Modelling and Identification

Prof. Dr. Ping Zhang

Institute for Automatic Control

WS 2023/24
 Organisation of this course

Chapter 1: Introduction
Chapter 2: Theoretical Modelling
Chapter 3: Experimental modelling
Chapter 4: Least-Squares methods
Chapter 5: Prediction error methods
Chapter 6: Instrumental variable methods
Chapter 7: Subspace identification methods (SS model!)
Chapter 8: Some practical aspects

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 2

 Organisation of this course

Chapter 1
Introduction

Chapter 2 Chapter 3
Theoretical Modelling Experimental Modelling

Chapter 4
Least-Squares methods

Chapter 5 Chapter 6
Prediction error methods Instrumental variable methods

Chapter 7
Subspace identification

Chapter 8
Some practical aspects

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 3

 Chapter 3
Experimental Modelling

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 4

 Step response approach – systems with
integral action
Approximation by a -th system with integral action

Step response ( )
7

0
0 2 4 6 8 10

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 5

 Step response of 2. order system
Given a second order system described by

Step response ( )
n
2
=0
1.8

1.6 =0.2

1.4

1.2 =0.5

1 =0.8
y

=1
0.8
=1.5
0.6
=2
0.4

0.2

0
0 2 4 6 8 10 12
t

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 6

 Step response of 2. order system

Step response of the underdamped system

Step response ( )
Final value:
3

2
Half period of the oscillation:

Overshoot: 0
0 20 40

 e nt 

y (t )  aK 1  sin( 1   2 n t  arccos  
 1   2 
 

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 7

 Step response approach – 2. order system
Assume that the system is approximated by a second order
system

Step response ( ) Read characteristic

3
values:
 final value
2  peak value
 time
1

0
0 20 40

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 8

 Step response approach – 2. order system

Identification procedure:

1. Calculate the gain .

2. Read the overshoot and the period from the step response
3. Calculate the damping ratio based on .
4. Calculate based on and .

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 9

 Step response approach – 2. order system

Example:
Step response ( )
3
2.8
2.6
2.4
2.2
2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 10

 Step response approach – 2. order system

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 11

 Review: Step response of 2. order system

Step response of the overdamped system

K 1
  1: T1  T2
T1s  1 T2 s  1
T2
T 
1.5 Ta  T1  2  T2 T1

 T1 
tangent line
T1T2 T
1 Tu  ln 2  Ta  T1  T2
T2  T1 T1
T
 2
0.5 T1
Inflection point 
Ta
  1
T1
0
0 10 20 30 40 Ta 1

Tu 
  
  1
1    ln    1
  1 
Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 12
 Step response approach – 2. order system

Key of identification: Nomogram for the calculation of

20

Ta/T1
18
Ta/Tu
16

14

12

10

0 
0 1 2 3 4 5 6 7 8 9 10

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 13

 Step response approach – 2. order system

Identification procedure:

1. Calculate the gain .

2. Determine the inflection point and the tangent line in the step
reponse
3. Read and

4. Determine based on according to the Nomogram (see the blue

curve)
5. Determine according to the Nomogram (see the red curve)
6. Calculate .

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 14

 Step response approach – 2. order system

Example:
Step response ( )
1.6

1.4

1.2

0.8

0.6

0.4

0.2

0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 15

 Step response approach – 2. order system

20

Ta/T1
18
Ta/Tu
16

14

12

10

0 
0 1 2 3 4 5 6 7 8 9 10

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 16

 Step response approach – 2. order system

20

Ta/T1
18
Ta/Tu
16

14

12

10

0 
0 1 2 3 4 5 6 7 8 9 10

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 17

 Küpfmüller approach

Approximation by a first order system with time delay

Step response ( )
2.5

Küpfmüller approach:
2
tangent line - Gain
1.5

1 - Time delay
Inflection point
0.5
- Time constant
0
0 20 40 60 80

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 18

 Küpfmüller approach

Example
Step response ( )
2.5

1.5

0.5

0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 19

 Strejc approach

Approximation by a -th order system with equal time constants

Step response ( )
2.5
Read characteristic
2 values:
 final value
1.5
 time
1  time

0.5

0
0 20 40 60 80

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 20

 Strejc approach

Identification procedure:

1. Calculate the gain .

2. Draw the tangent at the inflection point
3. Read and

4. Based on , read the order from the following table. A rough

estimate of can also be got by .

5. Based on , get the time constant from the table.

2 3 4 5 6 7 8 9 10
9.65 4.58 3.13 2.44 2.03 1.75 1.56 1.41 1.29
2.72 3.69 4.46 5.12 5.70 6.23 6.71 7.16 7.59
0.28 0.80 1.42 2.10 2.81 3.55 4.30 5.08 5.87

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 21

 Strejc approach

Example
Step response ( )
2.5

1.5

0.5

0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 22

 Strejc approach

Approximation by a -th system with equal time constants

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 23

 Schwarze approach

Approximation by a -th system with equal time constants

Step response ( )
2.5

2 Schwarze approach
make use of the
1.5
characteristic values:
1

0.5

0
0 20 40 60 80

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 24

 Schwarze approach

Key of identification:

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 25

 Schwarze approach

Identification procedure:
1. Get the characteristic values from the step
response

2. The gain is got by .

3. Based on , determine the order of the system .

4. Based on and one of the curves , determine the time constant .

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 26

 Schwarze approach

Example:
Step response ( )
2.5
2.4
2.2
2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0 2 4 6 8 101214161820222426283032343638404244464850525456586062646668707274767880

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 27

 Schwarze approach

Example:

The gain of TF

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 28

 Schwarze approach

Example:

The gain of TF

x
As

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 29

 Schwarze approach

Example:

The gain of TF

x
As

From Figure a it can be seen that

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 30

 Schwarze approach

Example:

From Figure b it can be seen that

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 31

 Schwarze approach

Example:

From Figure b it can be seen that

Hence, the time constant is

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 32

 Comparison of approaches

Available approaches:
- Nomogram approach
Second order system with two
real poles
- Küpfmüller approach
First order system with time
delay
- Strejc approach
N-th order system with the
same time constants
- Schwarze approach
N-th order system with the
same time constants
Step response ( )
Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 33
 Comparison of approaches

Available approaches:
- Nomogram approach
Tangent based approach
- Küpfmüller approach
Tangent based approach
- Strejc approach
Tangent based approach
- Schwarze approach
Time-Percent based
approach

Step response ( )
Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 34
 Comparison of approaches

Results of identification:
- Nomogram approach
2.4
𝐺(𝑠) =
(1 + 5.028𝑠)(1 + 8.044𝑠)

- Küpfmüller approach
2.4 .
𝐺(𝑠) = 𝑒
1 + 18.1𝑠

- Strejc approach
2.4
𝐺(𝑠) =
(1 + 6.54𝑠)

Step response ( )
Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 35
 Comparison of approaches

Results of identification:

- Schwarze approach

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 36

 Comparison of approaches

Validation of model for the above example

Scenario 1:
True
Step-response system
output
(amplitude of the step
input: 0.5)

True
system
output

Scenario 2:
Response to sinusoidal
input

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 37

 Get transfer function from frequency response

Measured frequency response

Bode Diagram
Determine the structure of the
80
70 transfer function
60 dB/decade
50
40
30
Magnitude (dB)

20
10
0
-10
-20
-30
-40
-50
-60
dB/decade
-70
-80
-90

-120
Phase (deg)

-150

-180
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/s)

Corner frequency:
Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 38
 Get transfer function from frequency response

Measured frequency response

Bode Diagram
Determine the structure of the
80
70 transfer function
60 dB/decade
50
40
30
Magnitude (dB)

20
10
0
-10
-20
-30
Read corner frequency
-40
-50
-60
dB/decade
-70
-80
-90
Hence, the time constant is

-120
Phase (deg)

-150

-180
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/s)

Corner frequency:
Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 39
 Get transfer function from frequency response

Measured frequency response

Bode Diagram
80
70
60 dB/decade Read
50
40 at frequency
30
Magnitude (dB)

20
10
0
-10
-20
-30
-40
-50
-60
dB/decade
-70
-80
-90

-120
Phase (deg)

-150

-180
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/s)

Corner frequency:
Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 40
 Get transfer function from frequency response

Example 2:
-20
Bode Diagram
Determine the structure of the
dB/decade transfer function
-25
dB/decade
Magnitude (dB)

-30

-35
dB/decade
-40
0
Phase (deg)

-30

-60
-2 -1 0 1
10 10 10 10
Frequency (rad/s)

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 41

 Summary of Chapter 3

 Measurement of non-parametric models:

• Step response
• Impulse response
• Frequency response

 Get parametric model from non-parametric model

Lehrstuhl für Automatisierungstechnik Ping Zhang, [email protected] 42