You are on page 1of 13

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/325865472

Fuzzy nonlinear programming problem for inequality constraints with alpha


optimal solution in terms of trapezoidal membership functions

Article  in  International Journal of Pure and Applied Mathematics · January 2018

CITATIONS READS

2 259

2 authors, including:

Kaliyaperumal Palanivel
Vellore Institute of Technology (VIT) Vellore
19 PUBLICATIONS   133 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Fuzzy Optimization Techniques View project

LINEAR PROGRAMMING View project

All content following this page was uploaded by Kaliyaperumal Palanivel on 26 July 2018.

The user has requested enhancement of the downloaded file.


International Journal of Pure and Applied Mathematics
Volume 119 No. 9 2018, 53-63
ISSN: 1311-8080 (printed version); ISSN: 1314-3395 (on-line version)
url: http://www.ijpam.eu
Special Issue
ijpam.eu

Fuzzy nonlinear programming problem


for inequality constraints with alpha
optimal solution in terms of trapezoidal
membership functions
R.Saranya#1 and Palanivel Kaliyaperumal∗2
#
Research Scholar
1,2
School of Advanced sciences,Department of Mathematics
VIT University, Vellore, India

drkpalanivel@gmail.com

Corresponding author
October 16, 2017

Abstract
In general, nonlinear programming presents much greater
difficulties than linear programming. Even the case when
all the constraints are linear and only the objective function
is nonlinear is often complicated. Here we have introduced
a new approach for solving nonlinear Programming Prob-
lem in terms of fuzziness using trapezoidal fuzzy member-
ship functions and its arithmetic operations and also stated
Kuhn Tuckers necessary and sufficient condition in terms
of fuzziness for finding alpha optimal solution of the prob-
lem. As a final point the alpha optimal solution with in
the nature of fuzzy numbers and also the same optimal so-
lution justified within the description of trapezoidal fuzzy
membership functions.
AMS Subject Classification:03B52,90C70,90C30
Key Words and Phrases:Fuzzy nonlinear program-
ming problem − Fuzzy nonlinear constraints − Kuhn Tuck-
ers conditions − Trapezoidal fuzzy numbers − Arithmetic
operations.

53
International Journal of Pure and Applied Mathematics Special Issue

1 Introduction
Mostly, the mathematical programming problem deals with opti-
mum use or allocation of limited resource to meet a desired objec-
tive. The fuzzy nonlinear programming problem is useful in solving
problems which are difficult, impossible to solve due to the im-
precise, subjective nature of the problem formulation or have an
accurate solution. Here we have an objective function which we
need to optimize under certain constraints. The concept of fuzzy
decision making was introduced by [1] and the maximum decision
[10] that is used in NLPP to find the optimal solution. A solution
that satisfies both, the constraints of the problem as well as the
objective function is termed as an optimal solution [2]. When the
objective function or the constraints are nonlinear we call it nonlin-
ear programming problem [7].Furthermore, this problem has fuzzy
objective function and fuzzy variables in the constraints [8], [6]and
[4] where the fuzzy left and right hand side coefficients on con-
straints [9]. In this article to solve such a nonlinear programming
problem, Kuhn Tuckers conditions are applied in terms of fuzzi-
ness to find an optimal solution. We have used trapezoidal fuzzy
membership function and its arithmetic operations and provided a
fuzzy approach to the generalized nonlinear programming problem
and also justified its solution with in the description of trapezoidal
fuzzy membership functions [3], [5].These kind of solving fuzzy op-
timization problem and its membership justification shows that the
novelty of our proposed methodology.

2 The Non Linear Programming Prob-


lem in Fuzzified Form
Fuzzy nonlinear programming problem is defined as the fuzzified
forms of nonlinear programmingproblem h which are stated
i below 
(1) (2) (3) (4)
 (1) (2) (3) (4)  x1 , x1 , x1 , x1 ,
Maximize z ,z ,z ,z = f  h (1) (2) (3) (4) i h (1) (2) (3) (4) i 
x2 , x2 , x2 , x2 ... xn , xn , xn , xn
Undertheh constraints i 
(1) (2) (3) (4) h i
x1 , x1 , x1 , x1 , (1) (2) (3) (4)
g i  h (1) (2) (3) (4) i h (1) (2) (3) (4) i  ≤ bi , bi , bi , bi (2.1)
x2 , x2 , x2 , x2 ... xn , xn , xn , xn

54
International Journal of Pure and Applied Mathematics Special Issue

0
where g i s are ‘m’ real valued functions of ‘n’ fuzzy variables and
0
bi s are ‘m’ fuzzy constants, and
h i
(1) (2) (3) (4)
xj , xj , xj , xj ≥ 0, i = 1, 2, 3, . . . , m; j = 1, 2, 3, . . . , n & m < n. (2.2)

Moreover, 0 (=) [−1 − δ, −δ, δ, 1 + δ] , where δ is a small positive


number.A feasible solution to the fuzzified nonlinear programming
problem is a fuzzy vector
 which
h satisfies conditions
i (2.1) and (2.2). 
(1) (2) (3) (4)
 (1) (2) (3) (4)  x 1 , x1 , x1 , x1 ,
X ,X ,X ,X =  h (1) (2) (3) (4) i h (1) (2) (3) (4) i 
x2 , x2 , x2 , x2 ... xn , xn , xn , xn

3 Computational Procedure
Kuhn Tuckers conditions form maximization can  be stated as
P ∂h(i)
∂f
∂xj
− λi ∂xj = 0  


i=1 
(i)
λi h (xj ) = 0 (3.1)


h(i) (xj ) ≤ 0 


λi ≥ 0
Kuhn Tuckers condition in fuzzified form may be expressed as 
h i Pm h ih i
(1) (2) (3)
fj , fj , fj , fj
(4)

(1) (2) (3)
λi , λi , λi , λi
(4) i(1) i(2) i(3)
hj , hj , hj , hj
i(4)
=0 



h i h i=1 i 

(1) (2) (3) (4) i(1) i(2) i(3) i(4) 

λ , λi , λi , λi h , hj , hj , hj = 0
h i ij (3.2)
i(1) i(2) i(3)
h , hj , hj , hj
i(4)
≤ 0 



h j i 

(1) (2) (3)
λi , λi , λi , λi
(4)
≥ 0 

We now compute the fuzzy membership function (f.m.f) of the
above conditions.Let
h i h i h i
(1) (2) (3) (4) (1) (2) (3) (4) i(1) i(2) i(3) i(4)
A = fj , fj , fj , fj ; B = λi , λi , λi , λi ; C = hj , hj , hj , hj (3.3)
Here, fuzzy membership function for each condition A, B and C
which are defined and then evaluate the fuzzy membership func-
tion of B(.)C. Next compute the α level of confidence interval for
the membership function of B and C. Then we proceed for the
computation of the fuzzy membership function of B(.)C. Hence the
membership function of {Bα (·)Cα } is

55
International Journal of Pure and Applied Mathematics Special Issue

   2  12

   (1) i(2) (2) i(1) (1) i(1)

 (1) i(2) (2) i(1)  λi hj + λi hj − 2λi hj − 

  λi hj + λi hj 
 



 − ± (2) i(2) (1) i(1) (2) i(1) (1) i(1)
4λi hj λi hj + 4λi hj λi hj + 


(1) i(1)
−2λi hj 




 (1) i(2) (1) i(1) (1) i(1) (1) i(1)

 4λi hj λi hj − 4λi hj λi hj

  

 (2) i(2) (2) i(1) (1) i(2)
2 λi hj −λi hj −λi hj +λi hj
(1) i(1)

   2  1
2
µλi hj (x) =   (3) i(4) (4) i(3) (4) i(4)
 (3) i(4) (4) i(3)  λi h j + λ i h j − 2λ i hj − 
 λ hj + λi hj  
 

 − i  ±

 (3) i(3) (4) i(4) (3) i(4) (4) i(4) 


 (4) i(4)
−2λi hj  4λi hj λi hj + 4λi hj λi hj + 

 
(4) i(3) (4) i(4) (4) i(4) (4) i(4)


 4λi hj λi hj − 4λi hj λi hj

  

 (3) i(3)
2 λi hj
(3) i(4)
−λi hj
(4) i(3)
−λi hj
(4) i(4)
+λi hj






0 otherwise
Next evaluate the fuzzy membership function of the Kuhn Tuckers
first condition is as follows;
∂f P
m
∂hi 0 P
m
0
∂xj
− λi ∂xj
= 0 ⇒ f − λi hi = 0
i=1 i=1
The fuzzified form of the  abovehcondition becomes,
P (1) (2) (3) (4) i
 0 (1) 0 (2) 0 (3) 0 (4)   λ , λi , λi , λi
f ,f ,f ,f − h 0i 0 (2) 0 (3) 0 (4)
i =0
 (1)
hi , hi , hi , hi

 0 (1) 0 (2) 0 (3) 0 (4)  h i


(1) (2) (3) (4)
Let A = f ,f ,f ,f , B = λi , λi , λi , λi ,
h 0 0 (2) 0 (3) 0 (4)
i
(1)
C = hi , hi , hi , hi
Again we need to state the fuzzy membership function for each
condition A, B and C in the similar manner of equation (3.3), now
proceeds to finding for {Aα (−)Bα (·)Cα } . Hence,

56
International Journal of Pure and Applied Mathematics Special Issue


   0  n  0 o2  0  1


0 0
− 2B1 − f (2) −f (1) ± 2B1 − f (2) −f (1) −8A1 f (1) −C1
2

 ,

 4A1

 0 (1) (1) 0 (1) 0 (2) (2) 0 (2)

 for f − λi hi ≤x≤f − λi hi





 0 (2) 0 (3)

 1, for f
0 (2)
− λi hi
(2)
≤x≤f
0 (3) (3)
− λi hi
µA(−)B(.)C (x) =

  n  12


  0 0
 0 0
o2  0

 − 2B2 − f (3) −f (4) ± 2B2 − f (3) −f (4) −8A2 f (4) −C2

 ,

 4A2

 for f
0 (3) (3)
− λi hi
0 (3)
≤x≤f
0 (4) (4) 0 (4)
− λi hi







0, otherwise

Where,

(2) 0 (2) (3) i(3) (2) 0 (2) (3) i(4) (2) 0 (2) (4) i(3) (2) 0 (2) (4) i(4)
A1 = λi hi λi hj − λi hi λi hj − λi hi λi hj + λi hi λi hj −
(2) 0 (1) (3) i(3) (2) 0 (1) (3) i(4) (2) 0 (1) (4) i(3) (2) 0 (1) (4) i(4)
λi hi λi hj + λi hi λi hj + λi hi λi hj − λi hi λi hj −
0
(1) (2) (3) i(3) 0
(1) (2) (3) i(4) 0
(1) (2) (4) i(3) (1) 0 (2) (4) i(4)
λi hi λi hj + λi hi λi hj + λi hi λi hj − λi hi λi hj +
(1) 0 (1) (3) i(3) (1) 0 (1) (3) i(4) (1) 0 (1) (4) i(3) (1) 0 (1) (4) i(4)
λi hi λi hj − λi hi λi hj − λi hi λi hj + λi hi λi hj

(2) 0 (1) (3) i(4) (2) 0 (1) (4) i(4) (2) 0 (1) (4) i(3) (2) 0 (1) (4) i(4)
B1 = λi hi λi hj − λi hi λi hj + λi hi λi hj − λi hi λi hj −
(1) 0 (1) (3) i(4) (1) 0 (1) (4) i(4) (1) 0 (1) (4) i(3) (1) 0 (1) (4) i(4)
λi hi λi hj + λi hi λi hj − λi hi λi hj + λi hi λi hj +
0
(1) (2) (3) i(4) 0
(1) (2) (4) i(4) 0
(1) (2) (4) i(3) (1) 0 (2) (4) i(4)
λi hi λi hj − λi hi λi hj + λi hi λi hj − λi hi λi hj −
(1) 0 (1) (3) i(4) (1) 0 (1) (4) i(4) (1) 0 (1) (4) i(3) (1) 0 (1) (4) i(4)
λi hj λi hj + λi hj λi hj − λi hj λi hj + λi hj λi hj .

(1) 0 (1) (4) i(4)


C1 = λi hi λi hj

and

(2) i(2) (3) 0 (3) (2) i(2) (3) 0 (4) (2) i(2) (4) 0 (3) (2) i(2) (4) 0 (4)
A2 = λi hj λi hi − λi hj λi hi − λi hj λi hi + λi hj λi hi −
(2) i(1) (3) 0 (3) (2) i(1) (3) 0 (4) (2) i(1) (4) 0 (3) (2) i(1) (4) 0 (4)
λi hj λi hi + λi hj λi hi + λi hj λi hi − λi hj λi hi −
0
(1) i(2) (3) (3) 0
(1) i(2) (3) (4) 0
(1) i(2) (4) (3) (1) i(2) (4) 0 (4)
λi hj λi hi + λi hj λi hi + λi hj λi hi − λi hj λi hi +
(1) i(1) (3) 0 (3) (1) i(1) (3) 0 (4) (1) i(1) (4) 0 (3) (1) i(1) (4) 0 (4)
λi hj λi hi − λi hj λi hi − λi hj λi hi + λi hj λi hi .

57
International Journal of Pure and Applied Mathematics Special Issue

(2) i(1) (3) 0 (4) (2) i(1) (4) 0 (4) (2) i(1) (4) 0 (3) (2) i(1) (4) 0 (4)
B2 = λi hj λi hi − λi hj λi hi + λi hj λi hi + λi hj λi hi −
(1) i(1) (3) 0 (4) (1) i(1) (4) 0 (4) (1) i(1) (4) 0 (3) (1) i(1) (4) 0 (4)
λi hj λi hi + λi hj λi hi − λi hj λi hi + λi hj λi hi +
(1) i(2) (3) 0 (4) (1) i(2) (4) 0 (4) (1) i(2) (4) 0 (3) (1) i(2) (4) 0 (4)
λi hj λi hi − λi hj λi hi + λi hj λi hi − λi hj λi hi −
0
(1) i(1) (3) (4) 0
(1) i(1) (4) (4) 0
(1) i(1) (4) (3) (1) i(1) (4) 0 (4)
λi hj λi hi + λi hj λi hi − λi hj λi hi + λi hj λi hi .

(1) i(1) (4) 0 (4)


C2 = λi hj λi hi

Using the above Kuhn Tuckers necessary and sufficient conditions,


we can obtain an optimal solution to a fuzzified nonlinear program-
ming problem.

4 Numerical example
Let us consider a nonlinear programming problem using Kuhn Tuck-
ers conditions [7]:

MaximizeZ = 2x1 2 + 12x1 x2 − 7x2 2

Subject to the constraints

2x1 + 5x2 ≤ 98, x 1 , x2 ≥ 0

Here,
f (x) = 2x1 2 + 12x1 x2 − 7x2 2 ,
h (x) = 2x1 + 5x2 − 98
The fuzzified form of the above nonlinear programming problem in
terms of trapezoidal fuzzy numbers is given below:

Maximize
[0, 1, 3, 4] (.) [x1 2, x1 − 1, x1 + 1, x1 + 2] (.) [x1 2, x1 − 1, x1 + 1, x1 + 2] (+) [10, 11, 13, 14]
(.) [x1 2, x1 − 1, x1 + 1, x1 + 2] (.) [x2 2, x2 − 1, x2 + 1, x2 + 2] (−) [5, 6, 8, 9]
(.) [x2 2, x2 − 1, x2 + 1, x2 + 2] (.) [x2 2, x2 − 1, x2 + 1, x2 + 2]
Subject to the constraints

[0, 1, 3, 4] (.) [x1 − −2, x1 − 1, x1 + 1, x1 + 2] (+)
≤ [96, 97, 99, 100]
[3, 4, 6, 7] (.) [x2 − −2, x2 − 1, x2 + 1, x2 + 2]

58
International Journal of Pure and Applied Mathematics Special Issue

Now the fuzzified form of the Kuhn Tuckers necessary condition


for maximizing the above nonlinear programming problem is
[0, 2, 6, 8] (.) [x1 ] (+) [10, 11, 13, 14] (.) [x2 ] (−) [0, 1, 3, 4] (.) [λ] = 0 (4.1)
[10, 11, 13, 14] (.) [x1 ] (−) [10, 12, 16, 18] (.) [x2 ] (−) [3, 4, 6, 7] (.) [λ] = 0 (4.2)
[0, 1, 3, 4] (.) [x1 ] (+) [3, 4, 6, 7] (.) [x2 ] (−) [96, 97, 99, 100] = 0 (4.3)
As per the computational procedure, the fuzzy membership func-
tion for the equation (4.1) which described as follows,
 1
 2 2 2
 (−x1 −x2 +λ−8±(x1 +x2 +λ −4x1 −24x2 −4λ+152) 2 )

 2







 f or 10x2 ≤ x ≤ 2x1 + 11x2 − λ






 1 , f or 2x1 + 11x2 − λ ≤ x ≤ 6x1 + 13x2 − 3λ
µA(.)B+C(.)D−E(.)F (x) =

 1

 (x1 +x2 −λ+16±(x21 +x22 +λ2 −4x1 −24x2 −4λ+152) 2 )

 2







 f or 6x1 + 13x2 − 3λ ≤ x ≤ 8x1 + 14x2 − 4λ





0, otherwise
Similarly, we have attained fuzzy membership function for the above
Kuhn Tucker condition of equation (4.2) and (4.3). Furthermore,
solve the above fuzzy membership function for all the Kuhn Tucker
conditions. We get the required fuzzified optimal solution of the
above nonlinear programming problem, which are;
(1) (2) (3) (4)
[x1 , x1 , x1 , x1 ] = [41, 42.5, 45.5, 47],
(1) (2) (3) (4)
[x2 , x2 , x2 , x2 ] = [−196, −97, 101, 200]
[λ(1) , λ(2) , λ(3) , λ(4) ] = [88, 94, 106, 112]
Since λ ≥ 0and hence the value of the objective function is
[Z(1) ,Z(2) ,Z(3) ,Z(4) ] = [844, 2872, 6928, 8956]

4.1 Comparison between the solutions obtained


from existing and proposed methods
To compare the existing and the proposed methodologies, the re-
sults of fuzzy nonlinear programming problem chosen in the above
numerical example obtained by using the existing and the proposed
methodologies are shown in the following table.

59
International Journal of Pure and Applied Mathematics Special Issue

Existing methodology based Proposed methodology based


on the classical methodology on the fuzzified nonlinear
[7]. programming problem [5].
The optimal solution is The fuzzy optimal solution is
MaxZ=4900. MaxZ = [844, 2872, 6928, 8956]
=R[844, 2872, 6928, 8956],
MaxZ= 4900.

On the basis of above result, same results are obtained in the


numerical example; it can be recommended that it is ideal to expend
the proposed methodology of existing methodology.

4.2 Results and discussion


Using the proposed methodology shows the overall fuzzified nonlin-
ear programming problem value is [844, 2872, 6928, 8956], which
may be physically understood as follows:
(i) According to decision maker the overall fuzzified nonlinear
programming problem values are going to be larger than 844 and
fewer than 8956.
(ii) Decision maker in favour of that the overall fuzzified non-
linear programming problem values are going to larger than or ad-
equate to 2872and fewer than or equal to 6928.
(iii) The proportion of the favourness of the decision maker for
the remaining values of overall fuzzified nonlinear programming
problem value is often obtained as follows:
The optimal solution of a fuzzified nonlinear programming prob-
lem values are going to be continuously bigger than 844 and fewer
than 8956 and most likelihood is that the values are going to be
between 2872 and 6928. The variations in cost with relevancy like-
lihood are shown in the following Fig.1.
Let x represent the value of overall fuzzified nonlinear program-
ming problem values and then the proportion of the favourness in
the decision maker for µmax (X) , where
 x−844

 for 844 ≤ x ≤ 2872
 2872−844
1 for 2872 ≤ x ≤ 6928
µmax (X) = x−8956

 for 6928 ≤ x ≤ 8956
 6928−8956
0 otherwise

60
International Journal of Pure and Applied Mathematics Special Issue

Figure 1: Trapezoidal membership function of the fuzzy optimum


solution µz (X)

5 Conclusion
Finally, the proposed solving procedure entirely considered about
a fuzzy nonlinear programming problem with the support of trape-
zoidal membership functions and its arithmetic operations [3,5].
The fuzzified version of the problem has been discussed with the
support of a numerical example using fuzzified form of Kuhn Tuck-
ers conditions and it is associated to showing that the proposed
procedure offers an efficient tool for handling nonlinear program-
ming problem instead of classical procedures. As a final point the
optimal solution with in the form of fuzzy numbers and justified its
solution with in the description of trapezoidal fuzzy membership
functions is explained with the result and discussion and also the
comparison of evaluations which might be a brand new attempt in
solving nonlinear programming problem in fuzzy nature. An equiv-
alent approach for solving the fuzzy issues may be utilized in future
studies of optimization techniques.

References
[1] R.E.Bellman, L.A.Zadeh, Decision-making in a
fuzzy environment, Management Science,(1970), DOI
10.1287/mnsc.17.4.B141

61
International Journal of Pure and Applied Mathematics Special Issue

[2] M. Das, H.K. Baruah, Solution of a linear programming


problem with fuzzy data, The journal of Fuzzy Mathematics,
(2004),12:793-811.

[3] Kaliyaperumal Palanivel, Fuzzy dynamic programming prob-


lem for single additive constraint with additively separable re-
turn by means of trapezoidal membership functions, Handbook
of Research on Fuzzy and Rough Set Theory in Organizational
Decision Making, (2016), 168-187.

[4] B.Kheirfam, F.Hasani, Sensitivity analysis for fuzzy linear


Programming problems with Fuzzy variables, Advanced Model
and Optimization, (2010), 12:257-272.

[5] K.Palanivel, Fuzzy commercial traveler problem of trapezoidal


membership functions within the sort of α optimum solu-
tion using ranking technique, Afrika Mathematika, Springer,
(2016), 27:263-277. DOI: 10.1007/s13370-015-0331

[6] V.Pandian, R.Nagarajan, Fuzzy linear programming:a modern


tool for decision making. Journal of Teknologi,(2002), 37:31-44.

[7] K.Swarup, P.K.Gupta,Manmohan, Operation research. Sultan


Chand and sons education publishers,India, (2004)

[8] H.C.Wu, Duality theory in fuzzy optimization problems, Fuzzy


Optimization and Decision Mathematics, (2004), 3:.345-365.

[9] M.Yenilmez, N.Rafail, K.Gasimor, Solving fuzzy liner pro-


gramming problems with linear membership function, Turk H,
(2002), Math 26:375-396.

[10] H.J.Zimmermann, Fuzzy Sets, Decision Making, and Expert


Systems, Kluwer-Nijhoff Publishing, Boston.

10

62
63
64

View publication stats

You might also like