27.12.2023.

ETF
Beograd

Fuzzy Logic
Fuzzy Inference

Aleksandar Rakić, [email protected]

Accompanying presentation for lectures 2023/2024

Contents

◼ Mamdani Fuzzy Inference

• Fuzzification of the input variables
• Rule evaluation
• Aggregation of rules’ output
• Defuzzification
◼ Sugeno Fuzzy Inference
◼ Mamdani or Sugeno?

1
 27.12.2023.

Mamdani Fuzzy Inference

◼ The most commonly used fuzzy inference technique is the so-called Mamdani
method.
◼ In 1975, Professor Ebrahim Mamdani of London University built one of the first
fuzzy systems to control a steam engine and boiler combination.
He applied a set of fuzzy rules supplied by experienced human operators.
◼ The Mamdani-style fuzzy inference process is performed in four steps:
1. Fuzzification of the input variables
2. Rule evaluation (inference)
3. Aggregation of the rule outputs (composition)
4. Defuzzification.

Mamdani Fuzzy Inference

◼ We examine a simple two-input one-output problem that includes three rules:
Rule 1: IF x is A3 OR y is B1 THEN z is C1
Rule 2: IF x is A2 AND y is B2 THEN z is C2
Rule 3: IF x is A1 THEN z is C3
◼ More realistic example for these kinds of rules:
Rule 1: IF project_funding is ADEQUATE OR project_staffing is SMALL THEN risk is LOW
Rule 2: IF project_funding is MARGINAL AND project_staffing is LARGE THEN risk is NORMAL
Rule 3: IF project_funding is INADEQUATE THEN risk is HIGH

◼ Fuzzy sets, defined over universes of input and output variables, are:

2
 27.12.2023.

Step 1: Fuzzification
◼ The first step is to take the crisp inputs, x1 and y1
(e.g., project funding and project staffing) and determine the degree
to which these inputs belong to each of the appropriate fuzzy sets.

A1 (x1) = 0.5 B1 (y1) = 0.1

A2 (x1) = 0.2 B2 (y1) = 0.7

Step 2: Rule Evaluation

(Evaluating Antecedents 1/2)
◼ The second step is to take the fuzzified inputs,
A1 (x) = 0.5, A2 (x) = 0.2, B1 (y) = 0.1 and B2 (y) = 0.7,
and apply them to the antecedents of the fuzzy rules.
Rule 1: IF x is A3 OR y is B1 THEN z is C1
Rule 2: IF x is A2 AND y is B2 THEN z is C2
Rule 3: IF x is A1 THEN z is C3

◼ If a given fuzzy rule has multiple antecedents, the fuzzy operator

(AND / OR) is used to obtain a single number that represents the
result of the antecedent evaluation.
RECALL: To evaluate the disjunction of the rule antecedents, we use the
OR fuzzy operation. Typically, fuzzy expert systems make use of the
classical (default) fuzzy operation union:
AB(x) = max [A(x), B(x)]
Similarly, in order to evaluate the conjunction of the rule antecedents,
we apply the AND fuzzy operation intersection, where default is:
AB(x) = min [A(x), B(x)] 6

3
 27.12.2023.

Step 2: Rule Evaluation

(Evaluating Antecedents 2/2)

Step 2: Rule Evaluation

(Fuzzy Implication)
Fuzzy implication is the application of the antecedent evaluation result
to the membership function of the consequent.
◼ The most common method is clipping, which cuts
the consequent membership function 𝝁𝑪𝑶𝑵 𝒊 𝒛 at
the level of the antecedent’s truth degree 𝑇𝐷𝐴𝑁𝑇 𝑖 .
◼ Since the top of the membership function is sliced,
the clipped fuzzy set loses some information.
◼ However, clipping is still often preferred for less
complex and faster mathematics, and it generates 𝝁𝑹𝒊 𝒛 = 𝐦𝐢𝐧(𝑻𝑫𝑨𝑵𝑻 𝒊 , 𝝁𝑪𝑶𝑵 𝒊 𝒛 )
an aggregated output that is easier to defuzzify.

◼ The other frequently used method is scaling,

where the membership function 𝝁𝑪𝑶𝑵 𝒊 𝒛 of
the rule’s consequent is multiplied by the value
of the rule’s antecedent truth degree 𝑇𝐷𝐴𝑁𝑇 𝑖 .
◼ This method, which generally loses less information,
offers a better approach for preserving the original
shape of the output fuzzy set. 𝝁𝑹𝒊 𝒛 = 𝑻𝑫𝑨𝑵𝑻 𝒊 ∙ 𝝁𝑪𝑶𝑵 𝒊 𝒛
8

4
 27.12.2023.

Step 2: Rule Evaluation

(Evaluating Output of Rules)

Step 3: Aggregation of Rules

◼ Aggregation is the process of
unification of the outputs of all rules.
◼ We take the membership functions of
all rule consequents previously clipped
or scaled and combine them into a
single fuzzy set.
◼ The input of the aggregation process is
the list of clipped or scaled consequent
membership functions, and the output
is one fuzzy set for each output
variable.
◼ The most common aggregation method
is maximum of rules’ output
membership functions:
AGG(z) = max [R1(z), R12(z), R3(z) ]

10

10

5
 27.12.2023.

Step 4: Defuzzification 1/2

◼ The last step in the fuzzy inference process is defuzzification.
◼ Fuzziness helps us to evaluate the rules, but the final output of a fuzzy
system has to be a crisp number.
◼ The input for the defuzzification process is the aggregate output fuzzy set
and the output is a single number.
◼ There are several defuzzification methods, but probably the most popular
one is the centroid method. It finds the point where a vertical line would
slice the aggregate set into two equal masses.
Mathematically this centre of gravity (COG) can be expressed as:
𝑧𝑚𝑎𝑥
‫𝑧=𝑧׬‬ 𝜇𝐴𝐺𝐺 𝑧 𝑧 ⅆ𝑧
𝑚𝑖𝑛
𝑧𝐶𝑂𝐺 = 𝑧𝑚𝑎𝑥 ,
‫𝐺𝐺𝐴𝜇 𝑧=𝑧׬‬ 𝑧 ⅆ𝑧
𝑚𝑖𝑛

where 𝑧𝑚𝑖𝑛 and 𝑧𝑚𝑎𝑥 are the end-points for output variable universe.
11

11

Step 4: Defuzzification 2/2

◼ Centroid defuzzification method finds a point representing the centre of gravity
of the aggregated fuzzy set AGG over the output universe [𝑧𝑚𝑖𝑛 , 𝑧𝑚𝑎𝑥 ].
◼ An estimate can be obtained by calculating it over a sample of points:
𝑛 𝑛
𝑧𝑚𝑎𝑥
‫𝑧=𝑧׬‬ 𝜇𝐴𝐺𝐺 𝑧 𝑧 ⅆ𝑧 ෍ 𝜇𝐴𝐺𝐺 𝑧𝑗 × 𝑧𝑗 × Δ𝑧 ෍ 𝜇𝐴𝐺𝐺 𝑧𝑗 × 𝑧𝑗
𝑚𝑖𝑛 𝑗=1 Δ𝑧 𝑗=1
𝑧𝐶𝑂𝐺 = 𝑧𝑚𝑎𝑥 ≈ 𝑛 = 𝑛
‫𝑧=𝑧׬‬ 𝜇𝐴𝐺𝐺 𝑧 ⅆ𝑧 Δ𝑧
𝑚𝑖𝑛 ෍ 𝜇𝐴𝐺𝐺 𝑧𝑗 × Δ𝑧 ෍ 𝜇𝐴𝐺𝐺 𝑧𝑗
𝑗=1 𝑗=1

AGG(z)
0.5

0.2
0.1

0 10 20 30 40 50 60 70 80 90 100 z
zmin zCOG=67.4 zmax

0.1 × 0 + 10 + 20 + 0.2 × 30 + 40 + 50 + 60 + 0.5 × (70 + 80 + 90 + 100)

𝑧𝐶𝑂𝐺 = = 67.4
0.1 + 0.1 + 0.1 + 0.2 + 0.2+0.2+0.2+0.5+0.5+0.5+0.5
12

12

6
 27.12.2023.

Sugeno Fuzzy Inference 1/2

◼ Mamdani-style inference, as we have just seen, requires us to find the

centroid of aggregated membership function by integrating this
function over the (output variable) universe.
◼ In general, this process is not exact, since defuzzification integrals are
approximated in calculation by sampling and defuzzified value is
effectively estimated by weighted sums.
◼ Michio Sugeno suggested to use a single spike, a singleton, as the
membership function of the rule consequent.
◼ RECALL: A singleton, or more precisely a fuzzy singleton,
is a fuzzy set with a defined membership only at a single particular
point on the universe of discourse.

13

13

Sugeno Fuzzy Inference 2/2

◼ Sugeno-style fuzzy inference is very similar to the Mamdani method.
◼ Sugeno changed only a rule consequent: instead of a continuous fuzzy set,
he used a fuzzy singleton whose position is determined as
a mathematical function of the input variable.
◼ The format of the Sugeno-style fuzzy rule is
IF x is A AND y is B THEN z is f (x, y)
where:
◼ x, y and z are linguistic variables;
◼ A and B are fuzzy sets on universe of discourses X and Y, respectively;
◼ f (x, y) is a mathematical function.
◼ The most common zero-order Sugeno fuzzy model applies fuzzy rules as:
IF x is A AND y is B THEN z is k
where the output membership function is a fuzzy singleton
positioned at constant value k in universe of the output variable z.
◼ In first-order Sugeno fuzzy model, fuzzy rules take form:
IF x is A AND y is B THEN z is ax+by+c
where the output fuzzy singleton’s position changes linearly with inputs applied.
14

14

7
 27.12.2023.

Sugeno Rule Evaluation

15

15

Sugeno Aggregation and

Defuzzification
MAX aggregation becomes
the sum of singleton spikes
AGG(z) = max [R1(z), R12(z), R3(z) ]
= R1(z) + R2(z) + R3(z)
= 0.1|k1 + 0.2|k2 + 0.5|k3

COG defuzzification becomes

Weighted Average (WA)
𝑛
෍ 𝜇𝐴𝐺𝐺 (𝑘𝑗 )×𝑘𝑗
𝑗=1
𝑧𝑊𝐴 = 𝑛
෍ 𝜇𝐴𝐺𝐺 (𝑘𝑗 )
𝑗=1

𝜇𝐴𝐺𝐺 𝑘1 ×𝑘1 +𝜇𝐴𝐺𝐺 𝑘2 ×𝑘2 +𝜇𝐴𝐺𝐺 (𝑘3 )×𝑘3

=
𝜇𝐴𝐺𝐺 𝑘1 +𝜇𝐴𝐺𝐺 𝑘2 +𝜇𝐴𝐺𝐺 (𝑘3 )
0.1×20+0.2×50+0.5×80
= = 65
0.1+0.2+0.5
16

16

8
 27.12.2023.

Mamdani or Sugeno?
◼ 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.
◼ On the other hand, Sugeno method cripples the fuzzy spirit to some
extent (output fuzzy sets in design are rather not fuzzy), but it is
◼ computationally effective,
◼ works well with optimization and adaptive techniques,
which makes it very attractive in ANFIS modeling and
control problems, particularly for nonlinear dynamic systems.
◼ Choosing Mamdani or Sugeno fuzzy inference is up to the
requirements of a specific application and a designer’s preference,
just like many other things in fuzzy logic. 17

17