Fuzzy Numbers

Definition
Fuzzy Number
Convex and normal fuzzy set defined on R
Equivalently it satisfies
Normal fuzzy set on R
Every alpha-cut must be a closed interval
Support must be bounded
Applications of fuzzy number
Fuzzy Control, Decision Making, Optimizations
 Examples

1 1
. .

3 1.5 3 4.5
. .
1 1
. .

1.5 3 4.5 1.5 3 4.5

. .
 Arithmetic Operations
Interval Operations A = [ a1 , a3 ] , B = [ b1 , b 3 ]
Addition
[ a1 , a3 ]( )[b1 , b3 ] [ a1 b1 , a3 b3 ]
Subtraction
[ a1 , a3 ]()[b1 , b3 ] [ a1 b3 , a3 b1 ]
Multiplication
[ a1 , a3 ]()[b1 , b3 ] [ a1 b1 a1 b3 a3 b1 a3 b3 ,
a1 b1 a1 b3 a3 b1 a3 b3 ]
Division
[ a1 , a3 ](/)[b1 , b3 ] [ a1 / b1 a1 / b3 a3 / b1 a3 / b3 ,
a1 / b1 a1 / b3 a3 / b1 a3 / b3 ]
execpt b1 b3 0
 Examples
Addition
[2,5]+[1,3]=[3,8] [0,1]+[-6,5]=[-6,6]
Subtraction
[2,5]-[1,3]=[-1,4] [0,1]-[-6,5]=[-5,7]
Multiplication
[-1,1]*[-2,-0.5]=[-2,2] [3,4]*[2,2]=[6,8]
Division
[-1,1]/[-2,-0.5]=[-2,2] [4,10]*[1,2]=[2,10]
Properties of Interval Operations
Commutative
A B B A A B B A
Assocoative
( A B) C A ( B C ) ( A B) C A ( B C )
Identity 0 [0,0] 1 [1,1]
A A0 0 A A A 1 1 A
Subdistributive
A (B C) A B A C
Inverse
0 A-A 1 A/A
Monotonicity for any operations *
If A E and B F then A*B E*F
 Arithmetic Operation on Fuzzy
Numbers
Interval operations of alpha-level sets

( A * B) A* B for any (0,1].
When * /, 0 B for all (0,1].

A* B ( A * B)
(0,1]

Note: The Result is a fuzzy number.

Example: See Text pp. 105 and Fig. 4.5
Arithmetic Operation by Extension
Principle
By Extension Principle
( A * B )( z ) sup min[ A( x), B ( y )]
z x* y

Note:
* can be any operations including arithmetic
operations.
Example A = { 1/2 , 0.5/3}, B = { 1/3, 0.8/4}
 Example
A+B = {1/5, 0.8/6, 0.5/7 }
i) z 5 No such case.
A( ) B ( z ) 0
ii ) z 5
x y 23
A (2) B (3) 1
iii ) z 6
x y 33 or x y 24
A (3) B (3) 0.5
A (2) B ( 4) 0.8 A( ) B (6) ( 0.5, 0.8 ) 0.8
6 3 3
62 4

iv ) z 7
A (3) B (4) min ( 0.5, 0.8 ) 0.5
 Example
Max (A,B) = { (3 , 1) , (4 , 0.5) }
i) z 2 No such case.
A( ) B ( z ) 0
ii ) z 3
x y 2 3 or x y 3 3
A ( 2) B (3) 1 1 1 A (3) B (3) 0.5 1 0.5
A( ) B (3) (1,0.5) 1
3 2 3
3 3 3

iii ) z 4
x y 2 4 or x y 3 4
A ( 2) B ( 4) 1 0.5 0.5 A (3) B ( 4) 0.5 0.5 0.5
A( ) B ( 4) (0.5,0.5) 0.5
4 2 4
4 3 4

iv ) z 5 No such case.
A( ) B ( z ) 0
 Typical Fuzzy Numbers
Triangular Fuzzy Number
Fig. 4.5
Trapezoidal Fuzzy Numbers: Fig. 4.4
Linguistic variable: Performance
Linguistic values (terms):
very small very large
Semantic Rules:
Terms maps on trapezoidal fuzzy numbers
Syntactic rules (grammar):
Rules for other terms such as not small
An Application of Optimal Decision
Decision Making
Which alternative are you going to choose?
weighted average.
How to make use of resulted fuzzy set
Similarity comparisons with model fuzzy sets

Criterion 1 Criterion 2 Criterion 3

Alternative 1 Low High High
Alternative 2 High High Low
Alternative 3 Low High Medium
Weight of Attribute high medium Low
 Lattice of Fuzzy Numbers
Lattice
Partially ordered set with ordering relation
Meet(g.l.b) and Join(l.u.b) operations
Example:
Real number and less than or equal to
Lattice of fuzzy numbers
MIN ( A, B)( z ) sup min[ A( x), B ( y )] MEET ( A, B)
z min( x , y )

MAX ( A, B )( z ) sup min[ A( x), B ( y )] JOIN ( A, B)

z max( x , y )
 Lattice of Fuzzy Numbers
Distributive lattice

MIN[A,MAX(B,C)]=MAX[MIN[A,B],MIN[A,C]]

MAX[A,MIN(B,C)]=MIN[MAX[A,B],MAX[A,C]]
, MIN, MAX or , is a distributive lattice,
where is the set of all fuzzy numbers.

A B iff MIN( A, B) A
A B iff MAX( A, B ) B

Example: See Fig. 4.6

Example: very small <= small <= <= very large

 Fuzzy Equations
Addition A X B
X = B-A is not a solution because A+(B-A) is
not B.
Conditions to have a solution
For any (0,1], let A [ a1 , a2 ], B [ b1 , b2 ] and X [ x1 , x2 ]
(i) b1 a1 b2 a2 for any (0,1]
(ii) implies b1 a1 b1 a1 b2 a2 b2 a2

Suppose X [ b1 a1 , b2 a2 ] is a solution of
Solution
A X B for any (0,1]. Then
X X
(0,1]
 Fuzzy Equations
Multiplication A X B
X = B/A is not a solution.
Conditions to have a solution
For any (0,1], let A [ a1 , a2 ], B [ b1 , b2 ] and X [ x1 , x2 ]
(i) b1 / a1 b2 / a2 for any (0,1]
(ii) implies b1 / a1 b1 / a1 b2 / a2 b2 / a2

Solution Suppose X [ b1 / a1 , b2 / a2 ] is a solution of

A X B for any (0,1]. Then
X X
(0,1]