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

net/publication/315753220

Bi-level Multi-objective Programming Problems with Fuzzy Parameters:

Modified TOPSIS Approach

Article · January 2016

DOI: 10.11648/j.ijmfs.20160205.11

CITATIONS READS

6 109

2 authors:

Ibrahim A. Baky Mohamed A. El Sayed

Benha University, Faculty of Engineering Benha University
15 PUBLICATIONS   534 CITATIONS    27 PUBLICATIONS   249 CITATIONS   

SEE PROFILE SEE PROFILE

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

Multi-level Multi-objective Fractional Programming Problem View project

stability analysis of multi-objective problems View project

All content following this page was uploaded by Mohamed A. El Sayed on 03 April 2017.

The user has requested enhancement of the downloaded file.

International Journal of Management and Fuzzy Systems
2016; 2(5): 38-50
http://www.sciencepublishinggroup.com/j/ijmfs
doi: 10.11648/j.ijmfs.20160205.11

Bi-level Multi-objective Programming Problems with Fuzzy

Parameters: Modified TOPSIS Approach
Ibrahim A. Baky1, 2, M. A. El Sayed2, *
1
Department of Mathematics, Faculty of Science, Tabuk University, Tabuk, Saudi Arabia
2
Department of Basic Engineering Sciences, Faculty of Engineering, Benha University, El Qalyoubia, Egypt

Email address:
Ibrahimbaky@yahoo.com (I. A. Baky), mohamedali20008@yahoo.com (M. A. El Sayed)
*
Corresponding author

To cite this article:

Ibrahim A. Baky, M. A. El Sayed. Bi-level Multi-objective Programming Problems with Fuzzy Parameters: Modified TOPSIS Approach.
International Journal of Management and Fuzzy Systems. Vol. 2, No. 5, 2016, pp. 38-x50. doi: 10.11648/j.ijmfs.20160205.11

Received: October 5, 2016; Accepted: November 30, 2016; Published: December 27, 2016

Abstract: In this paper, a modified TOPSIS (techniques for order preference by similarity to ideal solution) approach for
solving bi-level multi-objective programming (BL-MOP) problems with fuzzy parameters is presented. These fuzzy
parameters are assumed to be characterized by fuzzy numerical data, reflecting the experts' imprecise or fuzzy understanding
of the nature of the parameters in the problem formulation process. Firstly, the corresponding non-fuzzy bi-level programming
model is introduced based on the -level set. Secondly, a modified TOPSIS approach is developed, in which the fuzzy goal
programming (FGP) approach is used to solve the conflicting bi-objective distance functions instead of max-min operator. As
the FGP approach utilized to achieve the highest degree of each membership goal by minimizing the sum of the unwanted
deviational variables. Finally, an algorithm to clarify the modified TOPSIS approach, as well as Illustrative numerical example
and comparison with the existing methods, are presented.

Keywords: Bi-level Programming, Fuzzy Sets, Fuzzy Parameters, TOPSIS, Fuzzy Goal Programming,
Multi-objective Programming

farthest distance from the negative ideal solution (NIS). It

1. Introduction
Bi-level mathematical programming (BLMP) is defined as
mathematical programming that solves decentralized
planning problems with two decision makers (DMs) in two-
levels or hierarchical organization [5, 15, 17]. The basic
concept of BLMP is that the upper-level decision maker
(ULDM) (the leader) sets his goals and/or decisions and then
asks each subordinate level of the organization for their
optima which are calculated in isolation; the lower-level
decision maker (LLDM) (the follower) decisions are then
submitted and modified by the ULDM with the consideration
of the overall benefit of the organization; the process is
continued until a satisfactory solution is reached [6, 7].
Techniques for order performance by similarity to ideal
solution (TOPSIS), one of the known classical multiple
criteria decision-making (MCDM) method, bases upon the
concept that the chosen alternative should have the shortest
distance from the positive ideal solution (PIS) and the
MSC2010: 90C05; 90C29; 90C70.
was first developed by Hwang and Yoon [11] for solving a
multiple attribute decision-making problem. A similar
concept has also been pointed out by Zeleny [24]. Lai et al.
[12] extended the concept of TOPSIS to develop a
methodology for solving multiple objective decision-making
(MODM) problems. Also, Abo-Sinna [2] proposed TOPSIS
approach to solve multi-objective dynamic programming
(MODP) problems. As Abo-Sinna showed that using the
fuzzy max–min operator with non-linear membership
functions, the obtained solutions are always non-dominated
by the original MODP problems.
Further extensions of TOPSIS for large scale multi-
objective non-linear programming problems with block
angular structure was presented by Abo-Sinna et al. in [1, 3].
Deng et al. [10] formulated the inter-company comparison
process as a multi-criteria analysis model, and presented an
effective approach by modifying TOPSIS for solving such a
problem. Chen [9] extended the concept of TOPSIS to
develop a methodology for solving multi-person multi-
39 Ibrahim A. Baky and M. A. El Sayed: Bi-level Multi-objective Programming Problems with Fuzzy
Parameters: Modified TOPSIS Approach
criteria decision-making problems in a fuzzy environment.
Recently, Baky and Abo-Sinna [4] extended the TOPSIS
approach for solving bi-level multi-objective decision-
making problem. Baky et al. proposed the TOPSIS approach
for bi-level multi-objective programming problems with
fuzzy parameters [8].
Generally, TOPSIS provides a broader principle of
compromise for solving multiple criteria decision-making
problems. It transfers m-objectives, which are conflicting and
non-commensurable, into two objectives (the shortest
distance from the PIS and the longest distance from the NIS).
They are commensurable and most of time conflicting. Then,
the bi-objective problem can be solved by using membership
functions of fuzzy set theory to represent the satisfaction
level for both criteria and obtain TOPSIS’s compromise
solution. The max–min operator is then considered as a
suitable one to solve the confliction between the new criteria
[1, 3, 11].
It has been observed that, in most real-world situations, for
example, power markets and business management, the
possible values of the parameters are often only imprecisely
or ambiguously known to the experts and cannot be
described by precise values. With this observation, it would
be certainly more appropriate to interpret the experts
understanding of the parameters as fuzzy numerical data
which can be represented by means of fuzzy sets [7, 20, 23].
Therefore, this paper study the bi-level multi-objective
programming (BL-MOP) problems with fuzzy parameters.
Fuzzy goal programming (FGP) approach was viewed to
be an efficient methodology for solving multi-objective
programming problems by Mohamed [14]. Thereafter, the
concept of FGP approach was extended by Moitra et al. [22]
for obtaining a satisfactory solution for bi-level programming
problems. Recently, Baky extended FGP approach for
solving a decentralized BL-MOP problem [5] and MLP
problem [6]. Thus, we make use of the FGP approach to
solve the confliction between the bi-objective distance
functions instead of the conventional max-min decision
model.
This study presents a modified TOPSIS approach for
solving BL-MOP problem with fuzzy parameters. To
formulate the non-fuzzy model of the problem for a desired
value of , the uncertain model is converted into a
deterministic BL-MOP problem using the -level set of
fuzzy numbers. Then, a modified TOPSIS approach is used
to solve the deterministic model. Therefore, the confliction
between the bi-objective distance functions is solved by
using the FGP approach. Since, to overcome the
shortcomings of the classical approaches, a FGP model is
used to minimize the group regret of degree of satisfaction
for the bi-objective distance functions. In order to achieve the
highest degree (unity) of each of the defined membership
goal by minimizing their deviational variables and thereby
obtaining the most satisfactory solution for both of them.
Procedures for the developed modified TOPSIS approach are
also introduced.
The remainder of this paper is organized as follows.
Section 2 presents some preliminaries. Section 3 and section
4 briefly discuss problem formulation and crisp model
formulation of BL-MOP problem with fuzzy parameters. The
proposed modified TOPSIS approach and an algorithm for
solving BL-MOP problem with fuzzy parameters is
developed in section 5. The following section presents an
illustrative numerical example to demonstrate the proposed
modified TOPSIS approach. The concluding remarks are
made in section 7.
2. Preliminaries
This section presents some basic concepts of fuzzy set
theory and distance measures, for more details see [21, 23]
for fuzzy set theory and [1-3] for distance measures.
Fuzzy set à is a set of ordered pairs , à ∈ ,
Definition 2.1: Let R be the space of real numbers. A
where à : → [0,1] is called membership function of fuzzy
Definition 2.2: A convex fuzzy set, Ã , is a fuzzy set in
set.
which: ∀ x, y ∈ R, ∀ ∈ [0, 1] μà + 1− ≥
min μà , μà !.
fuzzy set which is defined as Ã= ( , μà ) where:
Definition 2.3: Triangular fuzzy number (TFN) is a convex
& a ≤ ≤ b
' ( –* +
$ ' , –* +
μà =
% ' 0 –, + b ≤ ≤ c
' 0 – (+
$
(1)
# 0 23ℎ56 7895
parameters (a, b, c) which are : ≤ ; ≤ < will be denoted
For convenience, TFN represented by three real
by the triangle a, b, c.
fuzzy set denoted by à = for which the degree of it's
Definition 2.4: The –level set of a fuzzy set à is a non-
α ∈ [0, 1], i.e. Ã = = ? | Ã ≥ A.
membership functions exceed or equal to a real number
The -level set of :B is then; :B= = [:B=C , :B=D ] that is :B=C =
1 − : + ;, and :B=D = 1 − < + ;, where, :B=C and :B=D
Consider the vector of objective functions E =
represent the lower and upper cuts respectively.
'FG , FH , … , FJ + to be minimized and ideal vector of
objective functions E ∗ = FG∗ , FH∗ , … , FJ∗
(ideal point-
reference point- positive ideal solution (PIS)) in the m-
solution of objective functions E L = FGL , FHL , … , FJL (anti-
objective space. And consider the vector of anti-ideal
Where FM∗ = N8OP FM and FML = N:
P FM , R = 1,2, … , N .
ideal point – nadir point – negative ideal solution (NIS)).
(∈Q (∈Q
"closeness", TU -metric is used. The TU -metric defines the
And G is a convex constraints feasible set. As the measure of
and E ∗ . If the objective
functions FM , R = 1,2, … , N , are not expressed in
distance between two points E
commensurable units, then a scaling function for every
objective function, usually, this dimensionless is the interval
[0, 1]. In this case, the following metric could be used:
 International Journal of Management and Fuzzy Systems 2016; 2(5): 38-50 40

∗ U à = G , H ∈ e be partitioned between the two decision

VU = W∑J
M^G 7M Y Z ∗ LZ \ ]
U Z[ LZ[ (
_ , b = 1,2, … , ∞. makers. The ULDM has control over the vector G ∈
[ [
e` and the LLDM has control over the vector H ∈ ef ,
(2)

where 7M , R = 1,2, … , N, are where O = OG + OH. Furthermore, assume that

Egh G, H : × → , 8 = 1,2
the relative importance
(weights) of objectives. e` ef Jj
(3)

3. Problem Formulation are the upper-level and the lower-level vector objective
functions, respectively. So the BL-MOP problem with fuzzy
Assume that there are two levels in a hierarchy structure parameters of minimization-type may be formulated as
with ULDM and LLDM. Let the vector of decision variables follows [4-8, 15]:
[1st Level]

P EgG
N8O G, H P FkGG
= N8O G, H , FkGH G, H , … , FkGJ` G, H , (4)
(` (`

7ℎ565 H 92lm59
[2nd Level]

P EgH
N8O G, H P FkHG
= N8O G, H , FkHH G, H , … , FkHJf G, H , (5)
(f (f

9n;R5<3 32

pkG + pkH ;g, ≥ 0, ;g ∈

q
∈o= = G, H ∈ e
G H
^
r
J
≠ t, (6)

where

FkhM = <̃GG + <̃GH + ⋯ + <̃Ge` + <̃HG + <̃HH + ⋯ + <̃Hef Hef ,

hM hM hM hM hM hM
GG GH Ge` HG HH (7)

and where Nh , 8 = 1,2 are the number of DMh 's objective

functions, N is the number of constraints, <̃y =
hM

<̃yG , <̃yH , … , <̃ye , { = 1,2 and <̃ye are constants, pkh are the
4. Crisp Model Formulation of BL-MOP
hM hM hM hM
z z
coefficient matrices of size N × Oh , 8 = 1,2 ., the control
Problem with Fuzzy Parameters
variables G = ' GG , GH , … , G ` + and H = ' HG , HH , … , H f + ,
e e The individual optimal solution of ULDM and LLDM
objective functions would be considered when scaling every
and G is the bi-level convex constraints feasible choice set. objective function. Then for a prescribed value of ,

FkhM 8 = 1,2 , R = 1,2, … , Nh can be replaced by the

minimization-type objective function [7, 16, 19, 20],

lower bound of its -level.

FkhM
C
hM C hM C
= '<̃G += G + '<̃H += H 8 = 1,2 , R = 1,2, … , Nh
=
(8)

Similarly, maximization-type objective function, FkhM 8 = 1,2 , R = 1,2, … , Nh can be replaced by the upper bound of its
-level (α-cut)

FkhM
D
hM D hM D
= '<̃G += G + '<̃H += H 8 = 1,2 , R = 1,2, … , Nh
=
(9)

∑eM^G'pkhM + ≤ ';gh += 8 = 6G + 1, … , 6H
C D
= M
The inequality constraints,

∑eM^G pkhM M ≥ ;gh 8 = 1,2, … , 6G

(13)
(10)

∑eM^G pkhM M ≤ ;gh 8 = 6G + 1, … , 6H

For equality constraints;
(11) ∑eM^G pkhM M = ;gh 8 = 6H + 1, … , N (14)
Can be rewritten by the following constraints as [7, 13, 19, Can be replaced by two equivalent constraints;

∑eM^G'pkhM + ≤ ';gh += 8 = 6H + 1, … , N
C D
23]:

∑eM^G'pkhM + M ≥ ';gh += 8 = 1,2, … , 6G

D C = M
(15)
=
∑eM^G'pkhM + ≥ ';gh += 8 = 6H + 1, … , N
(12)
D C
= M (16)
41 Ibrahim A. Baky and M. A. El Sayed: Bi-level Multi-objective Programming Problems with Fuzzy
Parameters: Modified TOPSIS Approach

For proof of equivalency of the above equations (14) and value of , the minimization-type -(BL-MOP) problem
(15)&(16), see Lee and Li [13]. Therefore, for a prescribed reduced to the following problem:
[1st Level]

P EgG P | FkGG , FkGH , … , FkGJ`

C C C C
N8O = N8O },
= = = =
(17)
(` (`

7ℎ565 H 92lm59
[2nd Level]

P EgH P | FkHG , FkHH , … , FkHJf

C C C C
N8O = N8O },
= = = =
(18)
(f (f

9n;R5<3 32

∑eM^G'pkhM + ≥ ';gh += 8 = 1,2, … , 6G , 6H + 1, … , N

D C
M
∈ o~ = • ∈ H
€ =
•
∑eM^G'pkhM + ≤ ';gh += 8 = 6G + 1, … , 6H , 6H + 1, … , N ≥ 0,
C D (19)
= M M

where, the set of constraints in (19) denoted by o~ .
5. The Modified TOPSIS Approach for
BL-MOP Problem with Fuzzy
Parameters
In this section, a modified TOPSIS approach is presented
in which a FGP approach, for more details see [5-7], is
considered for solving the bi-criteria of the shortest distance
membership values to the highest degree (unity) is not
possible due to conflicting objectives. Therefore, each DM
should try to maximize his or her membership function by
making them as close as possible to unity by minimizing
their deviational variables.
5.1. The Modified TOPSIS Model for the Upper-level MOP
Problem
Consider the upper-level of minimization type problem of
the -(BL-MOP) problem (17):

P | FkGG , FkGH , … , FkGJ`

C C C
N8O }
from the PIS and the farthest distance from the NIS instead
of the conventional max-min model. The FGP approach,
= = =
(20)
(`
developed by Mohamed [14], were recently extended to

9n;R5<3 32
solve BL-MOP problem with fuzzy demands by Baky et al
[7]. Since in decision-making situation, the aim of each DM
is to achieve highest membership value (unity) of the
associated fuzzy goal in order to obtain the satisfactory
solution. However, in real situation, achievement of all

∑eM^G'pkhM + ≥ ';gh += 8 = 1,2, … , 6G , 6H + 1, … , N

D C
M
∈ o~ = • ∈ H
€ =
•
∑eM^G'pkhM + ≤ ';gh += 8 = 6G + 1, … , 6H , 6H + 1, … , N ≥ 0,
C D (21)
= M M

Thus, the TOPSIS approach of Lai et al. [12] that solves Therefore, the bi-objective distance functions for the upper-
single-level MODM problems is considered, in this paper, to level MOP problem follows as [1-4, 12]:

N8O V‚‚ƒ„
…
solve the upper-level MOP problem at a desired value of .
(22)

N: V‚†ƒ„
…
(23)
9n;R5<3 32

∑eM^G'pkhM +
D
g C
M ≥ ';h += 8 = 1,2, … , 6G , 6H + 1, … , N
∈ o~ = • ∈ H
€ =
•
∑eM^G'pkhM + ≤ ';gh += 8 = 6G + 1, … , 6H , 6H + 1, … , N ≥ 0,
C D (24)
= M M

where
 International Journal of Management and Fuzzy Systems 2016; 2(5): 38-50 42

‰ U à
Zk`[ ( LZ`[
∗

V‚ = •∑M^G 7M ‡ Š •
‚ƒ„ … J` U ˆ
\ LZ ∗
Z`[
(25)
`[

‰ U à
Z \ L Zk`` (
V‚ " •∑M^G 7M ‡ Š •
†ƒ„ … J` U `[ ˆ
\
Z LZ ∗ (26)
`[ `[

P FkGM then FkGM " FkGM

C C
and where FGM∗ " N8O . Let E Œ " 'FGG , FGH , … , FGJ + and
∗ ∗∗ ∗
`
= =
, is the individual positive
E Œ " 'FGG , FGH , … , FGJ + . Assume that the membership
(∈Q‹ \ L L L
P FkGM
D
FGML " N:
`
ideal solutions,
=
, is the individual functions ( G and H
are linear between 'VUΠ+ and 'VUΠ+ which are:
(∈Q‹ ∗ L
of the two objective functions
negative ideal solutions and 7M , is the relative importance
(weights) of objectives. As the problem of minimization type
… ∗
' V‚‚ƒ„ + " N8O
P V‚‚ƒ„ and the solution is ,
… ‚
(27)
(∈Q‹

… ∗
' V‚†ƒ„ + " N:
P V‚†ƒ„ ,
… †
and the solution is (28)
(∈Q‹

… L … L
' V‚‚ƒ„ + " V‚‚ƒ„ and ' V‚†ƒ„ + " V‚†ƒ„ (29)
… † … ‚

≡
… L
Also, as proposed in [1-4, 12] that ' V‚‚ƒ„ + ••‘
Ž•
… H ’•‘
Ž•
…
… L … L
' V‚†ƒ„ + can be taken as ' V‚‚ƒ„ + " N:
P V‚‚ƒ„
and and can be obtained as (see:
…
and Fig. 1):
(∈Q‹
… L
' V‚†ƒ„ + " N8O
P V‚†ƒ„ ≡
…
, respectively. Thus G
(∈Q‹
… ∗
1 8F V‚‚ƒ„ “ ' V‚‚ƒ„ +
…
&
$ ••‘ … ••‘ … ∗

" 1
Ž• ( L Ž• … ∗ … L
8F ' V‚‚ƒ„ + . V‚‚ƒ„ . ' V‚‚ƒ„ +
…

%
G \
••‘… L Ž ••‘…
∗
Ž•
(30)
$
•
… L
# 0 8F ' V‚‚ƒ„ + “ V‚‚ƒ„
…

… ∗
1 8F V‚†ƒ„ ” ' V‚†ƒ„ +
…
&
$ ’•‘ … ∗ ’•‘ …
" 1 . ' V‚†ƒ„ +
Ž• L Ž• ( … L … ∗
8F ' V‚†ƒ„ + . V‚†ƒ„
…

%
H ’•‘ … ∗ ’•‘ … \
Ž• L Ž•
(31)
$ … L
# 0 8F V‚†ƒ„ “ ' V‚†ƒ„ +
…

ULDM problem having flexible membership goals with the

aspired level unity can be represented as [5-7, 15]:
••‘ … ••‘ … ∗
Ž• ( L Ž•
1 •G‚ƒ„ •G‚ƒ„ " 1
\ –
\ ∗
••‘…
Ž• ••‘…
L Ž•
(32)

’•‘ … ∗ ’•‘ …
Ž• L Ž• (
1 •H†ƒ„ •H†ƒ„ " 1
\ –
’•‘ … ∗ ’•‘ … \
Ž• L Ž•
(33)

where •G‚ƒ„ , •H†ƒ„ and •G‚ƒ„ , •H†ƒ„ represent the under-

\ \ – –

••‘
Fig. 1. The membership functions of Ž•
Appling the FGP approach developed by Baky et al. [7], to
solve the conflicting bi-objective distance functions, to
’•‘
and Ž• . and over-deviations from the aspired levels, respectively. The
FGP approach of Mohamed [14] that solves single-level
multi-objective linear programming problem is considered to
solve the ULDM problem as follows:

N8O — " 7G‚ƒ„ •G‚ƒ„ 7H†ƒ„ •H†ƒ„

achieve highest membership value (unity) of the associated – \
fuzzy goals and obtain the satisfactory solution. So the (34)
membership functions defined in (30) and (31) for the 9n;R5<3 32
43 Ibrahim A. Baky and M. A. El Sayed: Bi-level Multi-objective Programming Problems with Fuzzy
Parameters: Modified TOPSIS Approach

∑eM^G'pkhM + ';gh += 8 " 1,2, … , 6G , 6H 1, … , N (37)

••‘ … ••‘ … ∗ D C
Ž• ( L Ž•
1 •G‚ƒ„ •G‚ƒ„ " 1,
\ – M
\ ∗ =
••‘… L Ž ••‘…
Ž•
(35)
∑eM^G'pkhM + . ';gh += 8 " 6G
•
1, … , 6H , 6H 1, … , N (38)
C D

’•‘…
∗
’•‘… = M
Ž• L Ž• (
1 ∗
’•‘… L Ž ’•‘…
\ •H †ƒ„ \
•H †ƒ„ –
"1
Ž•
(36)
•

where the numerical weights 7G‚ƒ„ and 7H†ƒ„ represent Mohamed [14] can be used to assign the values of 7G‚ƒ„ and
7H†ƒ„ as follows:
the
relative importance of achieving the aspired levels of the
respective fuzzy goals subject to the constraints set in the
decision situation. The weighting scheme suggested by

7G‚ƒ„ " , :OV 7H†ƒ„ "

G G
••‘ … \ ••‘ … ∗ ’•‘ … ∗ ’•‘ … \
Ž• L Ž• Ž• L Ž•
(39)

Based on the basic concepts of the bi-level programming, give the lower-level decision maker an extent feasible region
the ULDM sets his goals and/or decisions with possible to search for the satisfactory solution [15, 18].

fuzzy set theory. According to this concept, let 3yC and 3ỹ , OG components of decision vector GŒ " ' GG , GH , … , Ge +
∗ Œ∗ Œ∗ Œ∗
tolerances which are described by membership functions of The linear membership functions (Fig. 2) for each of the

{ " 1,2, … , OG are the maximum acceptable negative and

`
controlled by the ULDM can be formulated as:

by the ULDM, GŒ " ' GG , GH , … , Ge + . The tolerances

∗ Œ∗ Œ∗ Œ∗
positive tolerance values on the decision vector considered
`

… ∗
L™z‰
&
(`z L (`z
, 8F Œ∗
3yC . . Gy
Œ∗
$ ™z‰ Gy Gy

" …
(`z
∗
š™z› L(`z
% , 8F Gy . Gy . Gy 3ỹ , { " 1,2, … , OG
(`z Gy Œ ∗ Œ ∗ (40)
$ ™z›
# 0, 234567895
It may be noted that, the decision maker may desire to shift the range of Gy . Following Pramanik and Roy [15] and Sinha
[18], this shift can be achieved.

Fig. 2. The membership function of the decision variable Gy .

positive ideal solution, V‚‚ƒ„ , and the distance function from

•

5.2. The Modified TOPSIS Model for œ-(BL-MOP)

the negative ideal solution, V‚†ƒ„ , are modified as both PIS
•

(E ∗ ) and NIS (E L ) obtained from the lower and upper bound

Problem

distance family. Thus, V‚‚ƒ„ and V‚†ƒ„ can be proposed in

In order to obtain a compromise solution (satisfactory models, respectively, of the problem to normalize the
• •
solution) to the -(BL-MOP) problem at a specified value of
using the modified TOPSIS approach, the distance family this paper, for the objective functions of the upper and lower
of equation (2) to represent the distance function from the levels as follows [4, 8]:
 International Journal of Management and Fuzzy Systems 2016; 2(5): 38-50 44

‰ U ‰ U à
Zk`[ ( LZ ∗ Zkf[ ( LZ ∗
V‚‚ƒ„ " •∑M^G
J`
7GM
U
‡ Š + ∑J
M^G 7HM
U
‡ ˆ f[
Š •
• ˆ `[ f
\ LZ ∗
Z`[ \ LZ ∗
Zf[
`[ f[
(41)

‰ U ‰ U à
\
Z`[ L Zk`[ ( \
Zf[ L Zkf[ (
V‚†ƒ„ = •∑M^G 7GM ‡
J` U
Š + ∑M^G 7HM ‡
Jf U
Š •
• ˆ ˆ
Z \ LZ ∗ `[ `[ Z \ LZ ∗ f[ f[
(42)

where 7y , { = 1,2, … , NG + NH are the relative importance

P FkhM
C
(weights) of objectives in both levels. FhM∗ = N8O
approach for solving -(BL-MOP) problem.
In order to obtain a compromise solutions, the problem
=
,
(∈Q‹
transferred into the following bi-objective problem with two

P FkhM
D
FhML = N: , 8 = 1,2, R = 1,2, … , Nh , b=
commensurable (but often conflicting) objectives [1-3,12]:

N8O V‚‚ƒ„
= •
and
(∈Q‹
1,2, … , ∞. Let E = ∗
'FGG
∗
, FGH
∗
, … , FGJ
∗
, F ∗ , F ∗ , … , FHJ
∗
+,
(43)
` HG HH
N: V‚†ƒ„
f •
the

E L = 'FGGL
, FGH
L
, … , FGJ
L
, F L , F L , … , FHJ
L
+ , the individual
individual positive ideal solutions for both levels, and (44)
` HG HH f
9n;R5<3 32
case of b = ∞ , see [1-3,12] for the general form of the
negative ideal solutions for both levels. Also, for the special

distance functions that can be applied to the proposed

∑eM^G'pkhM +
D
g C
M ≥ ';h += 8 = 1,2, … , 6G , 6H + 1, … , N
∈ o~ = • ∈ H
€ =
•
∑eM^G'pkhM + ≤ ';gh += 8 = 6G + 1, … , 6H , 6H + 1, … , N ≥ 0,
C D (45)
= M M

ž ≡ ••‘ • Ÿ ≡ ’•‘ •
Ž• Ž•
Since these two objectives are usually conflicting to each functions ( and ) of
the two objective functions are linear between 'VU + and
other, it is possible to simultaneously obtain their individual ∗

'VU + , they take the following form [4,8]:

optima. Thus, we can use membership functions to represent L
these individual optima. Assume that the membership
• ∗
V‚‚ƒ„ = N8O
P V‚‚ƒ„ and the solution is ,
• ‚ƒ„
(46)
(∈Q‹

• ∗
V‚†ƒ„ = N:
P V‚†ƒ„ ,
• †ƒ„
and the solution is (47)
(∈Q‹

• L • L
V‚‚ƒ„ = V‚‚ƒ„ , 26 V‚‚ƒ„ = N:
P V‚‚ƒ„
• †ƒ„ •
(48)
(∈Q‹

• L • L
V‚†ƒ„ = V‚†ƒ„ , 26 V‚†ƒ„ = N8O
P V‚†ƒ„
• ‚ƒ„ •
(49)
(∈Q‹

• ∗ • ∗
And also, assume that 'VU + = V‚‚ƒ„ , V‚†ƒ„
∗
≡ ’•‘• . Therefore, as shown in Fig. 1,
assign a larger degree to the one with farther distance from
• L • L
NIS for Ÿ
and 'VU + = V‚‚ƒ„, V‚†ƒ„
L Ž•
ž Ÿ
. Then, based on the
and can be obtained as follows [1-3, 12]:
≡ ••‘•
preference concept, we assign a larger degree to the one with
shorter distance from the PIS for ž
Ž•
and
• ∗
& 1 8F V‚‚ƒ„ < V‚‚ƒ„
•

$ ••‘• ••‘•
∗
• ∗ • L
= 1−
Ž• ( L| Ž• }
8F V‚‚ƒ„ ≤ V‚‚ƒ„ ≤ V‚‚ƒ„
•

%
ž \ ∗
••‘• L Ž ••‘•
Ž•
(50)
$
•
• L
0 8F V‚‚ƒ„ < V‚‚ƒ„
•
#
45 Ibrahim A. Baky and M. A. El Sayed: Bi-level Multi-objective Programming Problems with Fuzzy
Parameters: Modified TOPSIS Approach

• ∗
1 8F V‚†ƒ„ > V‚†ƒ„
•
&
$ ’•‘•
∗
’•‘•
| Ž• } L Ž• ( • L • ∗
" 1− 8F V‚†ƒ„ ≤ V‚†ƒ„ ≤ V‚†ƒ„
•

%
Ÿ • ∗ • \
’•‘
Ž• ’•‘
L Ž•
(51)
$ • L
0 8F V‚†ƒ„ < V‚†ƒ„
•
#

satisfactory solution of, ∗ = G∗ , H∗ , the final proposed vector, GŒ = ' GG , GH , … , Ge +, follows as:
∗ Œ∗ Œ∗ Œ∗
Finally, as discussed in section 5.1, in order to generate the function equation (40) for the upper-level decision variables
`

N8O — = 7ž‚ƒ„ •ž‚ƒ„ + 7Ÿ†ƒ„ •Ÿ†ƒ„ + ∑y^G

e`
[7Gy •Gy + •Gy + 7Gy •Gy + •Gy˜š ]
– \ C CL Cš ˜ ˜L
model of the modified TOPSIS that includes the membership
(52)

9n;R5<3 32
••‘ • ••‘ • ∗
Ž• ( L| Ž• }
1− + •ž‚ƒ„ − •ž‚ƒ„ = 1
\ –
••‘ • \ ••‘ • ∗
Ž• L Ž•
(53)

’•‘ • ∗
’•‘ •
| Ž• } L Ž• (
1− + •Ÿ†ƒ„ − •Ÿ†ƒ„ = 1
\ –
∗ \
’•‘• L Ž ’•‘•
Ž•
(54)
•

… ∗
(`z L (`z L™z‰
+ •Gy
CL
− •Gy
Cš
= 1, { = 1,2 … , OG .
™z‰
(55)
…∗
(`z š™z› L(`z
+ •Gy
˜L
− •Gy
˜š
= 1, { = 1,2 … , OG .
™z›
(56)

∑eM^G'pkhM + ≥ ';gh += 8 = 1,2, … , 6G , 6H + 1, … , N

D C
= M
(57)

∑eM^G'pkhM + ≤ ';gh += 8 = 6G + 1, … , 6H , 6H + 1, … , N ≥ 0, R = 1,2

C D
= M M (58)

•Gy
L
= •Gy
CL
, •Gy
˜L
, •Gy
š
= •Gy
Cš
, •Gy
˜š

•ž , •Ÿ , •Gy , •Gy , •ž , •Ÿ , •Gy , •Gy ≥ 0

represents the fuzzy achievement function. Again, to assess
‚ƒ„ \ †ƒ„ \ CL ˜L ‚ƒ„ – †ƒ„ – Cš ˜š
where and

•ž‚ƒ„ × •ž‚ƒ„ = 0, •Ÿ†ƒ„ × •Ÿ†ƒ„ = 0, •Gy × •Gy = 0 and

the relative importance of the fuzzy goals properly, the
\ – \ – CL Cš
with

to assign the values to 7ž‚ƒ„ , 7Ÿ†ƒ„ , 7Gy

C
and 7Gy
˜
weighting scheme suggested by Mohamed [14] can be used
•Gy × •Gy = 0, { = 1,2, … , OG , represent the under- and
˜L ˜š

over-deviation, respectively, from the aspired levels. Also —

as:

7ž‚ƒ„ = , :OV 7Ÿ†ƒ„ =

G G
\ ∗ ∗ • \
••‘•
Ž• ••‘•
L Ž• ’•‘•
Ž• ’•‘
L Ž•
(59)

7Gy = , :OV 7Gy = , { = 1,2 … , OG .

C G ˜ G
™z‰ ™z›
(60)

5.3. The Modified TOPSIS Algorithm for BL-MOP satisfactory decision for the two DMs at the two levels.
Problem with Fuzzy Parameters Following the above discussion, the algorithm for the
proposed modified TOPSIS approach for solving BL-MOP
The modified TOPSIS model (52)-(58) provides a problem with fuzzy parameters is given as follows:
Step 1. Formulate the deterministic model of the BL-MOP problem, for a prescribed value of .
Step 2. Determine the individual maximum and minimum values from the upper and lower bound of -level.

E Œ = 'FGG , FGH , … , FGJ +, the individual positive ideal solutions.

∗ ∗ ∗ ∗
Construct the PIS payoff table of the ULDM problem (20)&(21) from the lower bound model and obtain
Step 3.
`

E Œ = 'FGG , FGH , … , FGJ +, the individual negative ideal solutions.

\ L L L
Construct the NIS payoff table of the ULDM problem (20)&(21) from the upper bound model and obtain
Step 4.
`
Use Eq. (25)&(26) to construct V‚‚ƒ„ and V‚†ƒ„
… …

Ask the DM to select p, ?b = 1,2, … , ∞A.

Step 5. .

Construct the payoff table of problem (27)-(29) and obtain 'VUΠ+ and 'VUΠ+ .
∗ L
Step 6.
Step 7.
 International Journal of Management and Fuzzy Systems 2016; 2(5): 38-50 46

••‘ … ’•‘ …
Ž• Ž•
Formulate and solve the model (34)-(38) for the ULDM problem to get Π= ' GΠ, HΠ+,
Step 8. Elicit the membership functions and .
∗ ∗ ∗

G = ' GG , GH , … , Ge` +.
Step 9. Œ∗ Œ∗ Œ∗ Œ∗

Set the maximum negative and positive tolerance values on the decision vector GŒ = ' GG , GH , … , Ge +, 3yC
∗ Œ∗ Œ∗ Œ∗
`
and 3ỹ , { = 1,2, … , OG .
Step 10.
Construct the PIS payoff table of the -(BL-MOP) problem from the lower bound model and obtain E ∗ =
'FGG
∗
, FGH
∗
, … , FGJ
∗
F ∗ , F ∗ , … , FHJ
` HG HH
∗
+, the individual positive ideal solutions for both levels.
Step 11.
Construct the NIS payoff table of the -(BL-MOP) problem from the upper bound model and obtain E L =
f

'FGG
L
, FGH
L
, … , FGJ
L
F L , F L , … , FHJ
` HG HH
L
+, the individual negative ideal solutions for both levels.
Step 12.
f
Use Eqs. (41) and (42) to construct V‚‚ƒ„ and V‚†ƒ„
• •

Construct the payoff table of problem (43)-(45) and obtain 'VU + and 'VU + .
∗ L
Step 13. , respectively.
Step 14.
Elicit the membership functions ••‘• and ’•‘• .
Ž• Ž•
Elicit the membership functions (`z Gy , { = 1,2, … , OG .
Step 15.

Formulate and solve the model (52)-(58) for the -(BL-MOP) problem to get ∗ = G∗ , H∗ .
Step 16.
Step 17.

Modify the maximum negative and positive tolerance values on the decision vector GŒ = ' GG , GH , … , Ge` +,
∗ Œ∗ Œ∗ Œ∗
Step 18. If the DM is satisfied with the candidate solution in Step 17, go to Step 20, otherwise go to Step 19.

3y and 3ỹ , { = 1,2, … , OG , go to Step 16.

C
Step 19.
Step 20. Stop with a satisfactory solution, ∗ = G∗ , H∗ , to the -(BL-MOP) problem.

Since the problem is minimization-type then replacing the

6. Illustrative Numerical Example fuzzy coefficient by their lower bound -cuts, for =0.5, the
BL-MOP problem with fuzzy parameters reduces to a
The following numerical example studied in [7, 8, 16] is deterministic BL-MOP problem as follows:
considered to illustrate the modified TOPSIS approach for [1st level]
‰
Zk`` ( ^ (` š H.§(f š (¤ š H.§(¥
solving BL-MOP problem with fuzzy parameters.

¹̧
‹.µ

[1st level]
‰
Zk`` ( ^ (` šž
£(f šH£(¤ šž£(¥ , N8O
P ³́ Zk`f (
‹.µ
^ (` š¶.§(f šH.§(¤ šŸ.§(¥

P ¢
N8O k`f ( ^ H£(` š¦g(f šž£(¤ š§£(¥ ,
¨
‰
Z (` ,(f Zk`¤ ( ^ H.§(` š ¶.§(f š ¶.§(¤ š(¥
² ·
Zk`¤ ( ^ ž£(` š¦g(f š¦g(¤ š (¥ ‹.µ
(` ,(f

nd nd
[2 level] [2 level]

P
N8O
Zkf` ( ^ ©
£(` šž
£(f šH£(¤ šH£(¥ , ‰
Zkf` ( ^§.§(` š H.§(f š (¤ š (¥
Zkff ( ^ §
N8O
P¢ ¨
£(` š¦g(f L¦g(¤ š©
£(¥ ‹.µ
(¤ ,(¥ ‰
(¤ ,(¥ Zkff ( ^Ÿ.§(` š¶.§(f L¶.§(¤ š§.§(¥
9n;R5<3 32
‹.µ

9n;R5<3 32
3g G− H+
g
ž +3
- , 2g
≤ 48 g
G +4
g
H +2
g
ž−2
-,
≤ 35
Ÿ Ÿ
2.5 G − H + ž + 2.5 Ÿ ≤ 48.5
g
G+2 − + -,
≥ 30 G, H, ž, Ÿ ≥ 0. + 2.5 H − ž + Ÿ ≥ 29
H ž Ÿ ∈ o~ = º G »
G + 3.5 H + ž − Ÿ ≤ 36
G , H , ž , Ÿ ≥ 0.
Here, the fuzzy numbers are assumed to be triangular
fuzzy numbers and are given as follows:

2g = 0, 2, 3 , 3g = 2, 3, 4 , 4g = 3, 4, 5 , 5g = 4, 5, 6 , 6g =
The individual minimum and maximum of each of the

5, 6, 7 , 8g = 6, 8, 10 , 9g = 8, 9, 10 , 30
- = 28, 30, 32 .
objective functions at both levels calculated from the lower

- -
35 = 33, 35, 37 , 48 = 45, 48, 49 .
and upper bound -cut model are given in Table 1.

Table 1. Minimum and maximum individual optimal solutions.

¼g½½ ¾ ¼g½Ã ¾ ¼g½Ä ¾ ¼gý ¾ ¼gÃà ¾

Á, Á, Á, Á, Á,Â

¿.À ¿.À ¿.À ¿.À ¿.À

Objective Function

N8O FkhM
C

~.§
29 48.862 48.862 29 55.875

FkhM
D
N:
~.§
155.47 315.79 268.46 152.1 342.34

We first obtain PIS and NIS payoff tables for the ULMD problem from the lower and upper bound model respectively
(Table 2 and 3):
47 Ibrahim A. Baky and M. A. El Sayed: Bi-level Multi-objective Programming Problems with Fuzzy
Parameters: Modified TOPSIS Approach

Table 2. PIS payoff table of the ULDM problem.

¼g½½ ¾ ¼g½Ã ¾ ¼g½Ä ¾

Á Á Á
¾½ ¾Ã ¾Ä ¾Å
¿.À ¿.À ¿.À
Objective Function

N8O FkGG
C
29∗
~.§
71 88.25 11.5 7 0 0

N8O FkGH
C
48.862∗
~.§
29.005 79.96 20.73 3.31 0 0

N8O FkGž
C
48.862∗
~.§
60.1 121.4 0 3.31 0 20.73

E Œ = FGG , FGH , FGž = 29, 48.862, 48.862

∗ ∗ ∗ ∗

Table 3. NIS payoff table of the ULDM problem.

¼g½½ ¾ ¼g½Ã ¾ ¼g½Ä ¾

  Â
¾½ ¾Ã ¾Ä ¾Å
¿.À ¿.À ¿.À
Objective Function

FkGG
D
N: 155.47∗
~.§
315.79 196.32 0 17.87 0 26.55

FkGH
D
N: 315.79 ∗
~.§
155.47 196.32 0 17.87 0 26.55

FkGž
D
N: 268.46∗
~.§
138.11 253.67 0 10.38 15.85 17.5

E Œ = FGG , FGH , FGž = 155.47, 315.79, 268.46

\ L L L

Assume that 7h = , 8 = 1, 2, 3, then the equations for V‚‚ƒ„ and V‚†ƒ„ when b = 2 are:
G … …
ž
GÆ
0.0000069[ G + 2.5 H + ž + 2.5 Ÿ − 29]H + H

EGŒ = VH‚ƒ„ = •0.00000156[ G + 8.5 H + 2.5 ž + 4.5 − 48.862]H +•

…
Ÿ
0.0000023[ 2.5 G + 8.5 H + 8.5 ž + Ÿ − 48.862]H
GÆ
0.0000069[155.47 − G + 2.5 H + ž + 2.5 Ÿ ]H + H

EHŒ = VH †ƒ„ …
= •0.00000156[315.79 − G + 8.5 H + 2.5 ž + 4.5 Ÿ ]H +•
0.0000023[268.46 − 2.5 G + 8.5 H + 8.5 ž + Ÿ ]H
Next to formulate model (34)-(38) we determine the following N: EGΠ= 0.41, N8O EGΠ= 0.044 , and N: EHΠ=
0.553, N8O EHΠ= 0.188. Thus we have VHΠ= 0.044, 0.553 and VHΠ= 0.41, 0.188 , therefore, the membership functions
∗ \

G :OV H can be obtained as:

Ç`… = 1.12 − 2.73EGŒ

Çf… = −0.515 + 2.74EHŒ

Then, the modified TOPSIS formulation for the ULDM problem is obtained as follows:

N8O — = 2.73 •G‚ƒ„ + 2.74•H†ƒ„

– \

9n;R5<3 32

1.12 − 2.73EGŒ + •G‚ƒ„ − •G‚ƒ„ = 1

\ –

−0.515 + 2.74EHŒ + •H†ƒ„ − •H†ƒ„ = 1

\ –

2.5 G − H + ž + 2.5 Ÿ ≤ 48.5

G + 2.5 H − ž + Ÿ ≥ 29

G + 3.5 H + ž − Ÿ ≤ 36

G, H, ž, Ÿ ≥ 0. •G‚ƒ„ , •G‚ƒ„ , •H†ƒ„ , •H†ƒ„ ≥ 0,

\ – \ –

The optimal solution of the ULDM problem is achieved at Π= 20.678, 3.315,0.0096, 0.044 . Let the upper-level DM
∗

decide GŒ = 20.678, and HŒ = 3.315, with positive tolerances 3G˜ = 3H̃ = 0.5 and weights of 7Gy = = 2, (one sided
∗ ∗ ˜ G
~.§
membership function [23, 28]).
We first obtain PIS and NIS payoff tables for the second level MOP problem from the lower and upper bound model
 International Journal of Management and Fuzzy Systems 2016; 2(5): 38-50 48

respectively (Tables 4 and 5):

Table 4. PIS payoff table of the LLDM problem.

¼gý ¾ ¼gÃà ¾
Á Á
¾½ ¾Ã ¾Ä ¾Å
¿.À ¿.À
Objective Function

N8O FkHG
C
29∗
~.§
102.6 0 10.83 0 1.917

N8O FkHH
C
55.875 ∗
~.§
60.16 0 10.83 15.58 17.5

E È = FHG , FHH = 29, 55.875

∗ ∗ ∗

Table 5. NIS payoff table of the LLDM problem.

¼gý ¾ ¼gÃà ¾
 Â
¾½ ¾Ã ¾Ä ¾Å
¿.À ¿.À
Objective Function

FkHG
D
N: 152.1∗
~.§
156.52 21.1 4.26 0 0

FkHH
D
N: 342.34∗
~.§
128.92 0 17.87 0 26.55

E È = FHG , FHH = 152.1, 342.34

\ L L

Assume that 7h = 8 = 1, 2, 3, 4, 5, then the equations for V‚‚ƒ„ and V‚†ƒ„ when b = 2 are:
G • •
§
GÆ
0.0000025[ G + 2.5 H + ž + 2.5 Ÿ − 29]H + H
& Ì
$ 0.00000056 [ G + 8.5 H + 2.5 ž + 4.5 Ÿ − 48.862]H + $
ÉG = VH‚ƒ„ = 0.00000083[ 2.5 G + 8.5 H + 8.5 ž + Ÿ − 48.862]H +
•

% Ë
$ 0.0000026[ 5.5 G + 2.5 H + ž + Ÿ − 29]H + $
# 0.00000049[ 4.5 G + 8.5 H − 8.5 ž + 5.5 Ÿ − 55.875] Ê
H

GÆ
0.0000025[155.47 − G + 2.5 H + ž + 2.5 Ÿ ]H + H
& Ì
$ 0.00000056 [315.79 − G + 8.5 H + 2.5 ž + 4.5 Ÿ ] + $
H

ÉH = VH†ƒ„ = 0.00000083[268.46 − 2.5 G + 8.5 H + 8.5 ž + Ÿ ]H +

•

% Ë
$ 0.0000026[152.1 − 5.5 G + 2.5 H + ž + Ÿ ] + $
H

#0.00000049[342.34 − 4.5 G + 8.5 H − 8.5 ž + 5.5 Ÿ ]H Ê

7.64 − 2 H + •H̃ L − •H̃ š = 1
N: ÉG = 0.303, N8O ÉG = 0.065 , and N: ÉH =
Next we determine the following

0.396, N8O ÉH = 0.175 2.5 G − H + ž + 2.5 Ÿ ≤ 48.5

VH = 0.065, 0.396 and VH = 0.303,0.175 , therefore,
∗ \
+ 2.5 − + ≥ 29
. Thus we have

the membership functions ž :OV Ÿ

G H ž Ÿ

+ 3.5 + − ≤ 36
can be obtained
as: G H ž Ÿ

= 1.27 − 4.23ÉG G, H, ž, Ÿ ≥ 0. •G‚ƒ„ , •G‚ƒ„ , •H†ƒ„ , •H†ƒ„ ≥ 0,

\ – \ –
‚`•

= −0.792 + 4.52ÉH •ž‚ƒ„ × •ž‚ƒ„ = 0, •Ÿ†ƒ„ × •Ÿ†ƒ„ = 0, and •ỹ L × •ỹ š =

\ – \ –
‚f•
0 { = 1,2
Finally, the modified TOPSIS formulation for the -(BL-
∗
= 20.68, 3.32, 0, 0.02 with objective function values
MOP) problem is obtained as: The satisfactory solution of the -(BL-MOP) problem is

N8O — = 4.23•ž‚ƒ„ + 4.52•Ÿ†ƒ„ + 2•G˜L + 2•G˜š + 2•H̃ L FGG = 29.03, FGH = 48.99, FGž = 79.94, FHG = 122.1,
– \

+ 2•H̃ š FHH = 121.4, and their corresponding membership functions

and

are GG = 0.9997, GH = 0.9995, Gž = 0.858, HG = 0.244,

9n;R5<3 32 and HH = 0.771, respectively.

1.27 − 4.23ÉG + •ž‚ƒ„ − •ž‚ƒ„ = 1

\ –
The comparison given in Table 6 between the modified
TOPSIS approach and the TOPSIS approach [8] shows that

−0.792 + 4.52ÉH + •Ÿ†ƒ„ − •Ÿ†ƒ„ = 1

\ –
the modified TOPSIS approach is greatly preferred than the
later approach.
42.356 − 2 G + •G˜L − •G˜š = 1
49 Ibrahim A. Baky and M. A. El Sayed: Bi-level Multi-objective Programming Problems with Fuzzy
Parameters: Modified TOPSIS Approach

Table 6. Comparison between the modified TOPSIS approach and the TOPSIS approach [8].

FGG " 29.03 GG = 0.999 FGG = 57.19 = 0.78 FGG = 29

The modified TOPSIS approach The TOPSIS approach [8] The optimal solution
GG
FGH = 48.99 GH = 0.999 FGH = 114.93 GH = 0.75 FGH = 48.862
FGž = 79.94 Gž = 0.858 FGž = 51.88 Gž = 0.986 FGž = 48.862
FHG = 122.1 HG = 0.244 FHG = 37.16 HG = 0.94 FHG = 29
FHH = 121.4 HH = 0.771 FHH = 140.1 HH = 0.71 FHH = 55.875

approach. And because the TOPSIS approach transfers Í

For indicating the merits of the modified TOPSIS
approach, the comparison is given in Table 7 among
modified TOPSIS approach, the method of Pramanik [16]
and the FGP algorithm [7] by Baky et al. to solve the BL-
MOP problem with fuzzy parameters shows that the values
of objective and membership functions of the BL-MOP
problem with fuzzy parameters obtained from the modified
TOPSIS are more preferred than that given by the TOPSIS
approach and FGP algorithm. Since the modified TOPSIS
Table 7. Comparison between the modified TOPSIS approach, the method of Pramanik et al. and FGP algorithm.
approach combines the advantages of both TOPSIS and FGP
objectives which are conflicting and non-commensurable into
two objectives (the shortest distance from the PIS and the
longest distance from the NIS), those which are
commensurable and most of time conflicting. Hence, FGP
approach solves the bi-objective problem to obtain the
satisfactory solution.

FGG = 29.03 GG = 0.999 FGG = 37 GG = 0.902 FGG = 29 =1

The modified TOPSIS approach The method of Pramanik [16] The FGP algorithm [7]
GG
FGH = 48.99 GH = 0.999 FGH = 90 GH = 0.815 FGH = 70.95 GH = 0.917
FGž = 79.94 Gž = 0.858 FGž = 108.25 Gž = 0.692 FGž = 88.27 Gž = 0.821
FHG = 122.1 HG = 0.244 FHG = 78.25 HG = 0.496 FHG = 80.9 HG = 0.578
FHH = 121.4 HH = 0.771 FHH = 105.5 HH = 0.795 FHH = 111.27 HH = 0.81

multi-objective dynamic programming problems under

7. Conclusion and Summary
This paper reveals how the concept of modified TOPSIS
approach can be efficiently used to solve the BL-MOP
problem with fuzzy parameters. In order to obtain a
compromise (satisfactory) solution to the BL-MOP problem
with fuzzy parameters using the modified TOPSIS approach,
firstly the crisp model is obtained at an -level. Then, the
distance function from the positive ideal solution and the
distance function from the negative ideal solution are
developed. Then, the bi-objective problem can be formulated
by using the membership functions of the fuzzy set theory to
represent the satisfaction level for both criteria. Hence, in the
modified TOPSIS approach, the FGP approach is used to
solve the confliction between the new criteria instead of the
conventional max-min decision model. An illustrative
numerical example is given to demonstrate the efficiency of
the proposed modified TOPSIS approach for BL-MOP
problem with fuzzy parameters. The comparison among the
proposed modified TOPSIS approach and the existing
methods is given in Tables 6&7. Finally, it is hoped that the
concept of solving BL-MOP problem with fuzzy parameters
presented here can contribute to future studies in the other
fields of MODM problems.
References
[1] Mahmoud. A. Abo-Sinna and A. H. Amer, Extensions of
TOPSIS for multi-objective large-scale nonlinear
programming problems, Appl. Math. Comput. 162 (2000)
243-256.
[2] Mahmoud. A. Abo-Sinna, Extensions of the TOPSIS for
fuzziness, Adv. Model. Anal. (AMSE Press, France) B 43 (4)
(2000) 1-24.
[3] Mahmoud. A. Abo-Sinna, A. H. Amer and A. S. Ibrahim,
Extensions of TOPSIS for large-scale multi-objective
nonlinear programming problems with block angular
structure, Appl. Math. Model. 32 (2008) 292-302.
[4] Ibrahim A. Baky and M. A. Abo-Sinna, TOPSIS for bi-level
MODM problems, Appl. Math. Model. 37 (3) (2012) 1004-1015.
[5] Ibrahim A. Baky, Fuzzy goal programming algorithm for
solving decentralized bi-level multi-objective programming
problems, Fuzzy Sets Systems, vol. 160, no. 18, pp. (2009)
2701-2713.
[6] Ibrahim A. Baky, Solving multi-level multi-objective linear
programming problems through fuzzy goal programming
approach, Applied Mathematical Modelling, 34 (9) (2010)
2377-2387.
[7] Ibrahim A. Baky, M. H. Eid, M. A. Elsayed, Bi-level multi-
objective programming problem with fuzzy demands: a fuzzy
goal programming algorithm, Journal of Operational Research
Society of India (OPSEARCH), 51 (2) (2014) 280-296.
[8] Ibrahim A. Baky, M. H. Eid, M. A. Elsayed, Bi-level multi-
objective programming problem with fuzzy parameters:
TOPSIS approach, the 38th International Conference for
statistics, computer science and its applications, Cairo, Egypt,
8-18 April, 13-33, 2013.
[9] C. T. Chen, Extensions of the TOPSIS for group decision-making
under fuzzy environment, Fuzzy Set Syst. 114 (2000) 1-9.
[10] H. Deng, C. H. Yeh and R. J. Willis, Inter-company
comparison using modified TOPSIS with objective weights,
Comput. Oper. Res. 17 (2000) 963-973.
[11] C. L. Hwang and K. Yoon, Multiple attribute decision making:
methods and applications, Springer-Verlag, Heidelberg, 1981.
 International Journal of Management and Fuzzy Systems 2016; 2(5): 38-50
[12] Y.-J. Lai, T.-J. Liu and C. L. Hwang, TOPSIS for MODM,
Euro. J. Oper. Res. 76 (1994) 486-500.
with Pareto optimum, Fuzzy Sets and Systems 53, 1993 275-
[13] E. S. Lee and R. J. Li, Fuzzy multiple objective programming
288.
and fuzzy programming, Fuzzy Sets and Systems, 89 1997
[14] R. H. Mohamed, The relationship between goal programming
215-222.
[15] S. Pramanik and T. K. Roy, Fuzzy goal programming approach
Operational Research, vol. 176, no. 2, pp 2007 1151-1166.
to multi-level programming problems, European Journal of
[16] S. Pramanik and P. P. Dey, Bi-level multi-objective
Journal of Computer Applications, 30 11 2011 13-20.
programming problem with fuzzy parameters. International
[17] B. B. Pal, B. N. Moitra, and U. Maulik, A goal programming
procedure for fuzzy multi-objective linear fractional
139 2 2003 395-405.
programming problem, Fuzzy Sets and Systems,
View publication stats
50
[18] S. Sinha, Fuzzy programming approach to multi-level
2003 189-202.
programming problems, Fuzzy Sets and Systems, 136
[19] M. Sakawa and I. Nishizaki, Interactive fuzzy programming for
Sets and Systems, 125 2002 301-315.
decentralized two-level linear programming problems, Fuzzy
[20] G. Zhang, J. Lu and T. Dillon, Decentralized multi-objective
Based System, 20 (5) 2007 , 495-507.
bi-level decision making with fuzzy demands, Knowledge-
[21] L. A. Zadeh, Fuzzy Sets, Inform. Control 8 (1965) 338-353.
[22] B. N. Moitra and B. B. Pal, A fuzzy goal programming
approach for solving bi-level programming problems, in
Advances in Soft Computing – AFSS, N. R. Pal and M.
Sugeno, Eds., vol. 2275 of Lecture Notes in Computer
Science, PP. 91-98, Springer, Berlin, Germany, 2002.
[23] M. Sakawa, Fuzzy sets and interactive multi-objective
optimization, Plenum Press, New York, NY, USA, 1993.
[24] M. Zeleny, Multiple criteria decision making, McGraw-Hill,
New York, 1982.