical gure of attack
helicopters as in Fig. 6.
2. From Section 4, we can evaluate the sub-
factors for the ve criteria, which are listed in
Tables 310.
(1) Technological advances.
(2) Logistic capability.
(3) Armament: The sub-factors of armament
are guns, anti-tank missiles, air-to-air missiles, and
rockets. For convenience and without loss of
generality, we assume equal weightings on these
sub-factors. The scores are listed in Table 7. In the
case of dierent weightings, the algorithm remains
unchanged except that the powers of dilation or
centralization could be weighted according to the
linguistic description. In addition, normalization is
used to transform this evaluation into discrete
membership functions, combined with other cri-
teria to select the best weapon system.
(4) Avionics.
(5) Subsist ability.
3. Normalize the total scores: The original
matrix is shown in Eq. (9), and the normalized
matrix is shown in Eq. (10)
Table 4
The scores of technological advances derived from Table 3
Item S1 S2 S3
1 Turbo-shafts 1 0.41 0.46
2 Weight empty 0.25 0.72 0.84
3 Max level speed 0.71 0.61 0.46
4 Max disc loading 0.49 0.67 0.25
5 Service ceiling 0.6 0.8 0.09
6 Max range standard fuel 0.3 0.41 0.54
7 Endurance with Max fuel 0.5 1 0.5
8 g-limits 0.5 1 0.25
9 Mission capable 0.9 0.93 0.92
Total 5.25 6.55 4.31
Table 5
Expert evaluations of logistic capability represented by linguistic terms
Item S1 S2 S3
1 Reliability Fair Good Good
2 Maintenance ability Very good Good Good
3 Convey Fair Very good Good
4 Economics Very good Good Very good
5 Flexibility for selecting weapon Good Very good Good
432 C.-H. Cheng et al. / European Journal of Operational Research 116 (1999) 423435
4. Determine the power of dilation or central-
ization: The decision-makers balance the impor-
tance of criteria, and determine the power of
dilation or centralization. The results are listed
below.
1. Weapon performance (X
1
) is ``very important'',
the power of centralization is 1.5.
Table 7
The data of armament for three attack helicopters and its judgement criteria
Item S1 S2 S3 Membership function
Gun Caliber (mm) 30 30 20 l
gc
=
0X5Y 20 Tx ` 30
1Y 30 = x
&
Firing rate (r/m) 900 625 650 l
fr
=
(x 500)a500Y 500 Tx T1000
1Y 1000 Tx
&
Feed 300 1200 750 l
gf
=
(x 200)a1000Y 200 Tx ` 1200
1Y 1200 Tx
&
Anti-tank missiles Feed 16 16 8 l
af
=
xa16Y 0 Tx ` 16
1Y 16 Tx
&
Firing range (km) 5 8 8 l
at
=
xa10Y 0 Tx ` 10
1Y 10 Tx
&
Firing accuracy (%) 80 76 87 l
aa
=
xa100Y 0 Tx ` 100
1Y 100 Tx
&
Air-to-air missiles Feed 8 4 4 l
at
=
xa10Y 0 Tx ` 10
1Y 10 Tx
&
Firing accuracy (%) 85 90 50 l
ia
=
xa100Y 0 Tx ` 100
1Y 100 Tx
&
ggRockets Feed 20 4 19 4 19 4 l
rf
=
xa100Y 0 Tx ` 100
1Y 100 Tx
&
Caliber (mm) 70 70 70 l
rf
=
xa100Y 0 Tx ` 100
1Y 100 Tx
&
Table 6
The scores of logistic capability derived from Table 5
Item S1 S2 S3
1 Reliability 0.5 0.75 0.75
2 Maintenance ability 1 0.75 0.75
3 Convey 0.5 1 0.75
4 Economics 1 0.75 1
5 Flexibility for selecting weapon 0.75 1 0.75
Total 3.75 4.25 4.0
C.-H. Cheng et al. / European Journal of Operational Research 116 (1999) 423435 433
2. Avionics (X
2
) is ``positively important'', the
power of centralization is 1.3.
3. Armament (X
3
) is ``important'', the power is 1.
4. Subsist ability (X
4
) is ``usually important'', the
power of dilation is 0.8.
5. Logistic capability (X
5
) is ``somewhat impor-
tant'', the power of dilation is 0.5.
5. Determine the best alternative:
1. Take the minimum in each alternative, we get
S1 =0.151 S2 =0.248 S3 =0.143.
2. Rank it orderings, S2 > S1 > S3.
3. Take the maximum among all comparative al-
ternatives, S2 is the best choice.
6. Conclusions
In the complicated systems of weapon evalu-
ating, some attributes are described by language or
ambiguous expressions and some quantitative re-
quirements are represented by quality. Hence it is
suitable to be analyzed by traditional models. Here
we utilize the hierarchy structure diagram to
structure complicated problems and combine fuz-
zy theory to deal with some vague or not well-
dened language variables and qualitative re-
quirements. In this paper, we use a new method for
evaluating weapon systems by AHP based on lin-
guistic variable weights. It is close to vague lan-
guage expressions of human beings and coincides
with to the model of human thinking with the
following three advantages:
1. Modify Cheng [7] and Yagers' [8] method and
eectively improve Chen's [6] drawback.
2. Improve the AHP method and smooth the eval-
uation process.
3. Reduce fuzzy computation to avoid informa-
tion from becoming more fuzzy and make cause
information loss.
We construct membership functions to evaluate
the qualitative requirements of attack helicopters.
This denes the expectation of decision-makers in
advance so that the nal decision cannot be swayed
by system alternatives and decision-makers' ob-
jective judgements. From numerical illustrations,
we have seen the thoughtfulness, exibility and
eciency of the proposed method, especially with
subjectivity valuations of the decision-makers.
Table 8
The scores of armament derived form Table 7
Item S1 S2 S3
Gun Caliber (mm) 1 0.5
Firing rate (r/m) 0.8 0.25 0.3
Feed 0.1 1 0.75
Anti-tank
missiles
Feed 1 1 0.75
Firing range (km) 0.5 0.8 0.8
Firing accuracy (%) 0.8 0.76 0.87
Air-to-air
missiles
Feed 0.8 0.4 0.4
Firing accuracy (%) 0.85 0.9 0.5
Rockets Feed 0.8 0.76 0.76
Caliber (mm) 0.7 0.7 0.7
Total 7.35 7.57 6.08
Table 10
The scores of subsist ability
Item S1 S2 S3
1 Armor-protection 0.75 0.75 0.75
2 Counter-detected 0.75 1 0.5
3 Pilot-protested 1 0.75 0.5
4 Noise 1 1 0.5
5 NBS protection 1 0.5 0.5
Total 4.5 4 2.75
Table 9
The scores of avionics
Item S1 S2 S3
1 Pilot night vision system 0.5 1 0.25
2 Target acquisition and
designation system
0.5 1 0.5
3 Integrate system 0.25 1 0.5
4 Global positioning system 0.5 1 0.5
Total 1.75 4 1.75
434 C.-H. Cheng et al. / European Journal of Operational Research 116 (1999) 423435
References
[1] T.L. Saaty, The Analytic Hierarchy Process, McGraw-
Hill, New York, 1980.
[2] T.L. Saaty, Decision Making for Leader: The Analytic
Hierarchy Process for Decisions in a Complex World,
RWS Publications, Pittsburgh, PA, 1988.
[3] D.L. Mon, C.H. Cheng, J.C. Lin, Evaluating weapon
system using fuzzy analytic hierarchy process based on
entropy weight, Fuzzy Sets and Systems 62 (1994) 127
134.
[4] C.H. Cheng, D.L. Mon, Evaluating weapon system by
analytic hierarchy process based on fuzzy scales, Fuzzy
Sets and Systems 63 (1994) 110.
[5] C.H. Cheng, Evaluating naval tactical missile system by
AHP based on the grade of membership function, Euro-
pean Journal of Operational Research 96 (2) (1997) 343
350.
[6] S.M. Chen, Evaluating weapon systems using fuzzy
arithmetic operations, Fuzzy Sets and Systems 77 (1996)
265276.
[7] C.H. Cheng, Determining main attack direction-quantify-
ing the Sun Tzu on the art of war, in: Proceedings of the
Fourth National Conference on Defense Management, pp.
11131124.
[8] R.R. Yager, Fuzzy decision making including unequal
objectives, Fuzzy Sets and Systems 1 (1978) 8795.
[9] A. Kaufmann, M.M. Gupta, Introduction to Fuzzy
Arithmetic Theory and Applications, Van Nostrand
Reinhold, New York, 1991.
[10] H.J. Zimmermann, Fuzzy Set Theory and its Applications,
Kluwer Academic Publishers, Boston, MA, 1991.
[11] E. Cox, The Fuzzy System Handbook, Academic Press,
New York, 1994.
[12] M. Lanbert, Janes all the world's craft 19931994, Janes
Information Group (1993).
C.-H. Cheng et al. / European Journal of Operational Research 116 (1999) 423435 435

Vous aimerez peut-être aussi