Fuzzy Control

– Configuration
– Design choices
– Takagi-Sugeno controller
 Direct Control

End-user

Controller

Rule
Ref Deviations base Actions Outputs
Plant
Inference
engine
 Building Blocks

Fuzzy controller

Rule
Prepro- Fuzzi- base Defuzzi Postpro
cessing fication -fication -
Inference cessing
engine
Nonlinear Input Scaling

100

50
scaled input

-50

-100

-5 0 5
measured input
 If-Then Rule Base

1. If error is Neg and change in error is Neg then output is NB

2. If error is Neg and change in error is Zero then output is NM
3. If error is Neg and change in error is Pos then output is Zero
4. If error is Zero and change in error is Neg then output is NM
5. If error is Zero and change in error is Zero then output is Zero
6. If error is Zero and change in error is Pos then output is PM
7. If error is Pos and change in error is Neg then output is Zero
8. If error is Pos and change in error is Zero then output is PM
9. If error is Pos and change in error is Pos then output is PB
 Relational Rule Format
Error Change in error Control
Pos Pos PB
Pos Zero PM
Pos Neg Zero
Zero Pos PM
Zero Zero Zero
Zero Neg NM
Neg Pos Zero
Neg Zero NM
Neg Neg NB
 Tabular Rule Format
Change in error

Neg Zero Pos

Neg NB NM Zero

Error Zero NM Zero PM

Pos Zero PM PB
 Connectives

A  B  min  A ( x),  B ( x)  minimum

A  B  max  A ( x),  B ( x)  maximum

A  B   A ( x)  B ( x) algebraic product
A  B   A ( x)   B ( x)   A ( x)  B ( x) probabilistic sum
 FLS I/O Families
Zero
Neg Pos
1
Membership

0.5

0
-1 -0.5 0 0.5 1
Input

1
Membership

0.5

0
-1 -0.5 0 0.5 1
Output
Examples Of Primary Sets
1 1 1 1 1

0.5 0.5 0.5 0.5 0.5

0 0 0 0 0
-100 0 100 -100 0 100 -100 0 100 -100 0 100 -100 0 100
(a) (d) (g) (j) (m)

1 1 1 1 1

0.5 0.5 0.5 0.5 0.5

0 0 0 0 0
-100 0 100 -100 0 100 -100 0 100 -100 0 100 -100 0 100
(b) (e) (h) (k) (n)

1 1 1 1 1

0.5 0.5 0.5 0.5 0.5

0 0 0 0 0
-100 0 100 -100 0 100 -100 0 100 -100 0 100 -100 0 100
(c) (f) (i) (l) (o)
Inference And Terminology
AND

a4

a5
Activation

Defuzzification

Accumulation
Aggregation
 Defuzzification

0.5

0
0 50 100

RM
LM
BOA
MOM
COG
Rule Based Controllers

1. If error is Neg then control is Neg

2. If error is Zero then control is Zero
3. If error is Pos then control is Pos
Mamdani Inference
error control
1 1

0.5 0.5

0 0
-100 0 100 -100 0 100
1 1 1

0.5 0.5 0.5

0 0 0
-100 0 100 -100 0 100 -100 0 100
1 1 u = -25.7

0.5 0.5

0 0
-100 0 100 -100 0 100
 FLS Inference
error control
1 1

0.5 0.5

0 0
-100 0 100 -100 0 100
1 1 1

0.5 0.5 0.5

0 0 0
-100 0 100 -100 0 100 -100 0 100
1 1 u = -29.7

0.5 0.5

0 0
-100 0 100 -100 0 100
 Sugeno Inference
error control
1 1

0.5 0.5

0 0
-100 0 100 -100 0 100
1 1 1

0.5 0.5 0.5

0 0 0
-100 0 100 -100 0 100 -100 0 100
1 1 u = -36.3

0.5 0.5

0 0
-100 0 100 -100 0 100
 Singleton Output

1. If error is Pos then control is 10

2. If error is Zero then control is 0
3. If error is Neg then control is -10
 First Order Output

1. If error is Pos then control is a2*error + b2

2. If error is Neg then control is a1*error + b1
Interpolation (Takagi-Sugeno)
150

100 1
output 2
50

0
0 50 100
(a)
1
membership

0.5

0
0 50 100
(b)
Rule Base To Table
 Look-Up Table
Change in error

-100 -50 0 50 100

100 0 40 100 100 200

50 -40 0 61 121 160

Error 0 -100 -61 0 61 100

-50 -100 -121 -61 0 40

-100 -200 -160 -100 -40 0

 Control Surface

200 0.8

0.6

membership
0
u

0.4
-200
100 0.2
100
0
0 0
CE -100 -100 E -100 -50 0 50 100
input family
 Linear Controller

200 0.8

0.6

membership
0
u

0.4
-200
100 0.2
100
0
0 0
CE -100 -100 E -100 -50 0 50 100
input family
Linear Rule Base
 Conditions For Linearity
• Triangular sets, crossing at  = 0.5
• Rules: complete -combination
• Define  as *
• Use conclusion singletons, positioned at sum of input
peak positions
• Use sum-accumulation and COGS defuzzification
 Simplification of 4 rules
1. If error is Neg and change in error is Neg then control is NB
3. If error is Neg and change in error is Pos then control is Zero
7. If error is Pos and change in error is Neg then control is Zero
9. If error is Pos and change in error is Pos then control is PB

is

u  ( Pos E    Pos CE   1) PB

 Simplification of 9 rules
1. If error is Neg and change in error is Neg then output is NB
2. If error is Neg and change in error is Zero then output is NM
3. If error is Neg and change in error is Pos then output is Zero
4. If error is Zero and change in error is Neg then output is NM
5. If error is Zero and change in error is Zero then output is Zero
6. If error is Zero and change in error is Pos then output is PM
7. If error is Pos and change in error is Neg then output is Zero
8. If error is Pos and change in error is Zero then output is PM
9. If error is Pos and change in error is Pos then output is PB

is

u   Pos E    Neg E    Pos CE    Neg CE  PB

1
2
 Summary Of Choices
• Rule-base related choices: # of inputs and outputs, rules,
universes, continuous or discrete, # of membership
functions, their overlap and width, singleton conclusions
• Inference engine choices: Connectives, modifiers,
activation operation, aggregation operation,
accumulation operation
• Defuzzification method: COG, COGS, BOA, MOM, LM,
RM
• Pre- and postprocessing: Scaling, quantization, sampling
time