Professional Documents
Culture Documents
Fuzzy Controls
Fuzzy Set A
Fuzzy Boundary
0
x
Fuzzy Fuzzy
Note: A crisp set is a special case of fuzzy set, where the membership function
(characteristic function) can take only two values 1 (membership) and 0 (non-membership)
SYMBOLIC REPRESENTATION AND EXAMPLES
A x A (x)
For discrete universe with elements xi :
A A x1 / x1 A x2 / x2 A xi / xi A xi
or A
xi X xi
A x
For continuous universe: A xX
x
Note: both series and integral are “symbolic” shorthand forms, not actual operation)
Example : A = 0.2/3 + 0.3/4 + 1.0/5 + 0.2/6 + 0.1/7
Fuzzy Labels: Linguistic terms "tall men," "beautiful women," and "fast cars are (represented by
fuzzy sets; their membership is subjective, not crisp).
Fuzzy Descriptors: Fuzzy terms “tall,” “beautiful,” and “fast”
Note: They are also linguistic variables, which take fuzzy values.
Linguistic Hedge: Fuzzy adjective “very” in linguistic variable “very tall.” Other Examples:
slightly, extremely, more or less, and highly.
FUZZY LOGIC OPERATIONS
Bivalent logic operations NOT, OR, and AND may be extended to fuzzy logic.
Fuzzy Implication needs particular attention.
Applicable connectives must be expressed in terms of set membership functions
Note: Isomorphism between fuzzy sets and fuzzy logic
Complement (Negation, NOT)
Fuzzy set A in universe X. Its complement A’ is a fuzzy set
A’
1.0
Hot
0 10 20 30 40
Temperature (C)
UNION (DISJUNCTION, OR)
Two fuzzy sets A and B in the same universe X.
1 A
B
AB
Temperature (C)
16 18 20 22 24 26
0 0.4 0.8 1.0 1.0 0.8 0.5
Relative 20 0.8 0.8 1.0 1.0 0.8 0.8
Humidity 40 1.0 1.0 1.0 1.0 1.0 1.0
(%) 60 0.6 0.8 1.0 1.0 0.6 0.6
80 0.4 0.8 1.0 1.0 0.8 0.8
INTERSECTION (CONJUNCTION, AND)
Two fuzzy set A and B in the same universe X.
Their intersection AB is a fuzzy set whose membership function is
A B ( x) min[( A ( x), B ( x)] x X
Corresponds to fuzzy-logic AND (Conjunction), denoted by A B, where A and B are
fuzzy states or fuzzy propositions
Rationale for min: Because element x must simultaneously belong to both sets, the
smaller possibility (membership grade) should govern the outcome.
Note: Consistent with intersection of crisp sets, using characteristic functions
(Note: min (1,0) = 0).
1 A
B
AB
0
ROBOTICS EXAMPLE
(b) Membership
(a) Membership Grade
Grade
1
1
0
0 0.0 10.0 20.0 Power (hp)
0.0 1.0 2.0 Speed (m/s)
(c) Membership
Grade
Speed
1
2.0
0.0
(a) Required speed; (b) Required power; (c) Required speed and power.
PROPERTIES OF FUZZY SETS
A’
Note: AA’ X
AA
0
Example: Let A denote the set of “my true statements.”
Complement of A ( A’) denotes the set of “my false statements.”
Universe = set of “all my statements,”
Logic proposition “All my statements are false:” A A ' A ' (a)
According to bivalent crisp logic this is a contradiction.
Proof: A A' X (for crisp logic) => From proposition (a): A' X => A .
But, if (a) is true, then A => contradiction
GENERALIZED FUZZY COMPLEMENT
Generalized complement C should satisfy the following axioms:
1. Boundary conditions: C() = X and C(X) = (Note X 1 ; 0).
2. Non-increasing: For two fuzzy sets with a = A(x) and b = B(x) in the same universe X,
if a b, then C(a) C(b) (or, if A B then C(A) C(B)).
3. Involutive: C(C(A)) = A (analogous to double negation)
Note: Special case of complement (1-) satisfies these.
Note: Complement is applied to a membership function and it generates another
membership function: C: [0,1] [0,1]
1 (1 y) if x y y if x y
Denoted by A B.
Grade of Inclusion (Partial inclusion) is given by membership function of A B:
A B ( x) 1 if A ( x) B ( x)
A ( x)T B ( x) otherwise
Note: sup denotes supremum operation. This obtains the maximum value of a function over a
continuous (or, piecewise continuous) interval of its variable.
In the above equation, c is varied from 0 to 1 continuously, and the maximum value of c
in this interval that satisfies the condition A ( x)Tc B ( x) , is used.
SET EQUALITY (A = B)
Special case of set inclusion.
Grade of Equality
This is the membership function of the fuzzy relation A = B. Specifically,
A B ( x ) 1 if A ( x ) B ( x )
A ( x)T B ( x) otherwise
Note: In a specific application use either min or product to represent the t-norm.
FUZZY IMPLICATION (IF-THEN)
Fuzzy set A in universe X and fuzzy set B in universe Y.
Fuzzy Implication “If A then B” is denoted by A B;
defined in Cartesian product space XY.
Fuzzy logic knowledge base uses if-then rules.
A fuzzy rule is a fuzzy relation.
A knowledge base (several fuzzy rules) is also a relation, formed by
combining (aggregating) rules according to how they are interconnected.
Useful Fuzzy Implications:
1. Larsen : AB ( x, y) A ( x) B ( y)
2. Mamdani: AB ( x, y) min[ A ( x), B ( y)]
Support Set
Crisp set containing all the elements (in the universe) whose membership grade is
greater than 0. S {x X | A ( x) 0}
Note: Each fuzzy state will have a modal point in membership function
Number of such modal points = number of fuzzy states fuzzy resolution
DEGREE OF FUZZINESS
Difficulty of ascertaining element membership.
Note: Most fuzzy elements are elements with membership grade is 0.5.
More Fuzzy
0.5
Less Fuzzy
0
FUZZY RELATION
A relation is a mathematical description of a situation where certain elements of sets are related to one another in
some way.
A membership function represents a fuzzy relation in its universe (or, domain, or space) of definition
Example: Two universes X1 ={x1} and X2 ={x2}.
Crisp Relation R: Crisp set consisting of a subset of ordered pairs (x1, x2) in 2-D Cartesian product space X1 X2.
Let X1 = {1,2,3} and X2 = {2,3,4}. Then the Cartesian product of X1 and X2 isthe set X1x X2 = {(1,2), (1,3), (1,4),
(2,2), (2,3), (2,4), (3,2), (3,3), (3,4)).
Fuzzy Relation R: Ordered pairs (x1, x2) with membership grades R(x1, x2) attached to them, in X1 X2.
Note: Can be extended to fuzzy relation in the n-dimensional Cartesian space X1 X2 … Xn.
Let X1 = {Peshawar, Karachi, Islamabad} and X2= {Mardan, Karachi}. We want to define the relational concept
“very far" between these two sets of cities. If we use a number in the interval [0,1] to represent the degree of
“very far," then the concept "very far" may be represented by the following (fuzzy) relational matrix:
CARTESIAN PRODUCT OF FUZZY SETS
Cartesian product A1 A2 of crisps sets
A1 and A2 in universes X1 and X2:
Rectangular area in Figure (a).
Subset of Cartesian product space X1 X2
(entire 2-D space or plane containing the
two axes x1 and x2)
x1 X1 , x2 X 2
Sup-Product Composition
Use the “product” operation in place of “min” for intersection (join):
( RS ) sup [ R S ]
Xm , , Xr
This composition is denoted by " " and is also known as the sup-dot composition.
0 0 0 0
0 0 0 0
Sup-dot composition: 0 0 0 0.8 0 0 0 0 0 0.08, 0.56, 0.64, 0.32
0.1 0.7 0.8 0.4
0 0 0 0
EXAMPLE
A fuzzy rulebase (relation) P( x, y ) is used to make an inference B( y ) from a context A(x ) .
From this information, derive a suitable rulebase R ( x, z ) that may be used to make an inference
C (z ) from context A( x ) .