EE-873: Fuzzy Control

Fuzzy Inference System

Dr. Sohail Iqbal

1
Contents
• Fuzzy Inference System (FIS)

• Fuzzification with examples

• Fuzzy Rule Base

• Inference Mechanism

• Defuzzification with applications

2
Fuzzy Inference System (FIS)
• Fuzzy Inference Systems take inputs and process
them based on the specified fuzzy rules to produce
the outputs.

3
Fuzzification
• Fuzzification is the process of making a crisp
quantity fuzzy by translating crisp value into a
membership function

• Define Linguistic variables and terms

• Construct the membership function
• Convert crisp input data to fuzzy values using
membership functions

4
Linguistic Variables
• Input/Output variables of system defined in natural
language
• Linguistic variables are decomposed into linguistic
terms that cover a range on the universe of
discourse
• e.g. Room Temperature

5
Selecting Membership Functions
• Used to map non-fuzzy input values to fuzzy
linguistic terms and vice versa
• Used during the process of Fuzzification and
Defuzzification

• Note: Values can belong to multiple sets

6
Types of Membership Functions
• Piece-wise linear functions
• Gaussian distribution function
• Sigmoid curve
• Quadratic and cubic polynomial curves

• Note: Non-linear membership function increase

computation time, because of this linear
membership functions are commonly used
7
Triangular Membership Function
• A triangular MF is specified by three parameters {a,
b, c} as follows:
• triangle (x; 20, 60, 80)

8
Trapezoidal Membership
Function
• A trapezoidal MF is specified by four parameters {a,
b, c, d} as follows:
• trapezoid (x; 10, 20, 60, 95)

9
Other Membership Functions
(MF)

10
The Basic Tipping Problem
• An engineer visits a restaurant for meal, in order to
automatically calculate the value of tip he design a
fuzzy controller.
• Although different input parameters can be
considered, he designs the controller for two inputs
service and food.

11
Fuzzy Inference System

Fuzzy Inference System for Tipping Problem

12
Tipping Example: Database

13
Tipping Example: Database

14
Tipping Example: Database

15
Tipping Example: Fuzzification
• According to customer, rating for food is 8.

16
Rule Base
• Set of rules that capture expert’s knowledge to
control the output of system
• IF-THEN rules with a condition and a conclusion
General Form: IF antecedent THEN consequent
• Antecedents are associated with fuzzy controller
input whereas consequent are the associated
actions or output of the fuzzy controller system
• Write down rules for all possible cases or total
combinations

17
Rule Base

• Fuzzy Rules for Basic Tipping Example

1 IF (service is poor) OR (food is rancid) THEN tip is cheap

2 IF (service is good) THEN tip is average
3
IF (service is excellent) OR (food is delicious) THEN tip is
generous

18
Rule Base
• If the antecedent of a rule has more than one part,
the fuzzy operator is applied to obtain one number
that represents the result of the rule antecedent.

• AND and OR logical operations are used.

19
Tipping Example: Apply Fuzzy
Operator

20
Inference Mechanism
• After evaluating result of each rule, result is
combined to obtain the final result through process
called inference.

21
Types of Fuzzy Inference System
• Mamdani Fuzzy Models

• Sugeno Fuzzy Models (Also known as Takagi-Sugeno

or TSK Fuzzy Model)

• Tsukamoto Fuzzy Models

22
Mamdani Fuzzy Inference System
• Mamdani-type inference produces the output
membership functions to be fuzzy sets

• Mamdani systems:
• Are intuitive.
• Have widespread acceptance.
• Are well-suited to human input.

23
Sugeno Fuzzy Inference System
• Sugeno-type inference produces the output
membership functions to be either linear functions
or constants
If Input 1 is x and Input 2 is y,
then Output is z = ax + by + c
• Sugeno systems:
• suitable for mathematical analysis
• computationally efficient.
• Work well with linear techniques, as well as
optimization and adaptive techniques
24
Implication Method
• The input for the implication process is a single
number given by the antecedent, and the output is
a fuzzy set. 
• Implication is implemented for each rule.
• Two built-in methods are supported, and they are
the same functions that are used by the AND
method: min (minimum), which truncates the
output fuzzy set, and prod (product), which scales
the output fuzzy set.

25
Tipping Example: Implication
Method

26
Aggregation of Output
• Aggregation is the process by which the fuzzy sets
that represent the outputs of each rule are combined
into a single fuzzy set.
• Aggregation only occurs once for each output
variable, which is before the final defuzzification step.
• The input of the aggregation process is the list of
truncated output functions returned by the
implication process for each rule. The output of the
aggregation process is one fuzzy set for each output
variable.

27
Tipping Example : Aggregation

28
Defuzzification
• Why? Final output communicated to machine must
be crisp

29
Defuzzification Methods
•  Max Membership Principle
• also known as height method

• Where is the defuzzified value.

30
Defuzzification Methods
•  Centroid Method

• Where denotes algebraic integration

31
Defuzzification Methods
•  Weighted Average Method

• Where denotes the algebraic sum and is the

centroid

32
Defuzzification Methods
•  Mean max Method

• Membership function is a plateau rather than single

point

33
Tipping Problem :
Defuzzification

34
An Application of FIS

Home Energy Management

System (HEMS) using
Fuzzy Inference System

35
Research Challenges in HEMS
• User Comfort Maximization

• Energy Consumption Minimization

• Cost Minimization

• PAR Minimization

36
System Model

37
Input Parameters

38
Input Parameters

39
Input Parameters

40
Sample Rule Base

1 L L HP A L L VL
2 L M OP P L L M
3 L H MP P H H M
4 M H OP A M H H
5 M L MP P H M M
6 H M OP A L M M
7 H H OP P H H VH

41
 Results: Energy Consumption
3000 4000

3500
2500

Energy Consumption (kWh)

Energy Consumption (kWh)

3000
2000
2500

1500 2000

1500
1000
Mamdani without Humidity 1000 Mamdani without Humidity
Sugeno without Humidity Sugeno without Humidity
500 Mamdani with Humidity Mamdani with Humidity
500
Sugeno with Humidity Sugeno with Humdity
0 0
1 2 3 4 1 2 3 4

Fig. Monthly Energy Consumption for hot city Fig. Monthly Energy Consumption of cold city

42
 Results: Total Cost
300 400

350
250
300
200
250
Cost ($)

Cost ($)
150 200

150
100
Mamdani without Humidity Mamdani without Humidity
100
Sugeno without Humidity Sugeno without Humidity
50 Mamdani with Humidity
50 Mamdani with Humidity
Sugeno with Humidity Sugeno with Humidity
0 0
1 2 3 4 1 2 3 4

Fig. Monthly Cost for hot city Fig. Monthly Cost for cold city

43
Conclusion
• Simple fuzzy logic model

• User comfort maintenance along with energy

conservation by adding humidity

44
Recap
• Fuzzy Inference System (FIS)

• Components of Fuzzy Inference System

• Application of Fuzzy Inference System in Home

Energy Consumption Scenarios

45
 EE-873: Fuzzy Control

Any Questions/Ideas?

46
 EE-873: Fuzzy Control

Thank You!

47