LINGUISTIC VARIABLES

 A linguistic variable is a variable of higher order than a fuzzy

CHAPTER 12 variable and its values are taken to be fuzzy variable

• The linguistic variable speed ranges between 0 and 300

km/h and includes the fuzzy sets slow, very slow, fast, …
• Fuzzy sets define the linguistic values.
FUZZY RULE BASE AND
A linguistic variable is characterized by
APPROXIMATE Name of the variable
REASONING Term set of the variable
Syntactic rule for generating values of the variable
Semantic rule for associating each value with its meaning

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

TRUTH VALUES AND TABLES IN FUZZY LOGIC

The degree of an element in a fuzzy set corresponds to the truth value of LINGUISTIC MODIFIERS OR HEDGES
a proposition in fuzzy logic systems.
 Hedges are qualifiers of a linguistic variable.
Fuzzy logic uses linguistic variable

The values of linguistic variables are words or sentences in a natural or

artificial language • All purpose: very, quite, extremely
Example • Probability: likely, unlikely
• Quantifiers: most, several, few
Height is a linguistic variable , it take values such as tall, medium , short
and so on • Possibilities: almost impossible, quite possible
The linguistic variable provide approximate characterization of complex
problem

The name of the variable, the universe of discourse and a fuzzy

subset of universe of discourse characterize a fuzzy variable
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 TRUTH TABLES
LINGUISTIC HEDGES (LINGUISTIC QUANTIFIERS)

 Hedges modify the shape of a fuzzy set. Truth tables define logic functions of two propositions. Let X and Y be
two propositions, either of which can be true or false.

The operations over the propositions are:

1. Conjunction (): X AND Y.

2. Disjunction (): X OR Y.

3. Implication or conditional (): IF X THEN Y.

4. Bidirectional or equivalence (): X IF AND ONLY IF Y.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

FUZZY PREPOSITIONS
FUZZY PREPOSITIONS
• Reasoning has logic as its basis, whereas prepositions
are text sentences expressed in any language
• Generally expressed in an canonical form as For extending reasoning capacity fuzzy logic uses
z is P fuzzy predicates, fuzzy predicate modifies fuzzy
• Z is Subject quantifiers and fuzzy qualifiers in the fuzzy
• P is predicate prepositions
• E a ple Lo do is i U ited Ki gdo The fuzzy prepositions make the fuzzy logic
• Lo do is su je t a d i u ited Ki gdo is differ from the classical logic
predicate The fuzzy prepositions are as follows
• Every preposition has its opposite called negation

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 FUZZY PREDICATES FUZZY QUANTIFIERS
• Most , several, many, frequently
In fuzzy logic predicate can be fuzzy
• Ma y people are educated
Example – tall, short , quick • Fuzzy quantifier can be interpreted as a fuzzy
 Pete is tall number or fuzzy preposition
It is obvious that most of the predicates in • Which provide an imprecise characterization of the
natural language are fuzzy rather than crisp cardinality of one or more fuzzy or non fuzzy sets
• These quantifiers can be used to manipulate
probabilities within the fuzzy logic

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

FUZZY PREDICATE MODIFIERS FUZZY QUALIFIERS

• Predicate modifiers that act as hedge • 4 model of fuzzy qualification
• Very, fairly, moderately, rather, slightly 1) Fuzzy truth qualification
• Necessary for generating the linguistic variable • Expressed as x is Ƭ
• The cli ate is oderately cool • Where Ƭ is a fuzzy truth value claim the degree of
truth of fuzzy preposition
• (Don is Young) is NOT VERY TRUE
• Here the qualified preposition is Don is Young
• Qualifying fuzzy truth value is NOT VERY true

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
2. Fuzzy probability Qualification 4. Fuzzy Usuality Qualification
– Denoted as x is λ – Denoted as usuality(x)=usuality(x is F)
– λ is fuzzy probability – X is a variable taking values in the universe of
– In crisp logic probability is either numerical or interval discourse U
– In fuzzy logic probability is expressed by – F is predicate, fuzzy subset of U
• Likely – The prepositions that are usually true or the events
• Very likely that have high probability of occurance
• Unlikely
• Around and so on
– Do is You g is Likel
– Qualifying fuzzy probability is likely
– These may be interpreted as fuzzy numbers , which
may be manipulated using fuzzy arithmetic
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

3. Fuzzy possibility Qualification

FUZZY RULES
– Denoted as x is π
A fuzzy rule is defined as the conditional statement of the form
– π is fuzzy possibility
– In fuzzy logic possibility is expressed by IF antecedent THEN consequence
• possible
• Quite possible
If x is A THEN y is B
• Almost impossible
– Do is You g is al ost i possi le where x and y are linguistic variables and A and B are linguistic values
determined by fuzzy sets on the universes of discourse X and Y.
– Qualifying fuzzy possibility is almost impossible
– These may be interpreted as labels of fuzzy subset of
real line

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 The decision-making process is based on rules with sentence
FORMATION OF FUZZY RULES
conjunctives AND, OR and ALSO. Three general forms are adopted for forming fuzzy rules. They are:

 Each rule corresponds to a fuzzy relation.

 Assignment statements,
 Rules belong to a rule base.
 Conditional statements,
 Example: If (Distance x to second car is SMALL) OR (Distance y
to obstacle is CLOSE) AND (speed v is HIGH) THEN (perform  Unconditional statements.
LARGE correction to steering angle ) ALSO (make MEDIUM
reduction in speed v).

 Three antecedents (or premises) in this example give rise to two

outputs (consequences).

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

FUZZY RULE FORMATION Assignment Statements Unconditional Statements

IF height is tall
THEN weight is heavy.

Here the fuzzy classes height and weight have a given range (i.e., the
universe of discourse).

range (height) = [140, 220]

range (weight) = [50, 250]

Conditional Statements

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 DECOMPOSITION OF FUZZY RULES AGGREGATION OF FUZZY RULES
Aggregation of rules is the process of obtaining the overall consequents
A compound rule is a collection of several simple rules combined
from the individual consequents provided by each rule.
together.

 Conjunctive system of rules.

 Multiple conjunctive antecedent,
 Disjunctive system of rules.
 Multiple disjunctive antecedent,

 Conditional statements (with ELSE and UNLESS).

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

DECOMPOSITION OF FUZZY RULES AGGREGATION OF FUZZY RULES

Multiple Conjunctive Antecedents Multiple disjunctive antecedent Conjunctive system of rules

Conditional Statements ( With Else and Unless)

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
Disjunctive system of rules Problem: Given

(a) membership functions for short, medium-height, tall, light,

medium-weight and heavy;
(b) The three fuzzy rules;
(c)the fact that John’s height is 180

estimate John’s weight.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

FUZZY RULE - EXAMPLE Solution:

(1) From John’s height we know that
Rule 1: If height is short then weight is light.
John is short (degree 0.3)
John is of medium height (degree 0.6).
Rule 2: If height is medium then weight is medium.
John is tall (degree 0.2).

Rule 3: If height is tall then weight is heavy. (2) Each rule produces a fuzzy set as output by truncating the
consequent membership function at the value of the antecedent
membership.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

 The cumulative fuzzy output is obtained by OR-ing the output from

each rule.

 Cumulative fuzzy output (weight at 6’1”).

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 Categorical reasoning:
1. De-fuzzify to obtain a numerical estimate of the output.
~ In this mode of reasoning, the premises are assumed
2. Choose the middle of the range where the truth value is to be in the canonical form
maximum.
" X is A " or in the conditional canonical form ,
3. John’s weight = 80 Kg. "If X is A then Y is B",
where A and B are fuzzy predicates.
~ Notations：
X, Y, Z, .... = (fuzzy) variables taking values in the
universes U,V, W.
A, B, C, .... = fuzzy predicates.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

FUZZY REASONING ~ Projection rule of inference:

There exist four modes of fuzzy approximate reasoning, which include:

1. Categorical reasoning,

2. Qualitative reasoning,

3. Syllogistic reasoning,

4. Dispositional reasoning.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

Syllogistic Reasoning
• In this reasoning antecedents with fuzzy quantifiers
are related to inference rules
• A fuzzy syllogism can be expressed as follows
X= k A’s a e B’s
Y=k C’s a e D’s
then
Z=k E’s a e F’s

• A,B,C,D,E are fuzzy predicates

• K1 and k2 are given fuzzy quantifiers
• K3 is a fuzzy quantifier which has to be decided

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 Dispositional Reasoning
SINGLE RULE WITH SINGLE ANTECEDANT
• In this reasoning antecedents are dispositions
that may contain implicitly or explicitly the
fuzz ua tifie usuall
• Usuality plays a major role in dispositional
reasoning
• Link together the dispositional and syllogistic
mode of reasoning

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

REASONING WITH FUZZY RULES

 In classical systems, rules with true antecedents fire.

 In fuzzy systems, truth (i.e., membership in some class) is relative,

so all rules fire (to some extent).

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
MULTIPLE ANTECEDANTS MULTIPLE CONSEQUENTS
IF x is A AND y is B THEN z is C IF x is A THEN y is B AND z is C
IF x is A OR y is B THEN z is C
Each consequent is affected equally by the membership in the
Use unification (OR) or intersection (AND) operations to calculate a antecedent class(es).
membership value for the whole antecedent.
E.g., IF x is tall THEN x is heavy AND x has large feet.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

MULTIPLE RULE WITH MULTIPLE ANTECEDANTS

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 • Single input single output Mamdani fuzzy
model with 3 rules: Single output-input Sugeno fuzzy model with
three rules
• If X is small then Y is small  R1 If X is small then Y = 0.1X + 6.4
• If X is medium then Y is medium  R2 If X is medium then Y = -0.5X + 4
If X is large then Y = X – 2
• Is X is large then Y is large  R3 If “small”, “medium” & “large” are nonfuzzy sets
X = input [-10, 10] Y = output [0,10] Using then
the overall input-output curve is a piece wise
centroid Defuzzification linear

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

Two input single-output Mamdani fuzzy model with 4 rules: Two-input single output fuzzy model with 4 rules
If X is small & Y is small then Z is negative large R1: if X is small & Y is small then z = -x +y +1
If X is small & Y is large then Z is negative small R2: if X is small & Y is large then z = -y +3
If X is large & Y is small then Z is positive small R3: if X is large & Y is small then z = -x +3
If X is large & Y is large then Z is positive large R4: if X is large & Y is large then z = x + y + 2
X = [-5, 5]; Y = [-5, 5]; Z = [-5, 5]

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 Construction and Working of Inference System
FUZZY INFERENCE SYSTEMS (FIS)
 Fuzzy rule based systems, fuzzy models, and fuzzy expert systems
Fuzzy inference system consists of a fuzzification interface,
are also known as fuzzy inference systems. a rule base, a database, a decision-making unit, and
 The key unit of a fuzzy logic system is FIS. finally a Defuzzification interface
 The primary work of this system is decision-making.
– a rule base containing a number of fuzzy IF–THEN rules;
 FIS uses “IF...THEN” rules along with connectors “OR” or “AND” for
making necessary decision rules. – a database which defines the membership functions of the
 The input to FIS may be fuzzy or crisp, but the output from FIS is fuzzy sets used in the fuzzy rules;
always a fuzzy set. – a decision-making unit which performs the inference
 When FIS is used as a controller, it is necessary to have crisp operations on the rules;
output.
 Hence, there should be a defuzzification unit for converting fuzzy
– a fuzzification interface which transforms the crisp inputs into
variables into crisp variables along FIS. degrees of match with linguistic values; and
– a Defuzzification interface which transforms the fuzzy results of
the inference into a crisp output.
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

BLOCK DIAGRAM OF FIS

WORKING OF FIS
• The crisp input is converted in to fuzzy by using
fuzzification method.
• After fuzzification the rule base is formed.
• The rule base and the database are jointly
referred to as the knowledge base.
• Defuzzification is used to convert fuzzy value to
the real world value which is the output.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 MAMDANI FUZZY INFERENCE SYSTEMS (FIS)
The steps of fuzzy reasoning (inference operations upon
fuzzy IF–THEN rules) performed by FISs are: • Mamdani’s fuzzy inference method is the most
commonly seen fuzzy methodology.
1. Compare the input variables with the membership • Mamdani’s method was among the first control
functions on the antecedent part to obtain the systems built using/fuzzy set theory.
membership values of each linguistic label(this step is
often called fuzzification.) • It was proposed by Mamdani (1975) as an attempt
2. Combine (through a specific t-norm operator, usually to control a steam engine and boiler combination by
multiplication or min) the membership values on the synthesizing a set of linguistic control rules
premise part to get firing strength (weight) of each rule. obtained from experienced human operators.
3. Generate the qualified consequents (either fuzzy or • Mamdani’s effort was based on Zadeh’s (1973) paper
crisp) or each rule depending on the firing strength. on fuzzy algorithms for complex systems and
4. Aggregate the qualified consequents to produce a crisp decision processes.
output. (This step is called Defuzzification.)
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

MAMDANI FUZZY INFERENCE SYSTEMS (FIS)

TYPES OF FIS
• Mamdani type inference expects the output membership functions to
There are two types of Fuzzy Inference Systems:
be fuzzy sets.
• After the aggregation process, there is a fuzzy set for each output
 Mamdani FIS(1975)
variable that need Defuzzification.
• It is possible, and in many cases much more efficient, to use a single
 Sugeno FIS(1985)
spike as the output membership function rather than a distributed
fuzzy set.
• This is sometimes known as a singleton output membership
function, and it can be thought of as a pre - defuzzified fuzzy set.
• It enhances the efficiency of the Defuzzification process because it
greatly simplifies the computation required by the more general
Mamdani method, which finds the centroid of a two-dimensional
function.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 MAMDANI FUZZY INFERENCE SYSTEMS (FIS)
Creating Fuzzy Rules
 Fuzzy rules are a collection of linguistic statements that describe
how the FIS should make a decision regarding classifying an input or
controling an output.
 Fuzzy rules are always written in the following form:
if (input 1 is membership function 1) and/or (input 2 is membership
function 2) and/or. . . then (output n is output membership function n).
 For example:
 if temperature is high and humidity is high then room is hot.
 There would have to be membership functions that define high
temperature (input 1), high humidity (input 2), and a hot room
(output 1).
 This process of taking an input such as temperature and processing
it through a membership function to determine “high” temperature
is called fuzzification and is discussed in section, “Fuzzification.”
 Also, “AND”/“OR” in the fuzzy rule should be defined.
 This is called fuzzy combination
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

MAMDANI FUZZY INFERENCE SYSTEMS (FIS) Fuzzification

To compute the output of this FIS given the inputs, six  The purpose of fuzzification is to map the inputs from a set of
sensors (or features of those sensors such as amplitude or
steps has to be followed: spectrum) to values from 0 to 1 using a set of input
1. Determining a set of fuzzy rules membership functions.
2. Fuzzifying the inputs using the input membership  In the example shown in Fig. 6.2, there are two inputs, x0 and
functions y0 shown at the lower left corner.
3. Combining the fuzzified inputs according to the fuzzy rules  These inputs are mapped into fuzzy numbers by drawing a line
to establish a rule strength up from the inputs to the input membership functions above
4. Finding the consequence of the rule by combining the rule and marking the intersection point.
strength and the output membership function  These input membership functions, as discussed previously, can
5. Combining the consequences to get an output distribution represent fuzzy concepts such as “large” or “small,” “old” or
6. Defuzzifying the output distribution (this step is only if a “young,” “hot” or “cold,” etc.
crisp output (class) is needed).  When choosing the input membership functions, the definition
of large and small may be different for each input.
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 Consequence Defuzzification of Output Distribution

The consequence of a fuzzy rule is computed using • In many instances, it is desired to come up with a
two steps: single crisp output from an FIS.
1.Computing the rule strength by combining the • For example, if one was trying to classify a letter
fuzzified inputs using the fuzzy combination drawn by hand on a drawing tablet, ultimately
process discussed in previous section. the FIS would have to come up with a crisp
number to tell the computer which letter was
This is shown in Fig. 6.2.
drawn.
In this example, the fuzzy “AND” is used to
• This crisp number is obtained in a process known
combine the membership functions to compute
as Defuzzification.
the rule strength.
2. Clipping the output membership function at the
rule strength.
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

Combining Outputs into an Output Distribution

 The outputs of all of the fuzzy rules must now be

combined to obtain one fuzzy output distribution.
 This is usually, but not always, done by using the fuzzy
“OR.”
• The output membership functions on the right-hand side
of the figure are combined using the fuzzy OR to obtain
the output distribution shown on the lower right corner
of the Fig. 6.2.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
MAMDANI FUZZY INFERENCE SYSTEMS (FIS) SUGENO FIS
 Fuzzify input variables:
• Determine membership values.

 Evaluate rules:
• Based on membership values of (composite) antecedents.

 Aggregate rule outputs:

• Unify all membership values for the output from all rules.

 Defuzzify the output:

• COG: Center of gravity (approx. by summation).

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

SUGENO FUZZY INFERENCE SYSTEMS (FIS)

Sugeno fuzzy model
The main steps of the fuzzy inference process namely,

1. fuzzifying the inputs and

 Sugeno fuzzy model is also know as Sugeno–
Takagi model.
2. applying the fuzzy operator are exactly the same as in MAMDANI
 A typical fuzzy rule in a Sugeno fuzzy model has
FIS.
the format
The main difference between Mamdani’s and Sugeno’s methods is that
Sugeno output membership functions are either linear or constant.

 where AB are fuzzy sets in the antecedent;

 Z = f(x, y) is a crisp function in the consequent
Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science
and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 Sugeno fuzzy model Sugeno fuzzy model
• For a zero-order Sugeno model, the output level z is
• Usually f(x, y) is a polynomial in the input variables x and y,
a constant (a = b = 0).
but it can be any other functions that can appropriately
describe the output of the system within the fuzzy region • The output level zi of each rule is weighted by the
specified by the antecedent of the rule. firing strength wi of the rule.
• When f(x, y) is a first-order polynomial, we have the first- • For example, for an AND rule with Input 1 = x and
order Sugeno fuzzy model. Input 2 = y, the firing strength is
• When f is a constant, we then have the zero-order Sugeno
fuzzy model, which can be viewed either as a special case of
the Mamdani FIS
• or a special case of Tsukamoto’s fuzzy model where each
rule’s consequent is specified by a membership function of a
step function centered at the constant.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

Sugeno fuzzy model Comparison Between Sugeno and Mamdani Method

• The main difference between Mamdani and Sugeno is that the Sugeno
• The first two parts of the fuzzy inference process, output membership functions are either linear or constant.
fuzzifying the inputs and applying the fuzzy operator, • Also the difference lies in the consequents of their fuzzy rules, and thus their
are exactly the same. aggregation and Defuzzification procedures differ suitably
• In Sugeno method a large number of fuzzy rules must be employed to
• The main difference between Mamdani and Sugeno is approximate periodic or highly oscillatory functions.
that the Sugeno output membership functions are • The minimal configuration of the TS fuzzy systems can be reduced and
either linear or constant. becomes smaller than that of the Mamdani fuzzy systems if nontrapezoidal
or nontriangular input fuzzy sets are used.
• A typical rule in a Sugeno fuzzy model has the form • Far fewer mathematical results exist for TS fuzzy controllers than do for
Mamdani fuzzy controllers, notably those on TS fuzzy control system
stability.
• Mamdani is easy to form compared to Sugeno method.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
 SUMMARY
Advantages of the Sugeno Method
 Advantages of fuzzy logic
• Allows the use of vague linguistic terms in the rules.
 It is computationally efficient.
 It works well with linear techniques (e.g., PID  Disadvantages of fuzzy logic
• Difficult to estimate membership function
control).
• There are many ways of interpreting fuzzy rules, combining the
 It works well with optimization and adaptive outputs of several fuzzy rules and de-fuzzifying the output.
techniques.
 It has guaranteed continuity of the output
surface.
 It is well suited to mathematical analysis.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

Advantages of the Mamdani Method

• It is intuitive.
• It has widespread acceptance.
• It is well suited to human input. HYBRID SOFT
COMPUTING
TECHNIQUES

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
HYBRID SYSTEMS The use of hybrid systems is growing rapidly with successful
applications in areas such as
Neural Network Systems
Neural networks are the simplified models of the human nervous
systems mimicking our ability to adapt to certain situations and to  engineering design
learn from the past experiences.  stock market analysis and prediction
 medical diagnosis
Fuzzy Logic  process control
Fuzzy logic or fuzzy systems deal with uncertainty or vagueness  credit card analysis and
existing in a system and formulating fuzzy rules to find a solution  few other cognitive simulations.
to problems.

Genetic Algorithm
Genetic algorithms inspired by the natural evolution process are
adaptive search and optimization algorithms.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

The main aim of the concept of hybridization is to overcome the

VARIOUS HYBRID SYSTEMS
weakness in one technique while applying it and bringing out the
strength of the other technique to find solution by combining them. In this text book, the following three different hybrid systems are
discussed:
Neural networks are good at recognizing patterns but they are not
good at explaining how they reach their decisions.  Neuro fuzzy hybrid system;

On the contrary, fuzzy logic is good at explaining the decisions but  neuron genetic hybrid system;
cannot automatically acquire the rules used for making the decisions.
Also, the tuning of membership functions becomes an important issue  fuzzy genetic hybrid systems.
in fuzzy modeling. Genetic algorithms offer a possibility to solve this
problem.

These limitations act as a central driving force for the creation of

hybrid soft computing systems where two or more techniques are
combined in a suitable manner that overcomes the limitations of
individual techniques.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
NEURO FUZZY HYBRID SYSTEMS TYPES OF NEURO FUZZY HYBRID SYSTEMS
Definition: NFSs can be classified into the following two systems:

A neuro-fuzzy hybrid system (also called fuzzy neural hybrid) is a 1. Cooperative NFSs.
learning mechanism that utilizes the training and learning algorithms
from neural networks to find parameters of a fuzzy system. 2. General neuro-fuzzy hybrid systems.

Advantages of neuro fuzzy hybrid systems:

 It can handle any kind of information (numeric, linguistic, logical,

etc.).
 It can manage imprecise, partial, vague or imperfect information.
 It can resolve conflicts by collaboration and aggregation.
 It has self-learning, self-organizing and self-tuning capabilities.
 It doesn’t need prior knowledge of relationships of data.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam

ARCHITECTURE OF NEURO FUZZY HYBRID CO-OPERATIVE NEURO FUZZY SYSTEMS

SYSTEMS
In this type of system, both artificial neural network (ANN) and fuzzy
The general architecture of neuro- system work independently from each other. The ANN attempts to
fuzzy hybrid system is as shown in learn the parameters from the fuzzy system.
Figure 1.

Fig 1

The architecture is a three-layer feed forward neural network model. It

can also be observed that the first layer corresponds to the input
variables, and the second and third layers correspond to the fuzzy
rules and output variables, respectively. The fuzzy sets are converted
to (fuzzy) connection weights.

Arsha J K,AP,Computer Science Arsha J K,AP,Computer Science

and Engineering, VJCET, and Engineering, VJCET,
Vazhakulam Vazhakulam
GENERAL NEURO FUZZY HYBRID SYSTEMS Fuzzy inference
General neuro-fuzzy hybrid systems (NFHS) resemble neural networks
where a fuzzy system is interpreted as a neural network of special kind.
The most commonly used fuzzy inference technique
Figure 2 illustrates an NFHS
is the so-called Mamdani method. In 1975,
Professor Ebrahim Mamdani of London University
In Fig 2, the rule base of
a fuzzy system is built one of the first fuzzy systems to control a
assumed to be a neural steam engine and boiler combination. He applied a
network; the fuzzy sets
are regarded as weights
set of fuzzy rules supplied by experienced human
and the rules and the operators.
input and output
variables as neurons.

Fig 2
Arsha J K,AP,Computer Science
and Engineering, VJCET,
Vazhakulam 11/23/2017 Intelligent Systems and Soft Computing 2

Lecture 5 Mamdani fuzzy inference

Fuzzy expert systems: n The Mamdani-style fuzzy inference process is
Fuzzy inference performed in four steps:

n Mamdani fuzzy inference l fuzzification of the input variables,

n Sugeno fuzzy inference l rule evaluation;

n Case study l aggregation of the rule outputs, and finally

n Summary l defuzzification.

11/23/2017 Intelligent Systems and Soft Computing 1 11/23/2017 Intelligent Systems and Soft Computing 3
We examine a simple two-input one-output problem that Step 2: Rule Evaluation
includes three rules:
The second step is to take the fuzzified inputs,
Rule: 1 Rule: 1
IF x is A3 IF project_funding is adequate
(x=A1) = 0.5, (x=A2) = 0.2, (y=B1) = 0.1 and (y=B 2) =
OR y is B1 OR project_staffing is small 0.7, and apply them to the antecedents of the fuzzy
THEN z is C1 THEN risk is low rules. If a given fuzzy rule has multiple antecedents,
Rule: 2 Rule: 2 the fuzzy operator (AND or OR) is used to obtain a
IF x is A2 IF project_funding is marginal single number that represents the result of the
AND y is B2 AND project_staffing is large antecedent evaluation. This number (the truth value)
THEN z is C2 THEN risk is normal
is then applied to the consequent membership
Rule: 3 Rule: 3 function.
IF x is A1 IF project_funding is inadequate
THEN z is C3 THEN risk is high

11/23/2017 Intelligent Systems and Soft Computing 4 11/23/2017 Intelligent Systems and Soft Computing 6

Step 1: Fuzzification To evaluate the disjunction of the rule antecedents,

we use the OR fuzzy operation. Typically, fuzzy
The first step is to take the crisp inputs, x1 and y1 expert systems make use of the classical fuzzy
(project funding and project staffing), and determine operation union:
the degree to which these inputs belong to each of the
appropriate fuzzy sets. A B(x) = max [A(x), B(x)]

Crisp Input Crisp Input

Similarly, in order to evaluate the conjunction of the
x1 rule antecedents, we apply the AND fuzzy operation
1 1 B1
intersection:
A1 A2 A3 B2
0.7
A B(x) = min [A(x), B(x)]
0.5
0.2 0.1
0 0
x1 X y1 Y
 (x = A1) = 0.5  (y = B1) = 0.1
 (x = A2) = 0.2  (y = B2) = 0.7

11/23/2017 Intelligent Systems and Soft Computing 5 11/23/2017 Intelligent Systems and Soft Computing 7
 Mamdani-style rule evaluation n While clipping is a frequently used method, scaling
1 1 1
offers a better approach for preserving the original
A3 B1
0.1 0.1
C1 C2 C3 shape of the fuzzy set. The original membership
0.0 OR
0 x1 X 0 y1 Y
(max)
0 Z
function of the rule consequent is adjusted by
Rule 1: IF x is A3 (0.0) OR y is B1 (0.1) THEN z is C1 (0.1)
multiplying all its membership degrees by the truth
1 1 1 value of the rule antecedent. This method, which
0.7
A2 0.2 B2 AND 0.2
C1 C2 C3 generally loses less information, can be very useful
(min)
0 x1 X 0 y1 Y 0 Z in fuzzy expert systems.
Rule 2: IF x is A2 (0.2) AND y is B2 (0.7) THEN z is C2 (0.2)
1 1
A1 0.5 0.5 C1 C2 C3

0 x1 X 0 Z
Rule 3: IF x is A1 (0.5) THEN z is C3 (0.5)

11/23/2017 Intelligent Systems and Soft Computing 8 11/23/2017 Intelligent Systems and Soft Computing 10

Now the result of the antecedent evaluation can be Clipped and scaled membership functions
applied to the membership function of the
consequent.
Degree of Degree of
n The most common method of correlating the rule Membership Membership
1.0 1.0
consequent with the truth value of the rule
antecedent is to cut the consequent membership C2 C2
function at the level of the antecedent truth. This
method is called clipping. Since the top of the 0.2 0.2
membership function is sliced, the clipped fuzzy set
0.0 0.0
loses some information. However, clipping is still Z Z
often preferred because it involves less complex and
faster mathematics, and generates an aggregated
output surface that is easier to defuzzify.
11/23/2017 Intelligent Systems and Soft Computing 9 11/23/2017 Intelligent Systems and Soft Computing 11
Step 3: Aggregation of the rule outputs Step 4: Defuzzification
Aggregation is the process of unification of the The last step in the fuzzy inference process is
outputs of all rules. We take the membership defuzzification. Fuzziness helps us to evaluate the
functions of all rule consequents previously clipped or rules, but the final output of a fuzzy system has to be
scaled and combine them into a single fuzzy set. a crisp number. The input for the defuzzification
process is the aggregate output fuzzy set and the
The input of the aggregation process is the list of
output is a single number.
clipped or scaled consequent membership functions,
and the output is one fuzzy set for each output
variable.

11/23/2017 Intelligent Systems and Soft Computing 12 11/23/2017 Intelligent Systems and Soft Computing 14

Aggregation of the rule outputs n There are several defuzzification methods, but
probably the most popular one is the centroid
technique. It finds the point where a vertical line
1 1 1
would slice the aggregate set into two equal masses.
C1 C2 C3
0.5 0.5 Mathematically this centre of gravity (COG) can
0.2 0.2
0.1
0 Z 0 Z 0 Z
0.1
0 Z
be expressed as:
z is C1 (0.1) z is C2 (0.2) z is C3 (0.5) 
b
A x x dx
COG  a
b
A x dx
a
11/23/2017 Intelligent Systems and Soft Computing 13 11/23/2017 Intelligent Systems and Soft Computing 15
n Centroid defuzzification method finds a point Sugeno fuzzy inference
representing the centre of gravity of the fuzzy set, A,
on the interval, ab. n Mamdani-style inference, as we have just seen,
n A reasonable estimate can be obtained by calculating requires us to find the centroid of a two-dimensional
it over a sample of points. shape by integrating across a continuously varying
function. In general, this process is not
1.0 computationally efficient.
0.8
n Michio Sugeno suggested to use a single spike, a
0.6 A
singleton, as the membership function of the rule
0.4 consequent. A singleton,, or more precisely a fuzzy
0.2 singleton, is a fuzzy set with a membership
a b function that is unity at a single particular point on
0.0 X
150 160 170 180 190 200 210 the universe of discourse and zero everywhere else.
11/23/2017 Intelligent Systems and Soft Computing 16 11/23/2017 Intelligent Systems and Soft Computing 18

Centre of gravity (COG): Sugeno-style fuzzy inference is very similar to the

(0  10  20)  0.1  (30  40  50  60)  0.2  (70  80  90  100)  0.5
Mamdani method. Sugeno changed only a rule
COG   67.4
0.1  0.1  0.1  0.2  0.2  0.2  0.2  0.5  0.5  0.5  0.5 consequent. Instead of a fuzzy set, he used a
mathematical function of the input variable. The
Degreeof
Membership format of the Sugeno-style fuzzy rule is
1.0
IF x is A
0.8
AND y is B
0.6
THEN z is f (x, y)
0.4
0.2 where x, y and z are linguistic variables; A and B are
0.0 fuzzy sets on universe of discourses X and Y,
0 10 20 30 40 50 60 70 80 90 100
Z
respectively; and f (x, y) is a mathematical function.
67.4

11/23/2017 Intelligent Systems and Soft Computing 17 11/23/2017 Intelligent Systems and Soft Computing 19
The most commonly used zero-order Sugeno fuzzy Sugeno-style aggregation of the rule outputs
model applies fuzzy rules in the following form:
IF x is A 1 1 1 1
AND y is B 0.5 0.5
0.2
0.1 0.2 0.1
THEN z is k 0 k1 Z 0 k2 Z 0 k3 Z 0 k1 k2 k3 Z

where k is a constant. z is k1 (0.1) z is k2 (0.2) z is k3 (0.5) 

In this case, the output of each fuzzy rule is constant.
All consequent membership functions are represented
by singleton spikes.

11/23/2017 Intelligent Systems and Soft Computing 20 11/23/2017 Intelligent Systems and Soft Computing 22

Sugeno-style rule evaluation Weighted average (WA):

1 1 1
(k1)  k1 ( k 2)  k 2  (k 3)  k 3 0.1 20  0.2  50  0.5  80
A3 B1 WA    65
0.0
0.1 OR
(max)
0.1
( k1)  ( k 2)  ( k3) 0.1 0.2  0.5
0 x1 X 0 y1 Y 0 k1 Z

Rule 1: IF x is A3 (0.0) OR y is B1 (0.1) THEN z is k1 (0.1)

Sugeno-style defuzzification
1 1 1
0.7
A2 0.2 B2 AND 0.2
(min)
0 x1 X 0 y1 Y 0 k2 Z

Rule 2: IF x is A2 (0.2) AND y is B2 (0.7) THEN z is k2 (0.2)

1
0 z1 Z
1
A1 0.5 0.5

Crisp Output
0 x1 X 0 k3 Z z1
Rule 3: IF x is A1 (0.5) THEN z is k3 (0.5)

11/23/2017 Intelligent Systems and Soft Computing 21 11/23/2017 Intelligent Systems and Soft Computing 23
How to make a decision on which method
to apply – Mamdani or Sugeno?
n Mamdani method is widely accepted for capturing
expert knowledge. It allows us to describe the
expertise in more intuitive, more human-like
manner. However, Mamdani-type fuzzy inference
entails a substantial computational burden.
Single input single output antecedent & consequent MFs
n On the other hand, Sugeno method is
computationally effective and works well with
optimisation and adaptive techniques, which makes
it very attractive in control problems, particularly
for dynamic nonlinear systems.
11/23/2017 Intelligent Systems and Soft Computing 24 11/23/2017 Overall
Intelligentinput-output
Systems26
and Soft Computing curve

More Examples for Mamdani Fuzzy Models  Example #2 (Mamdani Fuzzy models )
 Example #1 Two input single-output Mamdani fuzzy
Single input single output Mamdani fuzzy
model with 4 rules:
model with 3 rules: If X is small & Y is small then Z is
If X is small then Y is small  R1 negative large
If X is medium then Y is medium  R2
If X is small & Y is large then Z is
Is X is large then Y is large  R3
negative small
X = input [-10, 10] Y = output [0,10]
If X is large & Y is small then Z is
Using centroid defuzzification, we obtain
the following overall input-output positive small
curve If X is large & Y is large then Z is
positive large
11/23/2017 Intelligent Systems25
and Soft Computing 11/23/2017 Intelligent Systems27
and Soft Computing
 X = [-5, 5]; Y = [-5, 5]; Z = [-5, 5] with max-min
composition & centroid defuzzification, we can More Examples for Sugeno Fuzzy Models
determine the overall input output surface

Example 1: Single output-input Sugeno fuzzy

model with three rules

If X is small then Y = 0.1X + 6.4

If X is medium then Y = -0.5X + 4
If X is large then Y = X – 2

If “small”, “medium” & “large” are nonfuzzy sets

then the overall input-output curve is a piece
wise linear

Two-input single output antecedent

28
& consequent MFs
11/23/2017 Intelligent Systems and Soft Computing 11/23/2017 Intelligent Systems30
and Soft Computing

11/23/2017
Overall input-output surface
Intelligent Systems29
and Soft Computing 11/23/2017 Intelligent Systems31
and Soft Computing
 However, if we have smooth membership
functions (fuzzy rules) the overall input-
output curve becomes a smoother one

Overall input-output surface

11/23/2017 Intelligent Systems32

and Soft Computing 11/23/2017 Intelligent Systems34
and Soft Computing

Example 2:
Two-input single output fuzzy model with 4 rules
Building a fuzzy expert system: case study
n A service centre keeps spare parts and repairs failed
R1: if X is small & Y is small then z = -x +y +1
ones.
R2: if X is small & Y is large then z = -y +3
R3: if X is large & Y is small then z = -x +3 n A customer brings a failed item and receives a spare
R4: if X is large & Y is large then z = x + y + 2 of the same type.
n Failed parts are repaired, placed on the shelf, and
thus become spares.
n The objective here is to advise a manager of the
service centre on certain decision policies to keep
the customers satisfied.

11/23/2017 Intelligent Systems33

and Soft Computing 11/23/2017 Intelligent Systems and Soft Computing 35
Process of developing a fuzzy expert system Linguistic variables and their ranges
Lingui stic Variable: Mean Delay, m
Lingui stic Va lue Notation Numerical Range (normalis ed)
1. Specify the problem and define linguistic variables. Very Short VS [0, 0.3]
Short S [0.1, 0.5]
Medium M [0.4, 0.7]
2. Determine fuzzy sets. Lingui stic Variable: Number of Servers, s
Lingui stic Va lue Notation Numerical Range (normalis ed)
Small S [0, 0.35]
3. Elicit and construct fuzzy rules. Medium M [0.30, 0.70]
Large L [0.60, 1]
Lingui stic Variable: Repair Utilisation Factor, r
4. Encode the fuzzy sets, fuzzy rules and procedures Lingui stic Va lue Notation Numerical Range
Low L [0, 0.6]
to perform fuzzy inference into the expert system. Medium M [0.4, 0.8]
High H [0.6, 1]
5. Evaluate and tune the system. Lingui stic Variable: Number of Spares, n
Lingui stic Va lue Notation Numerical Range (normalis ed)
Very Small VS [0, 0.30]
Small S [0, 0.40]
Rather Small RS [0.25, 0.45]
Medium M [0.30, 0.70]
Rather Large RL [0.55, 0.75]
Large L [0.60, 1]
Very Large VL [0.70, 1]

11/23/2017 Intelligent Systems and Soft Computing 36 11/23/2017 Intelligent Systems and Soft Computing 38

Step 1: Specify the problem and define Step 2: Determine fuzzy sets
linguistic variables Fuzzy sets can have a variety of shapes. However,
There are four main linguistic variables: average a triangle or a trapezoid can often provide an
waiting time (mean delay) m, repair utilisation adequate representation of the expert knowledge,
factor of the service centre r (is the ratio of the and at the same time, significantly simplifies the
customer arrival day to the customer departure rate) process of computation.
number of servers s, and initial number of spare parts
n.

11/23/2017 Intelligent Systems and Soft Computing 37 11/23/2017 Intelligent Systems and Soft Computing 39
 Fuzzy sets of Mean Delay m Fuzzy sets of Repair Utilisation Factor r
Degree of Degree of
Membership Membership
1.0 1.0

0.8 VS S M 0.8 L M H

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Mean Delay (normalised) Repair Utilisation Factor

11/23/2017 Intelligent Systems and Soft Computing 40 11/23/2017 Intelligent Systems and Soft Computing 42

Fuzzy sets of Number of Servers s Fuzzy sets of Number of Spares n

Degree of Degree of
Membership Membership
1.0 1.0

0.8 S M L 0.8 VS S RS M RL L VL

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Number of Spares (normalised)

11/23/2017 Intelligent Systems and Soft Computing 41 11/23/2017 Intelligent Systems and Soft Computing 43
Step 3: Elicit and construct fuzzy rules The rule table
To accomplish this task, we might ask the expert to Rule m s r n Rule m s r n Rule m s r n
describe how the problem can be solved using the 1 VS S L VS 10 VS S M S 19 VS S H VL
fuzzy linguistic variables defined previously. 2 S S L VS 11 S S M VS 20 S S H L
3 M S L VS 12 M S M VS 21 M S H M
Required knowledge also can be collected from 4 VS M L VS 13 VS M M RS 22 VS M H M
other sources such as books, computer databases, 5 S M L VS 14 S M M S 23 S M H M
flow diagrams and observed human behavior. 6 M M L VS 15 M M M VS 24 M M H S
7 VS L L S 16 VS L M M 25 VS L H RL
The matrix form of representing fuzzy rules is called 8 S L L S 17 S L M RS 26 S L H M
fuzzy associative memory (FAM). 9 M L L VS 18 M L M S 27 M L H RS

11/23/2017 Intelligent Systems and Soft Computing 44 11/23/2017 Intelligent Systems and Soft Computing 46

The square FAM representation

Rule Base 1
s 1. If (utilisation_factor is L) then (number_of_spares is S)
2. If (utilisation_factor is M) then (number_of_spares is M)
3. If (utilisation_factor is H) then (number_of_spares is L)
4. If (mean_delay is VS) and ( number_of_servers is S) then (number_of_spares is VL)
L M S VS 5. If (mean_delay is S) and (number_of_servers is S) then (number_of_spares is L)
6. If (mean_delay is M) and (number_of_servers is S) then (number_of_spares is M)
7. If (mean_delay is VS) and (number_of_servers is M) then (number_of_spares is RL)
M RL RS S 8. If (mean_delay is S) and (number_of_servers is M) then (number_of_spares is RS)
9. If (mean_delay is M) and (number_of_servers is M) then (number_of_spares is S)
10. If (mean_delay is VS) and (number_of_servers is L) then ( number_of_spares is M)
S VL L M 11. If (mean_delay is S) and (number_of_servers is L) then ( number_of_spares is S)
12. If (mean_delay is M) and (number_of_servers is L) then ( number_of_spares is VS)

VS S M m

11/23/2017 Intelligent Systems and Soft Computing 45 11/23/2017 Intelligent Systems and Soft Computing 47
 Cube FAM of Rule Base 2
s
s
VS VS VS L RL M RS
Step 5: Evaluate and tune the system
VS VS VS
L S S VS
VS VS VS M M M S
M
VS VS VS
VS VS VS
The last, and the most laborious, task is to evaluate
VS VS VS r S VL L M r
S
VS VS VS
VS VS VS H VS S M
H and tune the system. We want to see whether our
M
L s fuzzy system meets the requirements specified at
VS S M m
L M RS S
m the beginning.
M RS S VS

S S VS VS Several test situations depend on the mean delay,

VS S M
s
m
M
number of servers and repair utilization factor.
L S S VS

M VS VS VS
The Fuzzy Logic Toolbox can generate surface to
S VS VS VS
help us analyze the system’s performance.
L
VS S M
m
11/23/2017 Intelligent Systems and Soft Computing 48 11/23/2017 Intelligent Systems and Soft Computing 50

Step 4: Encode the fuzzy sets, fuzzy rules Three-dimensional plots for Rule Base 1
and procedures to perform fuzzy
inference into the expert system 0.6

To accomplish this task, we may choose one of 0.5

two options: to build our system using a 0.4

programming language such as C/C++ or Pascal,

0.3
or to apply a fuzzy logic development tool such as
MATLAB Fuzzy Logic Toolbox, Fuzzy Clips, or Fuzzy 0.2

Knowledge Builder. 0

0.2 1
0.8
0.4 0.6
0.4
0.6 0.2
mean_delay 0 number_of_servers

11/23/2017 Intelligent Systems and Soft Computing 49 11/23/2017 Intelligent Systems and Soft Computing 51
 Three-dimensional plots for Rule Base 1 Three-dimensional plots for Rule Base 2

0.5
0.6

0.4
0.5

0.3
0.4

0.3 0.2

0.2
0 0

0.2 1 0.2 1
0.8 0.8
0.4 0.6 0.4 0.6
0.4 0.4
0.6 0.2 0.6 0.2
mean_delay 0 utilisation_factor mean_delay 0 utilisation_factor

11/23/2017 Intelligent Systems and Soft Computing 52 11/23/2017 Intelligent Systems and Soft Computing 54

Three-dimensional plots for Rule Base 2

However, even now, the expert might not be
satisfied with the system performance.
0.35

To improve the system performance, we may use

additional sets - Rather Small and Rather Large –
0.3

0.25
on the universe of discourse Number of Servers,
0.2 and then extend the rule base.
0.15

0.2 1
0.8
0.4 0.6
0.4
0.6 0.2
mean_delay 0 number_of_servers

11/23/2017 Intelligent Systems and Soft Computing 53 11/23/2017 Intelligent Systems and Soft Computing 55
 Three-dimensional plots for Rule Base 3
Modified fuzzy sets of Number of Servers s
Degree of 0.35
Membership
1.0 0.3

0.8 S RS M RL L
0.25
0.6
0.2
0.4
0.2 0.15

0.0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0.2 1
Number of Servers (normalised) 0.8
0.4 0.6
0.4
0.6 0.2
mean_delay 0 number_of_servers

11/23/2017 Intelligent Systems and Soft Computing 56 11/23/2017 Intelligent Systems and Soft Computing 58

Cube FAM of Rule Base 3 Three-dimensional plots for Rule Base 3

s
s
VS VS VS L L M RS
VS VS VS
L S S VS
VS VS VS RL RL M RS
VS VS VS
RL S S VS 0.5
VS VS VS M M M S
VS VS VS
M VS VS VS
VS VS VS s RS VL RL RS
VS VS VS
RS VS VS VS 0.4
VS VS VS r L M RS S S VL L M r
VS VS VS
S VS VS VS H VS S M H
M RL M RS S
L m
VS S M 0.3
m M RS S VS

s RS S VS VS
0.2
L S S VS S S VS VS
M
RL VS S M
S S VS
m 0
M VS VS VS
0.2 1
RS VS VS VS 0.8
0.4 0.6
S VS VS VS 0.4
L 0.6 0.2
VS S M 0
mean_delay utilisation_factor
m
11/23/2017 Intelligent Systems and Soft Computing 57 11/23/2017 Intelligent Systems and Soft Computing 59
 Tuning fuzzy systems
1. Review model input and output variables, and if
required redefine their ranges.
2. Review the fuzzy sets, and if required define
additional sets on the universe of discourse. The
use of wide fuzzy sets may cause the fuzzy system to
perform roughly.
3. Provide sufficient overlap between neighboring sets.
It is suggested that triangle-to-triangle and trapezoid-
to-triangle fuzzy sets should overlap between 25% to
50% of their bases.
11/23/2017 Intelligent Systems and Soft Computing 60

4. Review the existing rules, and if required add new

rules to the rule base.
5. Examine the rule base for opportunities to write
hedge rules to capture the pathological behaviour
of the system.
6. Adjust the rule execution weights. Most fuzzy
logic tools allow control of the importance of rules
by changing a weight multiplier.
7. Revise shapes of the fuzzy sets. In most cases,
fuzzy systems are highly tolerant of a shape
approximation.
11/23/2017 Intelligent Systems and Soft Computing 61