4, November 2014
I.
INTRODUCTION
PRELIMINARIES
Definition
The characteristic function A of a crisp set A
~A
: X
~
A =
Definition
A fuzzy number
~
A = (m, , ) is said to be triangular
~
xj 0
mx
,
1-
~A (x) =
m- x<m,
1-
> 0
m
= (m, , ) is said to be a
x +m+
,
~A (x) =
> 0
x<m+,
Definition
The concept of Fuzzy decision making was first proposed
by Bellman & Zadeh, (1970) [4].A linear programming
problem is called fuzzy variable linear programming
problem(FVLPP), if some of the parameters are crisp, and
variables and right hand sides are fuzzy numbers. General
form of FVLPP as follows:
~z
cj ~
xj
=
Where
(F(R))
~
c j Rn , A ij (R)mn, Bi
~
x
m
n
j
(F(R))
Definition
Kolman & Hill,(1984) [10] was introduced a FFLP
problem. A linear programming problem is called fully fuzzy
linear programming problem (FFLPP), if parameters and
variables are all fuzzy numbers. General form of FFLPP as
follows:
Max (min)
~
if and only if R ( a
)R(
~
b )
~
~
a > b
~
if and only if R ( a
)R(
~
b )
~
a = ~
b
Let
~
if and only if R ( a
~
A = (a1, b1, c1) and
)=R(
~
b )
~
B = (a2, b2, c2) be two
(i)
~ ~
A B
(ii)
c2 - b2
~ ~
A B iff a1 a2 , b1- a1 b2 -a2, c1-b1
b2 -a2,c1 -b1
c2 - b2
~
~
(iii) A = B iff a1=a2 , b1= b2c1 = c2
I.
~ ~
Subject to A ij x j Bi ,
~
xj 0
~
~
a b
Max (min)
maps each fuzzy number into the real line, where a natural
order exists. We define orders on F(R) by:
Definition
m
> 0
0,
R :F(R) R which
Definition
x + m
,
~
x j (F(R)) n
Definition
An effective approach for ordering the elements of F(R) is
otherwise
~
A fuzzy number A
~
~
c j (F(R))n, ~
A ij (F(R))mn, Bi (F(R))
Where
xm
,m x<m+, > 0
0,
~~ ~
A ij x j Bi ,
Subject to
EARLIER METHOD
~
~z
cj ~
xj
=
~ ~
C x )
~
x , =, , ~
b
A ~
~
x is a non-negative triangular fuzzy number and
~
c j]1n ~
a ij]mn ~
x =[ ~
x j] n1
C =[ ~
A =[ ~
~
~
b = [ b i] m1
Subject
4x1 + 2x2+s1= 10
6x1+ 8x2+ s2= 8
2y1+ 2y2+ t1= 6
Example
6y1+6y2+ t2= 2
~
x 1+ (4, 4, 4) ~
x
Max (6, 4, 4)
~
x
Subject to (4, 2, 2)
+ (2, 2, 2)
~
x
(6, 6, 6)
~
x 1, ~
x
~
x
(10, 6, 6)
~
x
+ (8, 6, 6)
~
x 1= (1.33, 0.33, 0.33),
~z
~
x 1+ (4, 4, 4) ~
x
~
x
Subject to (4, 2, 2)
1
+ (2, 2, 2)
~
x
III.
+ (1, 1, 1)
~
x 2+ (1, 1, 1)
Algorithm
Step 1: Formulate the chosen problem into the following
fully fuzzy linear programming problem:
= (8, 2, 2)
~
x
~
x 2,
1,
~s
, ~s
PROPOSED METHOD
~
x
+ (8, 6, 6)
~
x 2 = (0, 0, 0) and Max
= 10.
= (10, 6, 6)
(6, 6, 6)
~s
y2 x2 0
(8, 2, 2)
Solution:
~s
y1 x1 0
Max (min)
~
~z
cj ~
xj
=
Subject to
~~ ~
A ij x j Bi ,
fuzzy numbers
~
xj 0
2:
Using
the
m+2
~
c ,~
A =
2
Ranking
function
R (
~
M method. Let the solution be x
Subject to
(4x1 + 2x2+s1, 2y1 + 2y2+ t1, 2z1 + 2z2+ u1) = (10, 6, 6)
(6x1 + 8x2+s2, 6y1 +6y2+ t2, 6z1 +6z2+ u2) = (8, 2, 2)
(x1, y1, z1), (x2, y2, z2), (s1, t1, u1) (s2, t2, u2) are non-negative
triangular fuzzy numbers.
The above FFLPP is converted into the following Crisp
linear programming:
1
Max ( 4
Step
~
FFLPPis x j .
Step 4: Convert all the inequality constraints into equations
by adding slack / surplus variable and the cost of this variable
zero.
Step 5: Compute the value of
j=1.n, If all
~
Z
0 j
~
Z
=CBYj-Cj
jB
~
Z < 0
and
Standard form
~
x k, which will be
~
x
~
( B }, i=1, 2.,m in maximization problem and k =
~
argmax { R ( B }, i=1, 2.,m in minimization
~
x
~
x 1, ~
x
+ (2, 2, 2)
~
x
~
x 1, ~
x 2, ~
x
Example 4.1
~
+0 x
~
x
~
+0 x
+ (1, 1, 1)
+ (8, 6, 6)
~
x 2+ (1, 1, 1)
= (8, 2, 2)
~
(6, 6, 6) x
= (10, 6, 6)
(6, 6, 6)
problem.
~
Subject to (4, 2, 2) x
~
x
Subject to (4, 2, 2)
~
~
Max (6, 4, 4) x 1+ (4, 4, 4) x
~
x 1+ (4, 4, 4) ~
x
Max (6, 4, 4)
~
x
3,
~
Max 5 x
~
x
1 + (2, 2, 2)
~
x
1 + (8, 6, 6)
(10, 6, 6)
~
+4 x
~
x
Subject to 3
~
6 x
(8, 2, 2)
~
x
+2
~
+5 x
~
x 1, ~
x 2, ~
x
~
+0 x
3,
~
+0 x
~
x
~
x 4= (8, 2, 2)
~
x
= (10, 6, 6)
Solution:
Table 4.1.1
~
x
Basis
~
x
~
x
~
x
~
x
~
x
RHS
~ m+2
B =
2
(10, 6, 6)
(8, 2, 2)
-5
-4
(0, 0, 0)
~z
~z
, so
~
x 1is an
5. So
~
x
6.
Table: 4.1.2
Basis
~
x
~
x
~
x
~z
~
x
~
x
~
x
3
6
5
6
1
6
1
6
5
6
3
6
RHS
36
30
30
,
,
6
6
6 )
8
2
2
6 , 6 , 6 )
40
10
10
,
,
6
6
6 )
Since
problem is
~
Z 0,the fuzzy optimal solution of the FVLP
~
x
= (
8
6 ,
2
6 ,
2
6 )),
~
x
40
10
10
6 , 6 , 6 ) = 10
= (
IV.
Table: 5.1
Earlier method
Sl.
no
1
2
3
4
5
6
7
8
9
10
Number of
constraints
Number of
variables
2
2
2
3
3
4
4
5
5
6
2
2
2
3
3
4
4
5
5
6
Number of
Iterations
Fuzzy optimal
9
10
10
18
14
17
16
20
19
26
10
53
62
154
216
75
112
214
198
327
[3]
[4]
[5]
[6]
[7]
[9]
[10]
[11]
[12]
[13]
REFERENCES
[2]
value
[8]
CONCLUSION
[1]
Proposed method
[14]
[15]
[16]
[17]
[18]
~z
Number of
Iterations
Fuzzy optimal
2
2
2
2
2
3
3
4
6
8
10
53
62
154
216
75
112
214
198
327
value
~z