Research Article  Open Access
A MOLP Method for Solving Fully Fuzzy Linear Programming with LR Fuzzy Parameters
Abstract
Kaur and Kumar, 2013, use Meharâ€™s method to solve a kind of fully fuzzy linear programming (FFLP) problems with LR fuzzy parameters. In this paper, a new kind of FFLP problems is introduced with a solution method proposed. The FFLP is converted into a multiobjective linear programming (MOLP) according to the order relation for comparing the LR flat fuzzy numbers. Besides, the classical fuzzy programming method is modified and then used to solve the MOLP problem. Based on the compromised optimal solution to the MOLP problem, the compromised optimal solution to the FFLP problem is obtained. At last, a numerical example is given to illustrate the feasibility of the proposed method.
1. Introduction
The research on fuzzy linear programming (FLP) has risen highly since Bellman and Zadeh [1] proposed the concept of decision making in fuzzy environment. The FLP problem is said to be a fully fuzzy linear programming (FFLP) problem if all the parameters and variables are considered as fuzzy numbers. In recent years, some researchers such as Lofti and Kumar were interested in the FFLP problems, and some solution methods have been obtained to the fully fuzzy systems [2â€“4] and the FFLP problems [5â€“13]. FFLP problems can be divided in two categories: (1) FFLP problems with inequality constraints; (2) FFLP problems with equality constraints. If the FFLP problems are classified by the types of the fuzzy numbers, they will include the next three classes: (1) FFLP problems with all the parameters and variables represented by triangular fuzzy numbers; (2) FFLP problems with all the parameters and variables represented by trapezoidal fuzzy numbers; (3) FFLP problems with all the parameters and variables expressed by fuzzy numbers (or flat fuzzy numbers).
Fuzzy programming method is a classical method to solve multiobjective linear programming (MOLP) [14, 15]. In this paper, the fuzzy programming method is modified and then used to obtain a compromised optimal solution of the MOLP. The modified fuzzy programming method is shown in Steps 4â€“10 of the proposed method in Section 3.
Dehghan et al. [2â€“4] employed several methods to find solutions of the fully fuzzy linear systems. Hosseinzadeh Lotfi et al. [6] used the lexicography method to obtain the fuzzy approximate solutions of the FFLP problems. Allahviranloo et al. [7] and Kumar et al. [5, 8] solved the FFLP problem by use of a ranking function.
Fan et al. [12] adopted the cut level to deal with a generalized fuzzy linear programming (GFLP) probelm. The feasibility of fuzzy solutions to the GFLP was investigated and a stepwise interactive algorithm based on the idea of design of experiment was advanced to solve the GFLP problem.
Kaur and Kumar [9] introduced Meharâ€™s method to the FFLP problems with fuzzy parameters. They consider the following model: where the parameters and variables are flat fuzzy numbers and the order relation for comparing the numbers is defined as follows.(i) if and only if ,(ii) if and only if ,(iii) if and only if .Here and are two arbitrary flat fuzzy numbers.
In our study, we consider a new kind of FFLP problems with flat fuzzy parameters as follows: where the parameters and variables are flat fuzzy numbers and the order relation shown in Definition 4 is different from the one above.
In this paper, we modify the classical fuzzy programming method. The FFLP is changed into a MOLP problem solved by the modified fuzzy programming method. We get the compromised optimal solution to the MOLP and generate the corresponding compromised optimal solution to the FFLP.
The rest of the paper is organized as follows. In Section 2, the basic definitions and the FFLP model are introduced. In Section 3, we propose a MOLP method to solve the FFLP problems. Some results are discussed from the solutions obtained by the proposed method. In Section 4, a numerical example is given to illustrate the feasibility of the proposed method. In Section 5, we show some short concluding remarks.
2. Preliminaries
2.1. Basic Notations
Definition 1 ( fuzzy number, see [2]). A fuzzy number is said to be an fuzzy number if where is the mean value of and and are left and right spreads, respectively, and function means the left shape function satisfying(1);(2) and ;(3) is nonincreasing on .
Naturally, a right shape function is similarly defined as .
Definition 2 ( flat fuzzy number, see [9, 16]). A fuzzy number , denoted as , is said to be an flat fuzzy number if its membership function is given by
Definition 3 (see [5, 9]). An flat fuzzy number is said to be nonnegative flat fuzzy number if and is said to be nonpositive flat number if .
We define as an fuzzy number with membership function and denote as .
2.2. Arithmetic Operations
Let and be two flat fuzzy numbers, . Then the arithmetic operations are given as follows [9, 16]:
It is easy to verify that the operator satisfies associative law. Hence, the formula is reasonable, where are flat fuzzy numbers.
2.3. Order Relation for Comparing the LR Flat Fuzzy Numbers
For comparing the flat fuzzy numbers, we introduce the order relation as follows.
Definition 4. Let and be any flat fuzzy numbers. Then(i) if and only if , , , ;(ii) if and only if , , , ;(iii) if and only if , , , .
Based on the definition of order , we may obtain that (i) is nonnegative if and only if ; (ii) is nonpositive if and only if .
The following propositions are given to show the properties of the order relation defined above.
Proposition 5. Let be four arbitrary flat fuzzy numbers and an arbitrary real number. Then
Proof. Suppose , , , and .
(1) It is obvious that , . Since , we get
So
This indicates that .
(2) It is clear that , . From , we get
Therefore,
for , and
for . This indicates that
Proposition 6. Let be three arbitrary flat fuzzy numbers. Then(1);(2), ;(3), .
Proof. Suppose , , and .
(1) Obviously, ; hence, we have .
(2) Since , we get
This means
That is
Therefore, we have .
(3) From , , we get
This indicates
Therefore, we have .
From Proposition 6, we know that the order relation is a partial order on the set of all fuzzy numbers.
2.4. Fully Fuzzy Linear Programming with LR Fuzzy Parameters
In this paper, we will consider the following model; that is, or where , , , and represent fuzzy matrices and vectors and , , , and are flat fuzzy numbers. The order relations for comparing the flat fuzzy numbers both in the objective function and the constraint inequalities are as shown in Definition 4.
3. Proposed Method
Steps of the proposed method are given to solve problem (20) as follows. This method is applicable to minimization of FFLP problems, and the solution method of maximization problems is similar to that of minimization ones.
Step 1. If all the parameters are represented by flat fuzzy numbers , , , and , then the FFLP (20) can be written as
Step 2. Calculate and , respectively, and suppose that and ; then the FFLP problem obtained in Step 1 can be written as
Step 3. According to the order relation defined above, the problem obtained in Step 2 is equivalent to
We denote , , , , and satisfies the constraints of programming (23)}. Programming (23) may be written as the programming (24), below for short, as follows: Obviously, programming (24) is a crisp multiobjective linear programming problem. In fact we have .
Step 4. Solve the subproblems where . We find optimal solutions , , , and , respectively. And the corresponding optimal values will be , , , and .
Step 5. Let , , and the membership function of is given by where .
Step 6. Let ; the MOLP problem obtained in Step 3 can be equivalently written as Suppose is one of the optimal solutions (if there exits only one optimal solution, is the unique one), and is the optimal objective value (in fact, the optimal solution should be written as . Since is an auxiliary variable, we denote as for simplicity). Then for at least one in . ( is an arbitrary element in the set ).
Step 7. Let , and solve the following crisp programming: If is one of the optimal solutions and is the optimal objective value, then for at least one in .
Step 8. Let , and solve the following crisp programming: Suppose is one of the optimal solutions and is the optimal objective value. Then for at least one in .
Step 9. Let , and solve the following crisp programming: Suppose is one of the optimal solutions, and is the optimal objective value. Then with in .
Step 10. Take as the compromised optimal solution to programming (23) and generate the compromised optimal solution to programming (21) by . Assuming we may obtain and the corresponding objective value .
Remark 7 (). Some properties of the solutions obtained in Steps 6â€“10 are shown in the following proposition.
Proposition 8. Suppose ,â€‰â€‰, (), and are the notations described in Steps 1â€“10; then
Proof. (1) From the results of Steps 6â€“9, it is obviously clear that
and with . Since is an optimal solution to programming (30), we know that satisfies the constraints of programming (30), and so , , and .
(2) In fact, is an optimal solution to programming (27); therefore, it is a feasible solution. We have
and it is obvious that from the result of Step 6. Hence, is a feasible solution to programming (28). The objective value of is , and the optimal objective value of programming (28) is ; so we get . It is similar to prove and .
4. Numerical Example
In this section, we present a numerical example to illustrate the feasibility of the solution method proposed in Section 3.
We aim to find the compromised optimal solution and corresponding objective value of the following fully fuzzy linear programming problem: where and .
According to Steps 1 and 2 in the proposed method, we obtain the following programming:
By Step 3, the programming above is transformed into the following programming: Programming (38) can be abbreviated to the following programming: where .
Solve the following subproblems: respectively, and we obtain the optimal objective value and one of the optimal solutions as shown in Table 1.

According to , we acquire the lower objective values , , , and , with corresponding membership functions given below. Consider
The optimal objective value is , and one of the optimal solutions is .
Calculate the value of the membership function of () at , and we get , , , and .
Solve the following problem:
The optimal objective value is , and one of the optimal solutions is .
Calculate the value of the membership function of () at , and we get , , , and .
Solve the following problem:
The optimal objective value is , and one of the optimal solutions is .
Calculate the value of the membership function of at , and we get , , , and .
Solve the following problem:
The optimal objective value is , and one of the optimal solutions is .
Calculate the value of the membership function of () at , and we get , , , and .
Generate by (), following Step 10, and calculate the value of