You are on page 1of 21

Classical Linear Programming model

Maximize (Minimize) Z = cjxj


Subject to
j=1 n

,m

aijxj = bi
j=1

i= 1,2,

Case I :Fuzzification of Objective Function (Z fuzzification)


MAX = 0.5*X1 + 0.2*X2; X1 + X2 < 300,400; 2*X1 + X2 < 400,500;
Fuzzification of Constraints (fuzzification of Bi)

Ref: Fuzzy sets and fuzzy logic theory and applications by Klir and Yaun

Crisp max problem


MAX = 0.5*X1 + 0.2*X2; X1 + X2 < 400; 2*X1 + X2 < 500;

Answer Z= 125 X1 = 250 X2=0

Crisp min problem


MAX = 0.5*X1 + 0.2*X2; X1 + X2 < 300; 2*X1 + X2 < 400; Answer Z=100 X1=200 X2=0

Formulation
Max = lamda ( satisfaction level) Subject to (Z - Zmin)/(Zmax Zmin) > lamda and crisp constraints ( as original problem)

Fuzzified problem
MAX = lamda; ((0.5*X1 + 0.2*X2) - 100) / ( 125 - 100 ) > lamda; X1 + X2 < 400; 2*X1 + X2 < 500; Lamda = satisfaction level Lamda = 1 Z=125 X1=250 X2=0

Satisfaction level

1.00

0.0 0

100

125

objective function value

Case II: Fuzzification of Constraints (fuzzification of Bi)


MAX = lamda; ((0.5*X1 + 0.2*X2) - 100)/ ( 125-100) > lamda ; (400- (X1 + X2))/(400-300) > lamda; (500 - (2*X1 + X2))/(500-400) > lamda;

Fuzzification of Objective Function (Z fuzzification)

Result Lamda = 0.5 Zopt = 112.5 X1 = 225 X2 = 0

Case III : Fuzzification of Technological Coefficients (Fuzzification of Aij)


Max = 2*x1 + 3* x2 Subject to (1,1) x1 + (2,3) x2 < 4 (3,2) x1 + (1,3) x2 < 6

Ref: Rafail Gasiman and Kursat Yenilmez (2002), " Solving fuzzy linear programming problems with linear membership function"Turk J Math 26(2002),pp 375-396.

Crisp1_problem
MAX = 2*X1 + 3*X2; X1 + 2*X2 < 4; 3*X1 + X2 < 6; Z1 = 6.8 X1 =1.6 X2 = 1.2

Crisp2_problem
MAX = 2*X1 + 3*X2; 2*X1 + 5*X2 < 4; 5*X1 + 4*X2 < 6;

Z2=3.0588 X1=0.8235 X2= 0.471

Formulation
Max = lamda (Z- Zmin)/(Zmax-Zmin) > lamda (Bi AijXij)/ dijXij > lamda

Fuzzified problem
Max = lamda Subject to (2*X1+3*X2 - 3.0588)/(6.8 - 3.0588) > lamda; (4 - (X1 + 2*X2))/(X1+3*X2) > lamda; (6 - (3*X1 + X2))/(2*X1 + 3*X2) > lamda;

This formulation is nonlinear we can make it linear by putting Lamda = value (between 0 to 1) By using bisection method we can find the maximum value of lamda i.e. maximum satisfaction level (after 20 iterations)
Lamda = 0.39737 X1 =1.147 X2= 0.7506 Zopt = 4.5458

Case IV : Fuzzification of Technological coefficients and resources (fuzzification of Aij and Bi) MAX = X1 + X2;

(1,1)X1 + (2,1)X2 < (3,2); (2,2)X1 + (3,2)X2 < (4,3);

Crisp1_problem
MAX = X1 + X2; 2*X1 + 3*X2 < 3; 4*X1 + 5*X2 < 4;

Z1= 1 X1 = 1 X2 = 0

Crisp2_problem
MAX = X1 + X2; X1 + 2*X2 < 5; 2*X1 + 3*X2 < 7;

Z2 =3.5 X1=3.5 X2=0

Formulation
Max = lamda (Z- Zmin)/(Zmax-Zmin) > lamda (Bi AijXij)/ dijXij > lamda

Fuzzified problem
MAX = lamda; (X1+X2 - 1)/(3.5 - 1) > lamda; (3 - (X1 + 2*X2))/(X1+X2) > lamda; (4 - (2*X1 + 3*X2))/(2*X1 + 2*X2) > lamda;

Using bisection method we get (after 14 iterations) Lamda = .2433 X1 = 1.6083 X2 = 0 Zopt = 1.6083

You might also like