Int. J. Contemp. Math. Sciences, Vol. 7, 2012, no.

4, 159 - 170

Solving Nonlinear Programming Problem

in Fuzzy Environment
Ali F. Jameel

and Amir Sadeghi

School of Mathematical Sciences, Universiti Sains Malaysia

11800 USM, Penang, Malaysia
Abstract
In this paper, the fuzzy nonlinear programming problem is discussed.
In order to obtain more accurate solution, the properties of fuzzy set
and fuzzy number with linear membership function and fuzzy maximum
decision maker is utilized to fuzzifying the crisp problem . An example
is provided to show the eectiveness of the proposed method.

Keywords: Fuzzy numbers, Nonlinear programming, Optimization problem, Fuzzy maximum decision making

Introduction

Fuzzy nonlinear programming problem (FNLPP) is useful in solving problems

which are dicult, impossible to solve due to the imprecise, subjective nature
of the problem formulation or have an accurate solution. In this paper we
will discuss the concepts of fuzzy decision making introduced by [1] and the
maximum decision [15] that is used in NLPP to nd the optimal decision
(solution). This decision making used in fuzzy linear programming problem
[8] and [7]. Furthermore, this problems has fuzzy objective function and fuzzy
variables in the constraints [13], [10] and [5] where the fuzzy left and right
hand side coecients on constraints [14]. In addition, the fuzzy NLPP is used
in quadratic programming [9, 11] which has fuzzy multi objective function and
fuzzy parameters on constraints so in our NLPP that have fuzzy properties on
1

Corresponding author (alifareed1@yahoo.com)

160

A. F. Jameel and A. Sadeghi

 )
 and have fuzzy linear membership function. The outline
the inequality (,
of this paper is as follows: In Section 2 we introduce some essential denitions
that are useful in our problem. Section 3 we state the nonlinear programming
in fuzzy environment by transform the crisp problem. A numerical example
will be given in Section 4 and the conclusions will be presented in Section5.

Preliminary

In this section several necessary basic concepts are recalled.

Definition 2.1: [15] If x is a collection of objects denoted generally by X,
 in X is a set of order pairs
then a fuzzy set A
 = {(x, ) : x X}
A
A

(2.1)

where A : x [0, 1] is called the membership function or grade of membership

 which maps x to the membership range M (when M contains only
of x in A
 is non fuzzy and is identical to the characteristic
the two points 0 and 1), A
A
function of crisp set. It should be emphasized that the range of membership
function is a subset of the non-negative real numbers. The elements with a
zero degree of membership are normally not listed.
Definition 2.2: [2] The function L : X [0, 1] is a function with two parameters dened as:

if x <

1,
L(x; , ) =

+x

0,

if x < +

(2.2)

if x >

It is called the trapezoidal linear membership function. This type of fuzzy

number is very useful which has a large non convex fuzzy rejoin set. Clearly,
it gives us the high degree of A . (See the Fig. 1)
 and
Definition 2.3: [1] Given a fuzzy goal (fuzzy objective function) G
 in a space of alternatives X. The G
 and C
 combine to
fuzzy constraints C
 which is a fuzzy set resulting from intersection of G
 and
form a decision, D,

Solving nonlinear programming problem in fuzzy environment

Figure 1: L-function

161

Figure 2: Fuzzy decision

 Furthermore, D
 =G
C
 is the membership function of D
 can be dened
C.
1 , . . . , G
n and m
as D = min{G , C }. In general, if we have n goals G
1 , . . . , C
m ,then, the resultant decision can be dened as
constraints C
=G
1 . . . G
n C
1 . . . C
m
D

(2.3)

Therefore, for j = 1, 2, . . . , n and i = 1, 2, . . . , m it can be written as follows

D = min{min{Gj }, min{Ci }}

= min{G1 , . . . , Gn , C1 , . . . , Cm }
= min{Gj , . . . , Ci }

(2.4)

(See the Fig. 2)

Remarks: If the decision-maker wants to have crisp decision proposal, it is
appropriate to suggest which of them have the highest degree of membership
in the fuzzy set decision. The maximizing decision is dened by
Xmax = max MD (x) = max{Dj (x), Ci (x)}
x

(2.5)

i are in Denition 2.3, for i = 1, 2, . . . , m, j = 1, 2, . . . , n [15].

 j and C
Where D

Fuzzy nonlinear programming problems

In this section we discuss the optimization problem with nonlinear fuzzy objective function and fuzzy exible nonlinear constraints. Consider the following
non linear programming problem:
Min/Maxf (x)
Subject to
gi (x) ()bi ,

i = 1, 2, . . . , m

(3.1)

162

A. F. Jameel and A. Sadeghi

For all x Rn and x 0. The fuzzy version of the problem (3.1) is

Min/Maxf (x)
Subject to
 )b
 i,
gi (x)(

i = 1, 2, . . . , m

(3.2)

For all x Rn and x 0. In problem (3.2), the tilde sign denotes a fuzzy
satisfaction of the constraints. It is clear that these constraints are exible
constrains. The fuzzy max (min) corresponds to achieving the highest (lowest)
possible aspiration level for the general f (x). This problem can be solved
by using the properties of fuzzy decision making problems (3.1) and (3.2) as
follows:
Step 1: Fuzzify the objective function by calculating the lower and the upper
bounds of the optimal values. The bounds of optimal values zl and zu can be
obtained by solving the standard crisp NLPP as follows:
z1 = Min/Maxf (x)
Subject to
gi (x) ()bi ,

i = 1, 2, . . . , m

(3.3)

For all x Rn and x 0 and

z2 = Min/Maxf (x)
Subject to
gi (x) ()bi + pi ,

i = 1, 2, . . . , m

(3.4)

For all x Rn and x 0. Where the objective function take the values between
z1 and z2 . Let zl = min(z1 , z2 ) and zu = max(z1 , z2 ), where zl and zu are the
 is the fuzzy set
lower and upper bounds of the optimal values. Suppose M
 = {(x, (x)) : x Rn },
representing the objective function f (x) such that M
M
where

if zu < f (x)

1,
f (x)zl
(3.5)
M
if zl < f (x) < zu
(x) =
zu zl

0,
if zl > f (x)

Solving nonlinear programming problem in fuzzy environment

163

Note that, pi is vector of relaxation and can be determined by fuzzifying bi

(denoted by bi ) by using the denition of L-function of the membership function as follows
bi = {(x, (x)) : x Rn }
bi
where

bi (x) =

1,

bi +pi x
pi

0,

Figure 3: Represent bi (x) membership function

if x < bi
if bi < x < bi + pi

(3.6)

if x > bi + pi

Figure 4: membership function of

3 (x)

i is the fuzzy
Step 2: Fuzzify the constraint gi (x), i = 1, 2, . . . , m. Let C
i = {(x, (x)) : x Rn }, where
set for ith constraints such that C
Ci

if gi (x) < bi
1,

Ci (x) = bi +pipgi (x) if bi < gi(x) < bi + pi

(3.7)
i

0,
if gi (x) > bi + pi
 be the fuzzy decision set, where
Let D
 =M
C
i ,
D

i = 1, 2, . . . , m

(3.8)

m , and D
 = {(x, (x)) : x Rn }. Then we
 =M
C
1 . . . C
Therefore, D
M
have
D (x) = min{M
(x), min{C1 (x), . . . , Cm (x)}}.
Now, if we suppose
= min{M
(x), min{C1 (x), . . . , Cm (x)}}

(3.9)

164

A. F. Jameel and A. Sadeghi

then we have the optimal decision:

x = max ,

x Rn

Now, the problem (3.2) becomes the following crisp NLPP

Max
Subject to
g 1 : M
(x) 0
g 2 : C1 (x) 0
..
.

(3.10)

g m : Cm1 (x) 0
g m+1 : Cm (x) 0
where 0 1, x 0 and x Rn . This is equivalent to the problem
Max
Subject to


g1 :

g2 :

f (z) zl
zu zl

b1 + p1 gl (x)
p1

..
.

g m+1 :

bm + pm gm (x)
pm

(3.11)

0

where 0 1, x 0 and x Rn . After that, we can obtain the optimal

solution x Rn and substitute problem (3.1) in the objective function. It
can be easily seen that
zl < zAF < zu
where zAF is the objective function after fuzziness.

Solving nonlinear programming problem in fuzzy environment

165

Numerical example

In the following example, we will illustrate presented theory. Our computation

is carried out by utilizing MATHLAB 7.2. Suppose we are going to make a
box for airport shipping [3]. The box is made of two materials; the top of the
box is made of a material costing $17 per square foot and the rest of the box is
made of material that cost $3 per square foot. Then, the baggage restrictions
require that the dimensions of the box end of it must sum to at most 3 foot.
The dimensions of the box of should be a maximum volume costing and not
more than $108. Mathematical description x1 , x2 and x3 are present the length,
width and height of the box, respectively. The NLP optimization problem is
Maxx1 x2 x3 = z
Subject to
g1 : x2 + x3 3
g2 : 17x1 x2 + 3(x1 x2 + 2x1 x3 + 2x2 x3 ) 108
x1 , x2 , x3 0

(4.1)

The optimal solution is x1 = 2.99526551, x1 = 1.00316204, and x1 = 1.99687905.

Therefore z = 5.99996738 satises the constraints of problem (4.1). Now, the
fuzzy version of the problem is
Maxx1 x2 x3 = z
Subject to

g1 : x2 + x3 3

g2 : 17x1 x2 + 3(x1 x2 + 2x1 x3 + 2x2 x3 )108
x1 , x2 , x3 0
Therefore, b1 = 3 and b2 = 108. In order to obtain p1 and p2 , we have

3 = {(x, 3 (x)) : x Rn }
where
3 (x) =

1,

5x

0,

if x < 3
if 3 x 5
if x 5

(4.2)

166

A. F. Jameel and A. Sadeghi

Figure 5: membership function of 1

08 (x)

It is shown in the Figure (4). Hence p1 = 2. Similarly p2 can be obtained by

= {(x,  (x)) : x Rn }
108
108
where

1
08 (x) =

1,

120x

12

0,

if x < 108
if 108 x 120
if x 108

We get p2 = 12. It is shown in the Figure (5). Now, we can nd zl and zu by

solving the two crisp NLPP as follows:
1. z1 = z. Since the problem is the same rst problem, and they have the same
solution, therfore z1 = 5.99996738$.
2. z2 = Maxx1 x2 x3
Subject to

x1 = 1.77424067,

x2 = 1.44655028,

x3 = 3.55344549.

 be the fuzzy
which is satisfy the constraints. Finally, z2 = 9.12001855. Let M
set of all objective function such that
 = {(x, (x)) : x R}
M
M

Solving nonlinear programming problem in fuzzy environment

167

and

M
(x) =

1,

5.999987738 < x1 x2 x3

if

x1 x2 x3 5.999987738

9.120018555.999987738

0,
if

if 5.9999 < x1 x2 x3 < 9.1200

x1 x2 x3 < 9.12001855

1 be the fuzzy set for g1 such that

In addition, let C
1 = {(x, (x)) : x R}
C
C1
where
C1 (x) =

1,

5(x1 +x3 )
2

0,

if x2 + x3 < 3
if 3 < x2 + x3 < 5
if 5 < x2 + x3

2 be the fuzzy set for g2 such that

and C
2 = {(x, (x)) : x R}
C
C2
where

C1 (x) =

1,

if 17x1 x2 + 3(x1 x2 + 2x1 x3 + 2x2 x3 ) < 108

0,

if 120 < 17x1 x2 + 3(x1 x2 + 2x1 x3 + 2x2 x3 )

120(17x1 x2 +3(x1 x2 +2x1 x3 +2x2 x3 ))

12

if 108 < 17x1 x2 + 3(x1 x2 + 2x1 x3 + 2x2 x3 ) < 120

The fuzzy decision making for this problem is

D (x) = min{M
(x), min{C1 (x), C2 (x)}}.

For = min{M
(x), min{C1 (x), C2 (x)}}, with optimal decision x = Max.
Finally, the crisp NLP corresponding with the fuzzy NLP is given by

Max
Subject to
g 1 : M
(x) 0
g 2 : C1 (x) 0
g 3 : C2 (x) 0

168

A. F. Jameel and A. Sadeghi

where x 0 and 0 1 which is equivalent to the following problem

MaxX =
Subject to


x1 x2 x3 5.999987738
g1 :
0
9.12001855 5.999987738

g2 :

g3 :

5 (x1 x2 )
2

120 (17x1 x2 + 3(x1 x2 + 2x1 x3 + 2x2 x3 ))

12

(4.3)

Where x1 , x2 , x3 0 and 0 1. Therefore, the solution of problem (4.3)

is
x1 = 2.3047,

x2 = 1.2152,

x3 = 2.7300,

and = 0.5274 which satisfy in the constraints while the result of crisp
problem before fuzziness is
x1 = 2.99526551,

x2 = 1.00316204,

x3 = 1.99687905.

Now, we can submit , x1 , x2 and x3 in the objective function of the crisp NLP.
It can be obtained
zAF = x1 x2 x3 = 7.64583302.
where zl < zAF < zu . Clearly, in comparison the crisp problem we have more
accurate solution.

Conclusion

In this work, the fuzzy solution of optimization problems and insensitive solution to the optimization problems are presented. Furthermore, it is proposed
that the results solution of fuzzy optimization is a generalization of the solution of the crisp optimization problem. In addition, the numerical experiment
show us the solution in fuzzied problems are more accurate than results in
crisp problems.

Solving nonlinear programming problem in fuzzy environment

169

References
[1] R.E. Bellman and L.A. Zadeh, Decision-making in a fuzzy environment.
Management Sci, Vol. 17, No. 4 (1970) pp. 141-164.
[2] D. Drinkov, H. Hellendoorn, and M. Reinfrank, An Introduction to Fuzzy
Control, Springer, Berlin, etc., 1993.
[3] Iyengar. P, Non-Linear Programming; Introduction, IEOR, Handout 19,
16 October 2002.
[4] Z. A. Kanaya , An Interactive Method for Fuzzy Multi objective Nonlinear
Programming Problems, JKAU Sci., Vol. 22, No. 1 (2010) pp. 103-112.
[5] B. Kheirfam F. Hasani Sensitivity analysis for fuzzy linear Programming
problems with Fuzzy variables, Advanced Model and Optimization, Vol.
12, No. 2, 2010.
[6] N. Mahdavi-Amiri and S.H. Nasseri, Duality in fuzzy variable linear programming, Transactions on Engineering, Computing and Technology, Vol.
6 (2005) pp. 115-118.
[7] S.H. Nasseri, A new method for solving fuzzy linear programming problems
by solving linear programming, Applied Mathematical Sciences Journal,
Vol. 50 (2008) pp. 2473-2480.
[8] S.H. Nasseri, E. Ardil, A. Yazdani and R. Zaefarian, Simplex method for
solving linear programming problems with fuzzy numbers, Transactions on
Engineering, Computing and Technology, Vol. 10 (2005) pp. 284-288.
[9] S.H. Nasseri, Fuzzy nonlinear optimization, The Journal of Nonlinear
Analysis and its Applications, Vol. 1, No. 4 (2008) pp. 230-235.
[10] V. Pandian, R. Nagarajan, and S. Yaacob, Fuzzy Linear Programming: A
Modern Tool for Decision Making, Jurnal Teknologi, Vol. 37 (2002) pp.
31-44.
[11] N. Ravi Shankar 1, G. Ananda Rao, J. Madhu Latha and V. Sireesha ,
Solving a Fuzzy Nonlinear Optimization Problem by Genetic Algorithm,
Int. J. Contemp. Math. Sciences, Vol. 5, No. 16, (2010) pp. 791 -803.

170

A. F. Jameel and A. Sadeghi

[12] E. K. Saadettin, K. D. Sanchoy, and Z. Gungor,A Mixed Integer

Programming Formulation For Scheduling Of Virtual Manufacturing
Cells(VCMS),International Journal Of Advance Manufacturing Technology, Vol. 2, No. 5 (2010) pp. 82-85.
[13] H. C. Wu, Duality Theory in Fuzzy Optimization Problems, Fuzzy Optimization and Decision Making, Vol. 3, (2004) pp. 345-365.
[14] M. Yenilmez, N. Rafail, and K. Gasimor, Solving Fuzzy Liner Programming Problems with Linear Membership Function, Turk H. Math,
TUBITAK, No. 26, (2002) pp. 375-369.
[15] H.J. Zimmermann. Fuzzy Sets, Decision Making, and Expert Systems,
Kluwer-Nijho Publishing, Boston (1987).
Received: May, 2011