An International

Journal
computers &
mathematics
PERGAMON Computers and Mathematics with Applications 45 (2003) 1601-1610
www.elsevier.nl/locate/camwa

The Canonical Representation of Multiplication

Operation on Triangular Fuzzy Numbers
CH.-CH. CHOU
Department of International Trade, Ta Hwa Institute of Technology
1, Ta Hwa Road, Chung-Lin 307, Hsin-Chu, Taiwan, R.O.C.
and
Department of Shipping and Transportation Management
National Taiwan Ocean University
2, Peining Road, Keelung 202, Taiwan, R.O.C.
chienchlQms58.hinet.net

(Received and accepted April 2002)

Abstract-The representation of multiplication operation on fuzzy numbers is very useful and

important in the fuzzy system such as the fuzzy decision making. In this paper, we propose a new
arithmetical principle and a new arithmetical method for the arithmetical operations on fuzzy num-
bers. The new arithmetical principle is the L-‘-Rml inverse function arithmetic principle. Based on
the L-‘-R-’ inverse function arithmetic principle, it is easy to interpret the multiplication operation
with the membership functions of fuzzy numbers. The new arithmetical method is the graded multi-
ple integrals representation method. Based on the graded multiple integrals representation method,
it is easy to compute the canonical representation of multiplication operation on fuzzy numbers.
Finally, the canonical representation is applied to a numerical example of fuzzy decision. @ 2003
Elsevier Science Ltd. All rights reserved.

Keywords-fizzy numbers, Fuzzy arithmetic, Fuzzy decision, Canonical representation of mul-

tiplication operation.

1. INTRODUCTION
The concept of fuzzy sets, which was introduced by Zadeh [l], led to the definition of the fuzzy
number and its implementation in fuzzy control and approximate reasoning problems. The
basic arithmetic structure for fuzzy numbers was developed by Mizumoto and Tanaka [2,3],
Nahmias [4], Dubois and Prade [5,6], Li [7], and Ma et al. [8]. The arithmetic operation was
established either by the extension principle or by observing the fuzzy number as a collection of
o-levels. Unfortunately, although there are many arithmetical operation approaches, none of these
approaches is easy to interpret the multiplication operation with the membership functions of
fuzzy numbers. On the other hand, none of these approaches is easy to compute the representation
of multiplication operation on fuzzy numbers.
In this paper, we propose a new arithmetical principle and a new arithmetical method for the
arithmetical operations on fuzzy numbers. The new arithmetical principle is the L-l-R-l inverse
function arithmetic principle. Based on the L-l-R-’ inverse function arithmetic principle, it is

08981221/03/S - see front matter @ 2003 Elsevier Science Ltd. All rights reserved. Typeset by &S-QX
PII: SO898-1221(03)00139-l
1602 CH.-CH.CHOU

easy to interpret the multiplication operation with the membership functions of fuzzy numbers.
The new arithmetical method is the graded multiple integrals representation method. Based on
the graded multiple integration representation method, it is easy to compute the representation of
multiplication operation on fuzzy numbers. In Section 2, we present the representation of addition
operation associated with the L-l-R-’ inverse function arithmetic principle. In Section 3, we
present the canonical representation of multiplication operation associated with the L-‘-R-’
inverse function arithmetic principle and the graded multiple integrals representation method. A
numerical example of fuzzy decision is shown in Section 4 followed by conclusion in Section 5.

2. THE REPRESENTATION OF ADDITION OPERATION

In this section, we introduce the addition operation on triangular fuzzy numbers associated
with the L-‘-R-’ inverse function arithmetic principle. First, we define the L-l-R-l inverse
function arithmetic principle of the addition operation as Definition 1. Second, we introduce
briefly the graded mean integration representation method. Third, according to the L-l-R-’
inverse function arithmetic principle and the graded mean integration representation method, we
define the representation of the addition operation on fuzzy numbers as Definition 2. Finally,
according to Definition 2, we prove the representation of the addition operation on triangular
fuzzy numbers (Property 1).
First, we define the L-l-R-’ inverse function arithmetic principle of the addition as follows.
DEFINITION 1. Let Al = (q,ul, b,) and AZ = (c2,a2,b2) be two triangular fuzzy numbers as
Figure 1. The addition of Al and A2 at h-level is

(
Am)@A2(w= Li:(h)+ Li:(h)’
Li:(h)+ Rii:(h)’
Rli:(h)
+ L&), R&) + Ri:(h) >
LAG and RAN are the functions L and R of fuzzy number Al, respectively. L&,,, and R$,,
are the inverse functions of functions LAG and RAN at h-level, respectively. LAG and RAN are the
functions L and R of fuzzy number AZ, respectively. LJ&,, and R,k(,) are the inverse functions
of functions LA2 and R,Q at h-level, respectively.
Suppose the membership function of Al = (cl, al, bl) is

fAl(z) = (x - h)
al L3:Ih,
(al - bl)’

0, otherwise.
Since

(x -4
LA1(x)= (al _ c1) 7 Cl 121a1,

RAIL = ;;l-;jj, al L x I bl,

and

LA:+, = ~1 + (ai - cl)h, O<h<l,

R&, = bl + (~1 - bl)h, O<h<l.

Similarly, suppose the membership function of A2 = (cz, ~2, b2) is

(r - c2)
~2 <xIa2,
(a2 - c2)'

fA2b) = (x -b2) a2 <% <b

2,
(a2-b2)’ - -
0, otherwise.
 Multiplication Operation on Triangular Fuzzy Numbers 1603

Since

(x - c2)
‘5A2c2) = (a2 _ C2j 7 c2 i 2 I a2,

(x - b2)
&2(x) = (a2 _ b2), a2 5 5 5 b2r

and

G(h)= ~2 + (a2 - c2)h Olhll,

K&h, = bz + (a2 - bz)h, O<h<l.

According to Definition 1, we have that

(
Aw @Aw = L&t) + L&&pL&h)+ R&q,q&) + L&), R&) + R-J&&) >
= [(Cl + (al -W)+(c2 +(a2 - c2)h),(c1 + (al - cl)h)

+ (b2 + (a2 - b2)h), (bl +( al - bl)h) + (~2 -t (a2 - cz)h), (bl + (al - bl)h)
-t (bz + (a2 - b@)].

L-l Aloo R’ AIW L-'rzaa$' AZW

L, v
mt R1w

(=R’ *1&w AZ(h))

Figure 1. The addition of A1 and AZ with L-‘-R-’ inverse function arithmetic

principle.

Second, we introduce briefly the graded mean integration representation method. Chen and
Hsieh [9] proposed the graded mean integration representation method of fuzzy numbers based
on the integral value of graded mean h-level of generalized fuzzy number. Her++ we describe the
meaning as follows.
In general, a generalized fuzzy number A is described as any fuzzy subset of the real line R,
whose membership function uA satisfies the following conditions.
(1) uA is a continuous mapping from R to the closed interval [0,11;
(2) uA(“) = 0, --oo < 5 5 c;
(3) uA(x) is strictly increskrg on [c, a];
(4) uA(x) = w, a 5 x < b, where 0 < w < 1;
(5) uA(x) is strictly decreasing on [b, d];
(6) ?‘A@) = 0, d 5 x 5 00.
Here a, 6, c, and d are real numbers. We denote generalized fuzzy number A in Figure 2 as
A = (c, a, b, d; wA)LR. When WA = 1, we simplify the notation as A = (c, a, b, d)LR.
1604 CR.-CH. CHOU

Figure 2. The graded mean h-level of fuzzy number A = (c, a, b, d; UJA)LR

Let L-l and R-’ be the inverse functions of the functions L and R, respectively; then the
graded mean h-level value of generalized number A is h(L~t + RT;L’))/‘Las in Figure 2. Then, the
graded mean integration representation of A is

s ((
WA
h LG\ + Rrhf /2 dh
> >
P(A) = O PJJA

I 0
hdh

WA (h(c + (a - c)h + d + (b - d)h)/2) dh

EC s 0 = ;(c+2a+2b+d),
WA
hdh
s0

where 0 < h 5 WA and 0 < WA 5 1. Formula (1) is equal to the formula proposed by Delgado et
al. [lo].
Generalized triangular fuzzy number Y = (c, a, b; w) is a special case of generalized trapezoidal
fuzzy number. The graded mean integration representation of the triangular fuzzy number Y
becomes
P(Y) = i(c+ 4a + b). (2)

Third, according to the L-l-R-l inverse function arithmetic principle (Definition 1) and the
graded mean integration representation method, we define the representation of the addition
operation on triangular fuzzy numbers as Definition 2.

DEFINITION 2. Let P(A1(h~ CBAZ(h)) be the representation of A1 CBAZ at h-level. Let P(A1 @AZ)
be the representation of Al $ AZ.

P (Al(h)@b(h)) = ;h (G(h)+2 G(h))+ (L&&,+2 G(h))+ (%&IL)

+ G(h))
2

+ (Ra:(h)
+K&h,)
2 ‘)
I
(3)

1
11
R-1 L-1 1
Al(h) A2(h) c&l) + G(h))
+ dh hdh.
0
2 2
 Multiplication Operation on Triangular Fuzzy Numbers 1605

PROPERTY 1.

%4+42)=;(cl+~2+4a1+4a~+bl+~~),

P(Ad + W2) = P(A1 CDA2).

PROOF. First, we prove

Wl @ A2) = f (cl+ c2 + 4al + da2 + bI + b2).

By formula (3), the representation of A1 CBA2 is

+(Gh, +~&4)+ (G(h)+G(h))

/J
dh 1 hdh
2 2 0
I

=
J [ l &

+‘,c:
(Cl

+ (al -
+ (a1

c1)h
-

+
Cl)h

b2 +
+
2
c2 +

(a2 -
(a2

bz)h)
- c2)h)

+ (bl + (al - b+ : c2(a2 - c2)h)

2

/J
+ (h + (al - h)h + bz(w - bz)h) dh l hdh
2 0
1

J
‘1
= -h[(cl + ~2 + bl + b2)

+
0 2
1
(-1 + 01 + al - bl - c2 + a2 + a2 - b2)h] dh
lJ hdh

= [(1/2)(1/2)h2(cl + ~2 + h + bz) + (1,2)(l,3)h3(-cl’+ al + al - bl - c2 + a2 + a2 _ b2)] 1;

( 0/2)h21:)

= $1 + C2 + 4a1 + 4~32+ bl + b2).

We have that
P(A1 @ A2) = f (CI+ cz + 4~x1+ 4a2 + bl + bz). (4)

Second, we prove P(A1) + P(A2) = P(A1 $ A2). By formula (2)

1 1
WI) = -(cl + 4~x1 + bl), J’(A2) = -(c2 + 4a2 + b2),
6 6

By formula (4),

J’(A1 @ A2) = f(c1 + c2 + 4al + 4@ + bl + b2).

We have that

%‘h) + W2) = J’(Al

@ A2) = f (cl+ c2 + 4al + 4U2+ bl + bz).

Hence, we complete the proof.

1606 CH.-CH. CHOU

For example, let AI = (q,al,bl) = (1,2,3) and AZ = ( Q,Q, b2) = (6,7,8) be two triangular
fuzzy numbers as in Figure 1. By formula (2), the representation of Al is P(A1) = 2, and the
representation of A2 is P(A2) = 7. P(Al)+P(As) = 2+7 = 9. By formula (4) with the L-‘-R-’
inverse function arithmetic principle, the representation of Al @ A2 is

P(AI 83 A2) = $1 + c2 + 4al f 4U2 + bl + b2) = 9.

We have that
P(A1) + P(A2) = 9 = P(A1 $ As).

3. THE CANONICAL REPRESENTATION

OF MULTIPLICATION OPERATION
Now, we introduce the canonical representation of multiplication operation associated with the
L-‘-W1 inverse arithmetic principle and the graded multiple integrals representation method.
First, we define the L-1-R-1 inverse arithmetic principle of the multiplication operation as
follows.

DEFINITION 3. Let A1 = (cl, al, bl) and A2 = (~2, ~2, b2) be two trianguhr fuzzy numbers as in
Figure 3. The multiplication of Al and A2 at h-level is

Aw) @Aw) = (Li:(h)Li:(h)~

L&)RA:(h)~ R&)R&q>
Ri:(h)La:ch)l

Figure 3. The multiplication of A1 and AZ with L-‘-R-’ inverse function arithmetic

principle.

Second, we define the graded multiple integrals representation method as follows.

DEFINITION 4. Let P(A l(h) @AZ(~)) be the representation of Al @ A2 at h-level. Let P(A1 @AZ)
be the representation of Al 8 AZ.
 Multiplication Operation on Triangular Fuzzy Numbers 1607

PROPERTY 2.

P(Al C+As) = ~(CI + 4al + bl)i(cz ‘+ 4az + bz),

P(& 8 -42) = P(AdP(A2).

PROOF. First, we prove

P(Al@ A2) = $1 +4Ul +bl);(c2 +4U2 + b2).

By formula (5), the representation of A1 @ A2 is

P(AI @Ad = l’lll; [(kd&) (hAZL&)) + (halL,;(,)) (k42R,:(,~)

+ (hA%tt~,)) (hA2Li;[,)) + (hAIR,:(,,) @42%:(q)]

hAlA2dhAldhA2dhtlA2

' (Jd1hn~dhm~1hazdha2~1hAla2dhnla2)
1 1
zz ’ +hAl[cl + (al - Cl)hAl]hA2[C2 + (a2 - C2)ht2]
S/J
0 0 04
+ hAl[Cl+(UI - +Al]hA2[b2 f(U2 - bz)hAz]
+ hAl[bl + (al - h)hAl]hA2[C2 + (a2 - cz)hAz]

+ hAl[bl + (a~ - h)hAl]hA2[b2 + (a2 - bz)hAz]}

hAlA dhAl dhA2 dhAlA2

’ (~1hAld~Al~lhA2dhA2~1hAlA2dhAlA2)

&cl + $&l - cl) hA2[C2 + (a2 - cz)hAz]

=1X{ [i I

+ $&cl
1 + ;h3,l(al - cl) hA2[b2 + (a2 - b2)hA2]

+$,A+$$(a~ 1hA2[‘4
-h) +
[

(a2 - cz)hAz]

+[;hk,bl+
;h;,(
al-h)
1hm[b
+
[

(a2 - b2)hA2]

h/nI;
dh/tz
dh/tlm
hA2[C2 + (a2 - cz)hAz]

hA2[b2 + (a2 - b2)hA2]

h/&
+ (a2 - cz)hAz]

hAlA2dh.42dhA1z-w
I hA2[b2 + (a2 - bz)hAz]
>
x( (l/2) (l/2)%,1;(1/2%,&)
ih2 1
= 2 A2c2+$2(U2-c2)
I[ I
1608 CH.-CH. CHOU

+
[
;c1+$1 - Cl)
I[ I
+$1+f(a1- IIlh2
[ 1 bl) 2 A2c2 +
1
$:,(a2 - c2)

[ I[ 11
1 1
+ yb1 + $Jl - f-Q) ;h;,bz + ;hi2(az - 62)

hm1:~hza,m
’ (WVWh2A21~
(1/2)hf4~/4
/’ -I[ II;c2
+$a2-cq)
I
ll
= Ic, + $1 - Cl)
04 2

+[+I+$q-Cl)
II;bz+ I ;(a2 - bz)

+ I[fc2+;(a2
-c2)
;bl+;(arbl)
I
+
[

hA1.a
[ I[
;bl

dhAl,u 11
+ ;(a~ - bl) ;bz + ;(a, - b2)

’ ((l/2)
(l/2)
(l/+;,A&)
I[;c2+;(a,
- 1
1
=- Ll + $(a1 - Cl) c2)
4 {[ 2

+[$1+f(q-q)
I[ I fb2 + ;(a2 - bz)

+ I[;c2+;(a2
$bl + $1 -I - bl) c2)

W’)h?,,,a~~
[

I[;bz+ 11(l/2)
( W) (l/W2 $2 - bz)
AlA21:)

={ [;@1+2d] [;+‘2+2az)] + [~(cI+~cQ)] [;(b2+2a2)]

+[f(b1+2Q)] [++2az)] + [;(61+2al)] [;(b2+2a2)]}

= f(Q + 4a1 + bl);(c2 + 4a2 + b2).

We have that

P(A1 CT9
-42) = $1 + 4~1 + b&2 + 4U2 f b2). (6)
Second, we prove P(A1) @ P(A2) = P(A1 8 AZ). By formula (2),

1 1
P(AI) = -(cl + 4~1 + bl), P(A2) = -(CZ + 4~2 + h),
6 6

P(A1) @ P(A2) = $(c~ + 4~1 + b&(ca + 4Q + b2).

By formula (6), P(A1 @AZ) = (1/6)(q + 4~1 + b1)(1/6)(c2 + 4a2 + b2). We have that P(A1) @
P(A2) = P(A1 @ AZ). Hence, we complete the proof. I
For example, let Al = (cl, al, bl) = (O,l, 2) and A2 = (~2, a2, b2) = (2,3,4) be two triangular
fuzzy numbers as in Figure 3. By formula (2), the representation of Al is P(A1) = 1, and the
 Multiplication Operation on Triangular Fuzzy Numbers 1609

representation of A2 is P(A2) = 3. P(A1) @ P(A2) = 1 x 3 = 3. By formula (6) with the graded

multiple integrals representation method, the representation of A1 @I A2 is

P(A1@A2)=~(~1+4a1+b1);(cz+4az+b~)=lx3=3.

We have that P(A1) 8 P(A2) = 3 = P(A1 @CIAZ). Finally, let us see more examples. Let
A = (1, 1,1) be a real number. B = (2,3,4) is a triangular fuzzy number.
EXAMPLE 1. A @ A. The representation of A is P(A). By formula (2),

P(A) = ;(c+4a+b) = 1, P(A) @I P(A) = 1 x 1 = 1.

By formula (6) with the graded multiple integrals representation method, the representation of
A1 @A2 is

P(A1 @A2) = $1 + 4~1 + b&c2 + 4U2 + b2), P(A@A)=lxl=l.

We have that P(A) @ P(A) = P(A @ A) = 1.

EXAMPLE 2. A @ B. By formula (2), P(A) = 1, and P(B) = 3. P(A) @I P(B) = 1 x 3 = 3. By
formula (6), P(A @I B) = 3. We have that P(A) @P(B) = P(A @I B).
EXAMPLE 3. B @aA. By formula (2), P(B) = 3, and P(A) = 1. P(B) @I P(A) = 3 x 1 = 3. By
formula (6), P(B @ A) = 3. W e have that P(B) CXJ
P(A) = P(B @A).
According to Examples 2 and 3, we have that P(A) @ P(B) = P(A @ B) = P(B) @ P(A) =
P(B @ A).
EXAMPLE 4. B @I B. By formula (2), P(B) = 3. P(B) @I P(B) = 3 x 3 = 9. By formula (6),
P(B QDB) = 9. We have that P(B) @P(B) = P(B @ B).

4. A NUMERICAL EXAMPLE
Alternatives: Al and Az. Criteria: Cr and C 2. Weights: Wi and Wz. All fuzzy numbers in
Table 1 are triangular fuzzy numbers.

Table 1. Basic data

Criteria
Alternative
Cl c2

Criteria weight = (0.3,0.4,0.8) Criteria weight = (0.4,0.5,0.9)

‘41
Preference value = (0.8,0.9,1.0) Preference value = (0.5,0.6,0.7)

Criteria weight = (0.1,0.4,1.0) Criteria weight = (0.3,0.5,1.0)

A2

Preference value = (0.6,0.7,0.8) Preference value = (0.7,0.8,0.9)

Let the total performance value of alternative 1 be P ~1. Let the total performance value of alter-
native 2 be P,Q. PAr = (Preference value)Arci x (Criteria weight)Arcr+(Preference value),Jrcz x
(Criteria weight)Arcz. PA2 = (Preference vahie)Azcr x (Criteria weight)Azci + (Preference
vahre),Qcz x (Criteria weight)AzCs.- PA1 = (0.8,0.9,1.0)(0.3,0.4,0.8) + (0.5,0.6,0.7)(0.4,0.5,
0.9) = 0.735; P_42 = (0.6,0.7,0.8)(0.1,0.4,1.0) + (0.7,0.8,0.9)(0.3,0.5,1.0) = 0.755. We have
that PAN < PAZ. The alternative Az is selected.

5. CONCLUSION
The representation of multiplication operation on fuzzy numbers is very useful and important
in the fuzzy systems. In this paper, we propose a new arithmetical principle (L-l-R-’ inverse
1610 CH.-CH. CHOU

function arithmetic principle) and a new arithmetical method (graded multiple integrals repre-
sentation method) for the multiplication operations on fuzzy numbers. Based on the L-l-R-l
inverse function arithmetic principle and graded multiple integrals representation method, we
can obtain easily the canonical representation of multiplication operation. Finally, the canonical
representation is applied to a numerical example of fuzzy decision. In fact, the canonical repre-
sentation of multiplication operation not only can be applied to the fuzzy decision, but also can
be applied to many fuzzy fields.

REFERENCES
1. L.A. Zadeh, Fuzzy sets, Inform. and Control 8, 338-353, (1965).
2. M. Mizumoto and K. Tanaka, The four operations of arithmetic on fuzzy numbers, Systems Comput. Controls
7 (5), 73-81, (1976).
3. M. Mizumoto and K. Tanaka, Some properties of fuzzy numbers, In Advances in Fuzzy Set Theory and Ap-
plications, (Edited by M.M. Gupta, R.K. Ragade and RR. Yager), pp. 156-164, North-Holland, Amsterdam,
(1979).
4. S. Nahmias, Fuzzy variables, FUL+YSets and Systems, 97-111, (1978).
5. D. Dubois and H. P&e, Operations on fuzzy numbers, J. Systems Sci. 9, 613-626, (1978).
6. D. Dubois and H. Prade, Fuzzy number: An overview, In The Analysis of Fuzzy Information, Volume 1:
Mathematics, (Edited by J.C. Bezdek), pp. 3-39, CRC Press, Boca Flaton, FL, (1987).
7. R.J. Li, Fuzzy method in group decision making, Computers Math. Applic. 38 (l), 91-101, (1999).
8. M. Ma, M. Friedman and A. Kandel, A new fuzzy arithmetic, Fuzzy Sets and Systems 108, 83-90, (1999).
9. S.H. Chen and C.H. Hsieh, Graded mean integration representation of generalized fuzzy number, In Proceeding
of 1998 Sixth Conference on Fuzzy Theory and Its Application, (CD-ROM), filename: 031. udl, pp. l-6,
Chinese Fuzzy Systems Association, Taiwan, (1998).
10. M. Delgado, M.A. Vila and W. Voxman, On a canonical representation of fuzzy numbers, FUZZYSeta and
Systems 93, 125-135, (1998).