NEURO-FUZZY LOGIC 1 X 0 A age 1 0 20 Crisp version for young age.
-
Upload
magdalene-young -
Category
Documents
-
view
217 -
download
0
Transcript of NEURO-FUZZY LOGIC 1 X 0 A age 1 0 20 Crisp version for young age.
NEURO-FUZZY LOGIC
1
X
0
A
age
1
0
20
A
Crisp version for young age
2010 16 age
1
0
A kid The teenager
A young man A
Crisp definitions of young age
1
age
0
65
A
Crisp version of old age
1
age
0
65 70 75
Elderly Old
Oldest
A
Crisp definitions of old age
FUZZY SETS
age
1
20 6525 45
0,9
0,2
0
A
Possible fuzzy set of young age
age
1
0
65 25 16 45
0,2
0
0,8
A
Possible fuzzy set of old age
MEMBERSHIP FUNCTIONS
65age
1
0,8
0,2
0
20
4525
Young Old
A
Possible membership functions for young and old ages
IF-THEN
LINGUISTIC RULES
age
1
0
16 65
young old
middle aged
40
A
IF a man have age less than 40 years old, THEN he is a young manIF a man have age more than 40 years old, THEN he is old manIF a man have age 40 years old, THEN he is middle aged man
1
0
negative positive
zero
max min
IF a man is old and his age is more than 40 years old, THEN level of car incidents protection is high (positive)
IF a man is young and his age is less than 40 years old, THEN level of car incidents protection is low (negative)IF a man is middle-aged and his age about 40 years old, THEN level of car incidents protection is normal (zero point).
1
age
less about 40 more
32
0,7
0,2
0
min max level
0
negativezero
positive
0,7
0,2
Center of gravity
A
32 years old age less than 40 with degree 0.7;
Level of car incident protection, in this age, is negative (low) with the same degree
32 years old age is about 40 with degree 0.2;
Level of car incident protection, in this age, is normal (zero) with the same degree
Center of gravity calculation is crisp value of car incident protection level for age 32 years old.
FUZZY LOGIC CONTROL SYSTEMS
distance
1
0
less than 10cm
more than 10 cm
about 10 cm
10 cm
A
IF distance between the robot and the obstacle is less than 10 cm, THEN steer for (a) -10 degr.IF distance between the robot and the obstacle is more than 10 cm, THEN steer for( a )+10degr.IF distance between the robot and the obstacle is 10 cm, THEN go straightforward
1
0
negative positive
zero
a max a min
IF distance between the robot and the obstacle is more than 10 cm, THEN turn to the right ( a is positive)
IF distance between the robot and the obstacle is less than 10 cm, THEN turn to the left ( a is negative)
IF distance between the robot and the obstacle is nearly 10 cm, THEN keep the direction
1
distance
less
nearly 10 cm
more
5 cm
0,7
0,2
0
a min a max0
negative
zero
positive
0,7
0,2
Center of gravity
A
5cm less than 10 cm with degree 0.7;
Steering angle has to be negative with the same degree
(turn to the left)
5 cm is nearly 10 cm with degree 0.2;
Steering angle has to be normal (zero) with the same degree
(keep the direction)
Center of gravity calculation is crisp output of control value
FUZZY LOGIC CONTROLLER
Forward FastForward FastSmallBigBigBigRule 3
Backward Medium
Forwad Medium
SmallRule 2
Forwad Medium
Forwad Medium
BigBigBigBigRule 1
Right motorLeft motorBackRightFrontLeft
Motor SpeedsDistancesRules
IF the distance to the left is Big and the distance in front is Big and the distance to the right is Big and the distance on the back is Big THEN left motor speed is Forward Medium and right motor speed is Forward Medium
Forward FastForward FastSmallBigBigBigRule 3
Backward Medium
Forwad Medium
SmallRule 2
Forwad Medium
Forwad Medium
BigBigBigBigRule 1
Right motorLeft motorBackRightFrontLeft
Motor SpeedsDistancesRules
IF the distance in front is Small (other distances are not considered) THEN left motor speed is Forward Medium and right motor speed is Backward Medium.
Forward FastForward FastSmallBigBigBigRule 3
Backward Medium
Forwad Medium
SmallRule 2
Forwad Medium
Forwad Medium
BigBigBigBigRule 1
Right motorLeft motorBackRightFrontLeft
Motor SpeedsDistancesRules
IF the distance to the left is Big and distance in front is Big and distance to the right is Big and distance to the back is Small THEN left motor speed is Forward Fast and right motor speed is Forward Fast
Small Medium Big
Small Medium Big
Small Medium Big
Small Medium Big
Membership functions
Input
Leftmotor
Left motor
Right motor
Logical Operations
Membership functions
Output
Distance
Left
Front
Right
Back
AND
AND
AND
Inference
Rightmotor
FUZZY LOGIC CONTROLLER
NEURAL NETWORK CONTROL SYSTEMS
summing unit
threshold1x
2x
0w
1w
2w
0x
PERCEPTRON
0.3-0.26 -0.51 -0.77 0.31 0.2 0.2 0.15 0.26Right motor (weights)
0.3 0.2 0.2 0.3 -0.72 -0.46 -0.2 0.26 0.15Left motor (weights)
ThresholdS1 S2 S3 S4 S5 S6 S7 S8 Sensors
SAMPLE OF PERCEPTRON FOR CONTROL
S1
S2
S3
S4
S5
S6
S7
S8
Sensors Summarizing of weights
Motors
Left
Forward
Right
Back
Threshold
Left
motor
Rightmotor
Distance
SAMPLE OF PERCEPTRON NETWORK FOR CONTROL
ADAPTIVE NEURO-FUZZY CONTROLLER
FLC MLP
errordesired performance
Learning parameter performance
actual performance
output
Rule 1: IF (Gradient of Error is Negative Big) AND (Change Gradient of Error is Negative Big) THEN Change of Learning Parameters is Negative Small…………Rule 13: IF (Gradient of Error is Zero Equal) AND (Change Gradient of Error is is Zero Equal) THEN Change of Learning Parameters is Positive Small …………. Rule 25: IF( Gradient of Error is Positive Big ) AND ( Change Gradient of Error is is Positive Big) THEN Change of Learning Parameters is Negative Small
FLC- Fuzzy Logic Controller
MLP- Multilayer Perceptron