You are on page 1of 9

European Journal of Operational Research 133 (2001) 548±556

www.elsevier.com/locate/dsw

Theory and Methodology

Fuzzy goal programming with di€erent importance and priorities


Liang-Hsuan Chen *, Feng-Chou Tsai
Department of Industrial Management Science, National Cheng Kung University, Tainan, Taiwan, ROC
Received 19 May 1999; accepted 6 July 2000

Abstract

This paper formulates fuzzy goal programming (FGP) incorporating di€erent importance and preemptive priorities
by using an additive model to maximize the sum of achievement degrees of all fuzzy goals. In contrast to previous
works, the proposed approach allows the decision-maker to determine a desirable achievement degree for each fuzzy
goal to re¯ect explicitly the relative importance of these goals. This approach can generate a set of achievement degrees
consistent with the decision-maker's expectations, even though the relative importance of the goals may change.
Furthermore, in this paper we incorporate the decision-maker's preemptive priority structure into a single formulation.
The resulting solutions satisfy both the preemptive priority structure and have the maximum achievement degrees in
sum. The proposed approaches' e€ectiveness and computational superiority over the existing approaches are demon-
strated and compared with examples from the literature. Ó 2001 Published by Elsevier Science B.V.

Keywords: Goal programming; Fuzzy goal programming; Preemptive priority; Relative importance

1. Introduction such as probability distribution, penalty function,


fuzzy numbers and various types of thresholds
The goal programming (GP) model is useful for [2,3,19,20,34]. For modeling goals with imprecise
decision-makers to consider simultaneously sev- nature, Martel and Aouni [23±25] and Aouni et al.
eral objectives in ®nding a set of acceptable solu- [1] have reformulated the standard GP model [5]
tions. However, determining precisely the goal by introducing the satisfaction degrees as a func-
value of each objective is dicult for decision- tion of the goal deviations in the objective function
makers, since possibly only partial information of the model. The indi€erence threshold concept is
can be obtained [41]. To incorporate uncertainty used in the satisfaction function for characterizing
and imprecision into the formulation, some ap- the imprecision associated with the goal values.
proaches are used to reformulate the GP models, This approach can allow the decision-maker to
express his preference structure in the formulation
in an explicit way.
*
Corresponding author. Tel.: +886-6-275-7575; fax: +886-6-
In order to specify imprecise aspiration levels of
236-2162. the goals in a fuzzy environment, Narasimhan [28]
E-mail address: lhchen@mail.ncku.edu.tw (L.-H. Chen). had initially proposed fuzzy goal programming

0377-2217/01/$ - see front matter Ó 2001 Published by Elsevier Science B.V.


PII: S 0 3 7 7 - 2 2 1 7 ( 0 0 ) 0 0 2 0 1 - 0
L.-H. Chen, F.-C. Tsai / European Journal of Operational Research 133 (2001) 548±556 549

(FGP) by using membership functions. This work achievement degree of each fuzzy goal to the
and some related studies [30,33,35] are actually model in an explicit way.
inspired by a fuzzy programming approach intro- Furthermore, decision-makers may have an
duced by Zimmermann [43]. The FGP formulation order in which to achieve the goals, i.e., a pre-
has widespread applications to various ®elds emptive priority structure, so that some goals are
[4,9,10,12,26,27,29,31±33,36,37]. Some researchers expected to be achieved before others in sequence.
[6,13±17,38±40] have provided further investiga- In the literature, the existing approaches categorize
tions of FGP regarding the problem formulation, fuzzy goals into k priority levels, where the number
the relative importance and the fuzzy priority of k is less than the number of the fuzzy goals [6,38,
the fuzzy goals, and associated solution algo- 39]. Then k subproblems are solved in sequence by
rithms. In the FGP formulations, most of the specifying the membership grades to be achieved
previous researchers except Tiwari et al. [39] have for the fuzzy goals that belong to the higher pri-
used the min-operator to ®nd the fuzzy decision ority levels as additional constraints in the sub-
that simultaneously satis®ed fuzzy goals and fuzzy problems for the lower priority levels. Although
constraints. Then the maximizing decision is de- the resulting answers can satisfy the decision-ma-
termined to be the maximum degree of member- ker's preemptive priority structure in the above
ship for the fuzzy decision. The current approach ways, obviously the computational eciency will
considers that the fuzzy decision is the intersection decrease when the number of priority levels in-
of fuzzy goals and constraints, so that there is no creases. The eciency for resolution procedure can
fundamental di€erence between them. Although be enhanced if we can incorporate the preemptive
the current approach is ecient in computation, priority structure into a single formulation.
the application of it may produce ``uniform'' This paper will focus on the reformulation of
membership degrees for fuzzy goals when the FGP using an additive model to eciently resolve
achievement of some goals is stringently required. the problem with di€erent importance levels and/
We will illustrate this condition in the following or with preemptive priorities. We will reformulate
section, and an additive model is adopted in this the FGP by expressing explicitly the decision-ma-
study to aggregate the fuzzy goals for further in- ker's desirable achievement degree of each fuzzy
vestigation of the FGP. goal and by incorporating the decision-maker's
Consideration of di€erent relative importance preemptive priority structure into a single formu-
and priorities of the goals in the FGP problem is lation. In the following section, we compare the
important because some goals are more important resulting solutions of a particular problem that is
than others. Narasimhan [28] has used linguistic formulated by using both the min-operator and an
variables, such as ``very important'' and ``moder- additive model to aggregate the fuzzy goals. Sec-
ately important'', to describe the fuzzy weights of tion 3 describes the FGP problem considering the
the goals, and de®ned the corresponding mem- fuzzy goals with di€erent desirable achievement
bership functions by specifying the desirable in- degrees. An FGP problem containing the decision-
tervals of membership degree to re¯ect the maker's preemptive priority structure is presented
importance. However, this approach may produce in Section 4. The eciency in resolving this prob-
a set of unreasonable answers [14]. Hannan [13] lem is shown in this section. Finally, we provide
and Tiwari et al. [39] have used di€erent weights conclusions in Section 5.
for the various goals in order to re¯ect the relative
importance of the goals, and considered the
weights as the coecients of the objective function. 2. Related models
This study illustrates that such approaches may
yield counter-intuitive results when the relative Since Narasimhan [28] had presented the initial
importance of the goals is changed. For actually FGP model and the solution procedure, a few
re¯ecting the relative importance of the goals in studies have proposed FGP models for improving
this paper, we add the decision-maker's desirable the computational eciency. Hannan [13,14] has
550 L.-H. Chen, F.-C. Tsai / European Journal of Operational Research 133 (2001) 548±556

introduced conventional deviation variables into The linear membership function li for the ith fuzzy
the model, so that only a conventional linear pro- goal based on Zimmermann [43,44] is de®ned as
gramming formulation is required; although, this 8
> 1 if Gi …x† P gi ;
increases the number of variables in the formula- < G …x† L
i i
tion. Based on piecewise linear approximation li ˆ if Li 6 Gi …x† 6 gi ; …4a†
>
: g i Li
(PLA), Yang et al. [40] have further formulated the 0 if Gi …x† 6 Li ;
problem with fewer variables, which can yield the
same solutions as Narasimhan's [28] and Hannan's or
[14]. Yang et al.'s model can be formulated as 8
> 1 if Gi …x† 6 gi ;
follows. Let Gk …x† denote the kth fuzzy goal with a <U Gi …x†
i
triangular membership function li ˆ if gi 6 Gi …x† 6 Ui ; …4b†
>
: Ui gi
8 0 if Gi …x† P Ui ;
> 0 if Gk …x† P bk ‡ dk2 ;
>
>
>
> G k …x† b k
>
> 1 if bk 6 Gk …x† 6 bk ‡ dk2 ; where Li (or Ui ) is the lower (upper) tolerance limit
< dk2 for the ith fuzzy goal Gi …x† J gi …Gi …x† K gi †. The
lk ˆ 1 if Gk …x† ˆ bk ;
>
> simple additive model is formulated as
>
> b k G k …x†
>
> 1 if bk dk1 6 Gk …x† 6 bk ; Xn
>
: dk1
0 otherwise; Maximize f …l† ˆ lk
kˆ1
…1† Gi …x† Li
subject to li ˆ for some i;
where bk is the aspiration level of the kth goal, and g i Li
dk1 and dk2 are the maximum allowable negative Ui Gi …x†
lj ˆ for some j; j 6ˆ i;
and positive deviations from bk , respectively. Then U i gi
the resulting linear programming formulation is Ax 6 b;
Maximize k li ; lj 6 1;
Gk …x† bk x; li ; lj P 0; i; j 2 f1; . . . ; ng;
subject to k61 ;
dk2 …5†
…2†
bk Gk …x†
k61 ; where Ax 6 b are the crisp system constraints in
dk1
vector. Note that problem (5) is to be optimized by
k; x P 0; for all k maximizing the sum of each goal's achievement
The models mentioned above commonly use degree li (and/or lj ). This use of an additive model
the min-operator for aggregating goals to deter- can obtain the maximum sum of goals' achieve-
mine the decision set, and then to maximize the ment degrees. And more importantly, the
set. Instead, Tiwari et al. [39] have presented a achievement degrees of some goals will not de-
simple additive model to formulate an FGP crease because of a particular goal that is dicult
problem. Adopting their notations, a solution set x to achieve. However, the achievement degrees of
is found for the following FGP problem that all goals will be lower if the min-operator is used
contains m fuzzy goals Gi …x†: for the same conditions. For illustration, we use an
example, given by Narasimhan [28], formulated as
Gi …x† J gi …or Gi …x† K gi †; i ˆ 1; . . . ; m two problems, both by PLA with the min-operator
subject to …3† as in (2) and by PLA with an additive model, to
Ax 6 b; x P 0; compare the resulting solutions. The example is to
®nd the solutions which best achieve pro®t goals
where Gi …x† J … K †gi indicates the ith fuzzy goal and sales goals for two kinds of products. The
approximately greater than or equal to (approxi- problem containing the three fuzzy goals can be
mately less than or equal to) the aspiration level gi . formulated as follows [14,28,38,40]:
L.-H. Chen, F.-C. Tsai / European Journal of Operational Research 133 (2001) 548±556 551

G1 : 80x1 ‡ 40x2  630; problem (8) are …l1 ; l2 ; l3 ) ˆ (1.0, 0.438, 1.0) and
G2 : x1  6; …6† (x1 ; x2 ) ˆ (5.88, 4.0). The solution k ˆ 0.64 to
problem (7) actually means (l01 ; l02 ; l03 ) ˆ (0.64,
G3 : x2  4;
0.64, 0.64), considering the respective achievement
where x1 and x2 denote the units of products 1 degree of the fuzzy goal. Based on the resulting
and 2, respectively, the symbol  represents the solutions of (8), the achievement degree of G2 is
linguistic term ``about'', and the right-hand side small because it is dicult to achieve; however, the
of each goal in (6) means the aspiration level of other two goals are completely achieved. In con-
the goal. In this formulation, G1 indicates the trast, the achievement degrees of G1 and G3 lessen
pro®t goal and the others are sales goods for and are equivalent to G2's degree (0.64) in (7). As
products 1 and 2. The maximum allowable a whole, the sum of achievement degrees in
negative and positive deviations of each goal's problem (8) is greater than in problem (7). This
aspiration level are the same, and are set as 10, advantage makes the additive model appealing,
2, and 2, respectively. As mentioned above, and this model will be applied in the following
Yang et al. [40] have solved this problem using sections.
PLA with the min-operator and had the same
solutions as [14,28]. As an illustration, if the
aspiration level of product 1 …G2† is changed 3. The model with di€erential achievement degrees
and, for example, increases to 7, i.e., x1  7, the
PLA formulation based on the same deviation In order to re¯ect the relative importance of the
constants is goals, the weighted average of deviations from
aspiration levels is usually used in conventional
Maximize k
goal programming [18,21,42]. Hannan [13,14] has
subject to k 6 8x1 4x2 ‡ 64; also introduced this approach to formulate the
k 6 8x1 ‡ 4x2 62; objective function of an FGP with fuzzy priority.
k6 0:5x1 ‡ 4:5; Tiwari et al. [39] have proposed a weighted addi-
…7† tive model that incorporates each goal's weight
k 6 0:5x1 2:5;
into thePobjective function of problem (5), i.e.,
k 6 0:5x2 ‡ 3; f …l† ˆ nkˆ1 wk lk , whereP wk denotes the weight of
k 6 0:5x2 1; the kth fuzzy goal, and wk ˆ 1. Weights in the
k; x1 ; x2 P 0: weighted additive model reveal the relative im-
portance of the fuzzy goals. However, the model
According to the same problem as (7), the PLA may produce undesirable solutions when the
with the additive model is formulated as weights are changed. As an illustration, Tiwari
et al.'s example [39] containing ®ve fuzzy goals
Maximize l1 ‡ l2 ‡ l3 with four variables and four system constraints is
subject to l1 6 8x1 4x2 ‡ 64; used here. The ®ve fuzzy goals in the problem are
l1 6 8x1 ‡ 4x2 62; described as
l2 6 0:5x1 ‡ 4:5;
…8† 4x1 ‡ 2x2 ‡ 8x3 ‡ x4 K 35;
l2 6 0:5x1 2:5;
l3 6 0:5x2 ‡ 3; 4x1 ‡ 7x2 ‡ 6x3 ‡ 2x4 J 100;
l3 6 0:5x2 1;
x1 6x2 ‡ 5x3 ‡ 10x4 J 120; …9a†
li ; xj P 0; i ˆ 1; 2; 3; j ˆ 1; 2:
5x1 ‡ 3x2 ‡ 2x4 J 70;
The solutions to problem (7) are k ˆ 0:64,
(x1 ; x2 ) ˆ (6.28, 3.28); however, the solutions to 4x1 ‡ 4x2 ‡ 4x3 K 40
552 L.-H. Chen, F.-C. Tsai / European Journal of Operational Research 133 (2001) 548±556

subject to X
5
Maximize li
iˆ1
7x1 ‡ 5x2 ‡ 3x3 ‡ 2x4 6 98; subject to the membership functions
7x1 ‡ x2 ‡ 6x3 ‡ 6x4 6 117; …li ; i ˆ 1; . . . ; 5† of …9a†;
x1 ‡ x2 ‡ 2x3 ‡ 6x4 6 130; …9b† the system constraints …9b†;
9x1 ‡ x2 ‡ 6x4 6 105; …10†
l1 P 0:8;
xi P 0; i ˆ 1; . . . ; 4: l2 P 0:7;
l3 P 0:5;
The tolerance limits and relative weights of l4 P 0:6;
the ®ve fuzzy goals are (55, 40, 70, 30, 10) and
l5 P 0:8:
(0.49, 0.131, 0.153, 0.114, 0.112), respectively.
After formulating the fuzzy goals in (9a) as the
membership functions in problem (5) and con- The above problem can be solved by conven-
sidering the crisp system constraints (9b), the tional linear programming, and the resulting
weighted additive model can be determined. achievement degrees for the ®ve fuzzy goals are
The resulting respective achievement degrees for (0.991, 0.988, 0.621, 0.768, 0.952), respectively, all
the ®ve fuzzy goals are (1.0, 0.977, 0.636, 0.761, of which satisfy the requirements. Then, assume
0.939). Instead, we place more emphasis on the that the ®rst and third goals are to be emphasized,
fourth fuzzy goal, increasing its weight to 0.7, with their desirable achievement degrees increasing
and the respective relative weights are (0.001, to 1.0 and 0.8, respectively, while the others re-
0.05, 0.2, 0.7, 0.049). The resulting achievement main unchanged. Applying this approach, the de-
degrees for the ®ve fuzzy goals will be (0.45, 1.0, termined achievement degrees are (1.0, 0.895, 0.8,
1.0, 0.675, 0.988). Note that the achievement 0.699, 0.844), which still satis®es the objective.
degree of the fourth fuzzy goal lessens, although The determination of a desirable achievement
its weight is signi®cantly increased. In other degree for a goal could be a dicult task for a
words, in this situation we obtain an undesirable decision-maker in a fuzzy environment when using
consequence. this method. For assessing desirable achievement
Instead of the weighted additive model, this degrees imprecisely, a useful method is to use lin-
study allows the decision-maker to determine ex- guistic terms, such as ``important'', ``somewhat
plicitly a desirable achievement degree for each important'', and ``very important'', and so on, to
fuzzy goal as the importance of the fuzzy goal. verbally describe the importance of each fuzzy
That is, the more important the goal, the higher goal. The associated membership functions are
the desirable achievement degree. This approach then de®ned. For example, we can de®ne l~I …a† to
can ensure that more important goals can have represent the membership function of ``impor-
higher achievement degrees. To do this, we refor- tant'', where l~I …a† 2 ‰0; 1Š, and a denotes the
mulate Tiwari et al.'s model, shown as (5), by variable taking an achievement degree in the in-
adding a set of desirable achievement degrees as terval of ‰amin ; amax Š, 0 6 amin 6 amax 6 1. Then some
constraints, i.e., li P ai , where ai is the desirable useful methods [7,8,11,22] for ranking fuzzy
achievement degree for the ith fuzzy goal. For il- numbers can be employed to map a membership
lustration, Tiwari et al.'s example described in (9a) function representing a fuzzy goal's importance to
and (9b) is still used. First the membership func- a real number in the range of ‰0; 1Š. The real
tions representing the fuzzy goals are determined number obtained can be considered as the desir-
using (4a) and/or (4b). Suppose the desirable able achievement degree for the fuzzy goal.
achievement degrees for the ®ve fuzzy goals are Thus, the application of desirable achievement
(0.8, 0.7, 0.5, 0.6, 0.8), respectively. The model is degrees to indicate the importance of fuzzy goals
formulated as is easy and ecient, and can produce satisfying
L.-H. Chen, F.-C. Tsai / European Journal of Operational Research 133 (2001) 548±556 553

solutions. However, the use of such an approach …l†pr ˆ …l †pr ; r ˆ 1; 2; . . . ; i 1;


may result in ``no feasible solutions'' when a de- ls 6 1;
cision-maker requires a very high desirable
achievement degree for each fuzzy goal. Under this x; lj P 0; j ˆ 1; . . . ; n;
circumstance, a compromise between goals should …11†
be made by the decision-maker. where …ls †pi represents the membership functions
of the goals in the ith priority level and …l †pr is the
achieved membership value in the rth priority le-
4. The model with preemptive priority
vel, r ˆ 1; 2; . . . ; i 1.
This study proposes a formulation to deal with
Decision-makers usually have a preemptive
such a problem; however, only one single problem
priority for achieving goals. That is, some goals
is necessary to be solved, no matter how many
have a higher priority for their achievement than
priority levels are decided. We incorporate the
the others under system constraints. To deal with
preemptive priority structure into this formulation
this problem, Hannan [13] adds achievement de-
to ®nd a set of solutions that not only satisfy the
grees for the fuzzy goals and their deviation vari-
desired structure, but also optimize the sum of
ables to a goal programming to ®nd the solutions
each fuzzy goal's achievement degree. To illustrate
that possibly satisfy the minimum achievement
the formulation, we still use Tiwari et al.'s [39]
degrees in sequence. Tiwari et al. [38,39] subdi-
example from the previous section to provide
vided such a problem into k subproblems, with k
comparable results. The ®ve fuzzy goals are
priority levels being set by a decision-maker. The
ranked into three priority levels:
desirable achievement degrees of the fuzzy goals
Priority level 1: G1 and G3,
belonging to the ®rst priority levels are achieved
Priority level 2: G2,
foremost. Then they are considered as additional
Priority level 3: G4 and G5.
constraints when solving the subproblems that
According to the above preemptive priority
contain the second-priority-level fuzzy goals. The
structure, this implies that the following relation-
k subproblems are resolved in the sequence of
ship exists for the respective achievement degrees
priority level. Although this approach can ®nd the
for the goals:
solutions that satisfy the preemptive priority
structure, k subproblems should be solved if k l1 P l2 ;
priority levels are considered. Chen [6] has modi- l3 P l2 ;
®ed Tiwari et al.'s [38] method to improve the l2 P l4 and
computational eciency for a FGP problem with l2 P l5 :
symmetrically triangular membership functions of
fuzzy goals and preemptive priority structure; Adding the above relationship to the problem,
however, the number of subproblems to be solved the FGP can be formulated as
is also equivalent to that of the priority levels. X5
For illustration, Tiwari et al. [39] have formu- Maximize li
lated the ith subproblem for the FGP with a pre- iˆ1
emptive priority structure using an additive model, subject to the membership functions
referring to the formulations in (3) and (4a), as …li ; i ˆ 1; . . . ; 5† of …9a†;
follows:
X the system constraints …9b†; …12†
Maximize f …l† ˆ …ls †pi l1 P l2 ;
s
l3 P l2 ;
Gs …x† Ls …12a†
subject to ls ˆ ; l2 P l4 ;
g s Ls
Ax 6 b; l2 P l5 ;
554 L.-H. Chen, F.-C. Tsai / European Journal of Operational Research 133 (2001) 548±556

The resulting achievement degrees for the ®ve are strongly required to be fully achieved, two
fuzzy goals are (1.0, 0.864, 0.864, 0.675, 0.807), additional constraints, l1 ˆ 1:0 and l3 ˆ 1:0,
respectively. Compared with Tiwari et al.'s solu- should be added to (12). In other words, they have
tions (1.0, 0.795, 1.0, 0.624, 0.727), except for G1, become crisp goals in this situation. Then the re-
the achievement degrees are not exactly the same. sulting achievement degrees are (1.0, 0.795, 1.0,
Instead of a partial achievement in (12), G3 is 0.624, 0.727), which are identical to Tiwari et al.'s
achieved fully in Tiwari et al.'s solution. However, solutions. Therefore, the solutions produced
the resulting achievement degrees from (12) com- by our approach also can be the same as that by
pletely satisfy the desired preemptive priority Tiwari et al.
structure, and the sum of achievement degrees
obtained is greater than Tiwari et al.'s.
Alternatively, the preemptive priority structure 5. Conclusion
can be incorporated into the problem in other
ways. First, the decision-makers can determine a This study has investigated FGP problems with
desirable achievement degree for each fuzzy goal di€erent importance levels and preemptive priori-
based on the preemptive priority structure, as de- ties. An additive model is suggested to maximize
scribed in the previous section. Fuzzy goals be- the total achievement degrees of fuzzy goals, be-
longing to the same priority level have an identical cause the achievement degree of each fuzzy goal
desirable minimum achievement degree. For ex- will decrease if a particular goal is dicult to
ample, the desirable minimum achievement de- achieve. Instead of the weighted additive model,
grees for the three priority levels in Tiwari et al.'s this study allows the decision-maker to determine
example could be determined as (0.9, 0.8, 0.65), explicitly a desirable achievement degree for each
respectively. The respective desirable achievement fuzzy goal as the importance of the fuzzy goal.
degree is then added to the formulation as con- This approach can ensure that the more important
straints. The problem is formulated as a fuzzy goal, the higher achievement degree it has,
even though a decision-maker may change the
X
5
relative importance of fuzzy goals. For dealing
Maximize li
iˆ1
with fuzzy ``weight (or importance)'', we suggest
that linguistic terms describing the importance of
subject to the membership functions
the fuzzy goals can be applied and that corre-
…li ; i ˆ 1; . . . ; 5† of …9a†; sponding real values are used as the achievement
the system constraints …9b†; degrees after employing some suitable ranking
…13†
l1 P 0:9; methods for fuzzy numbers.
To simplify the computational e€orts for FGP
l3 P 0:9;
with preemptive priority structure, this study for-
l2 P 0:8; mulates the decision-maker's structure into a sin-
l4 P 0:65; gle problem. We have shown the proposed
l5 P 0:65: approach's computational superiority over the
existing approaches. Although the resulting solu-
The achievement degrees for the ®ve fuzzy goals tions are not exactly the same as in previous
in the above problem are (1.0, 0.845, 0.9, 0.662, works, our approach still satis®es the preemptive
0.786), respectively, and all of them meet the re- priority structure and has the advantages of ob-
quirements. Furthermore, decision-makers cannot taining the maximum sum of achievement degrees.
only formulate preemptive priority structure for Furthermore, through the requirement of a mini-
the problem as in (12), but also can require mini- mum desirable achievement degree for the fuzzy
mum achievement degrees for some fuzzy goals goals belonging to a particular priority level, de-
belonging to the same priority level. For example, cision-makers can ®nd the desired achievement
if the ®rst-priority-level fuzzy goals, G1 and G3, degrees for the fuzzy goals.
L.-H. Chen, F.-C. Tsai / European Journal of Operational Research 133 (2001) 548±556 555

References [18] J.P. Ignizio, Goal Programming and Extension, Heath


Lexington Books, London, 1976.
[1] B. Aouni, O. Kettani, J.-M. Martel, Estimation through [19] M. Inuiguchi, Y. Kume, Goal programming problems with
imprecise goal programming model, in: R. Caballero, F. interval coecients and target intervals, European Journal
Ruiz, R.E. Steuer (Eds.), Advances in Multiple Objective of Operational Research 52 (3) (1991) 345±360.
and Goal Programming, Lecture Notes in Economics and [20] A.H. Kvanli, Financial planning using goal programming,
Mathematical Systems, No. 455, Springer, Berlin, 1997, pp. Omega 8 (1980) 207±218.
120±130. [21] S.M. Lee, Goal Programming for Decision Analysis,
[2] D. Bouyssou, Modelling inaccurate determination, uncer- Auerbach, Philadelphia, PA, 1972.
tainty, imprecision using multiple criteria, in: A.G. Locket, [22] T.-S. Liou, M.-J. Wang, Ranking fuzzy numbers with
G. Islei (Eds.), Improving Decision Making in Organiza- integral value, Fuzzy Sets and Systems 50 (1992) 247±255.
tions, Lecture Notes in Economics and Mathematical [23] J.-M. Martel, B. Aouni, Incorporating the decision-
Systems 335, Springer, Berlin, 1989, pp. 78±87. maker's preferences in the goal programming model,
[3] E.K. Can, M.H. Houck, Real-time reservoir operations by Journal of Operational Research Society 41 (12) (1990)
goal programming, Journal of Water Resources Planning 1121±1132.
Management 110 (1984) 297±309. [24] J.-M. Martel, B. Aouni, Diverse formulations of the
[4] G.A. Chalam, Fuzzy programming (FGP) approach to a imprecise goal programming model, Journal of Global
stochastic transportation problem under budgetary con- Optimization 12 (1998) 127±138.
straint, Fuzzy Sets and Systems 66 (3) (1994) 293±299. [25] J.-M. Martel, B. Aouni, Incorporating the decision-
[5] A. Charnes, W.W. Cooper, Management Models and maker's preferences in the goal programming model with
Industrial Applications of Linear Programming, Wiley, fuzzy goal values: A new formulation, in: M. Tamiz (Eds.),
New York, 1961. Multi-objective Programming and Goal Programming,
[6] H.-K. Chen, A note on a fuzzy goal programming Lecture Notes in Economics and Mathematical Systems,
algorithm by Tiwari Dharmar and Rao, Fuzzy Sets and Springer, Berlin, 1996, pp. 257±269.
Systems 62 (1994) 287±290. [26] H. Min, The dynamic expansion and relocation of capac-
[7] S.-H. Chen, Ranking fuzzy numbers with maximizing set itated public facilities: A multi-objective approach, Com-
and minimizing set, Fuzzy Sets and Systems 17 (1985) 113± puters and Operations Research 15 (3) (1988) 243±252.
129. [27] H. Min, A model-based decision support system for
[8] F. Choobineh, H. Li, An index for ordering fuzzy locating banks, Information and Management 17 (4)
numbers, Fuzzy Sets and Systems 54 (1993) 287±294. (1989) 207±215.
[9] A.K. Dhingra, S.S. Rao, H. Miura, Multiobjective decision [28] R. Narasimhan, Goal programming in a fuzzy environ-
making in a fuzzy environment with applications to ment, Decision Sciences 11 (1980) 325±336.
helicopter design, AIAA Journal 28 (4) (1990) 703±710. [29] J.B. Pickens, J.G. Hof, Fuzzy goal programming in
[10] A.K. Dhingra, S.S. Rao, V. Kumar, Nonlinear member- forestry: An application with special solution problems,
ship functions in multiobjective fuzzy optimization of Fuzzy Sets and Systems 39 (3) (1991) 239±246.
mechanical and structural systems, AIAA Journal 30 (1) [30] J.R. Rao, R.N. Tiwari, B.K. Mohanty, A preference
(1992) 251±260. structure on aspiration levels in a goal programming
[11] P. Fortemps, M. Roubens, Ranking and defuzzi®cation problem ± a fuzzy approach, Fuzzy Sets and Systems 25
methods based on area compensation, Fuzzy Sets and (1988) 175±182.
Systems 82 (1996) 319±330. [31] S.S. Rao, Description and optimum design of fuzzy
[12] M. Gen, K. Ida, Y. Tsujimura, C.E. Kim, Large-scale 0±1 mechanical systems, ASME Journal of Mechanisms,
fuzzy goal programming and its applications to reliability Transmissions and Automation in Design 109 (1) (1987)
optimization problem, Computers and Industrial Engi- 126±132.
neering 24 (4) (1993) 539±549. [32] S.S. Rao, Multi-objective optimization of fuzzy structural
[13] E.L. Hannan, Linear programming with multiple fuzzy systems, International Journal for Numerical Methods in
goals, Fuzzy Set and Systems 6 (1981) 235±248. Engineering 24 (6) (1987) 1157±1171.
[14] E.L. Hannan, On fuzzy goal programming, Decision [33] S.S. Rao, K. Sundararaju, B.G. Prakash, C. Balakrishna,
Sciences 12 (3) (1981) 522±531. Fuzzy goal programming approach for structural optimi-
[15] E.L. Hannan, Some further comments on fuzzy priorities, zation, AIAA Journal 30 (5) (1992) 1425±1432.
Decision Sciences 12 (1981) 539±546. [34] C. Romero, A note: E€ects of ®ve-side penalty functions in
[16] E.L. Hannan, Contrasting fuzzy goal programming and goal programming, Omega 12 (1984) 333.
fuzzy multicriteria programming, Decision Science 13 (2) [35] P.A. Rubin, R. Narasimhan, Fuzzy goal programming
(1982) 337±339. with nested priorities, Fuzzy Sets and Systems 14 (1984)
[17] J.P. Ignizio, Notes and communications on the (re)discov- 115±129.
ery of fuzzy goal programming, Decision Sciences 13 (2) [36] S.B. Sinha, K.A. Rao, B.K. Mangaraj, Fuzzy goal
(1982) 331±336. programming in multi-criteria decision systems: A case
556 L.-H. Chen, F.-C. Tsai / European Journal of Operational Research 133 (2001) 548±556

study in agricultural planning, Socio-Economic Planning [40] T. Yang, J.P. Ignizio, H.-J. Kim, Fuzzy programming with
Science 22 (2) (1988) 93±101. nonlinear membership functions: Piecewise linear approx-
[37] S. Suzuki, T. Yoshizawa, Multiobjective trajectory opti- imation, Fuzzy Sets and Systems 11 (1991) 39±53.
mization by goal programming with fuzzy decisions, [41] M. Zeleny, The pros and cons of goal programming,
Journal of Guidance, Control, and Dynamics 17 (2) Computers and Operations Research 8 (1981) 357±359.
(1994) 297±303. [42] M. Zeleny, Multiple Criteria Decision Making, McGraw-
[38] R.N. Tiwari, S. Dharmar, J.R. Rao, Priority structure in Hill, New York, 1982.
fuzzy goal programming, Fuzzy Sets and Systems 19 (1986) [43] H.J. Zimmermann, Fuzzy programming and linear pro-
251±259. gramming with several objective functions, Fuzzy Sets and
[39] R.N. Tiwari, S. Dharmar, J.R. Rao, Fuzzy goal program- Systems 1 (1978) 45±55.
ming ± an additive model, Fuzzy Sets and Systems 24 [44] H.J. Zimmermann, Fuzzy mathematical programming,
(1987) 27±34. Fuzzy Sets and Systems 10 (4) (1983) 291±298.