Lecture 5��State Estimation and Model Reduction for MPC

1

Instructor: Ercan Atam

Course: DSAI 586- Data-Driven Model Predictive Control

2

List of contents for this lecture

​

​

• Luenberger observer design basics

​

• Kalman filtering for LTI systems

​

• Emulator versus control models

​

• System identification and grey-box modelling

​

• POD-based model order reduction

3

State estimation for MPC (1)

4

State estimation for MPC (2)

5

Luenberger Observer (1)

6

Luenberger Observer (2)

7

Luenberger Observer (3)

8

Linear Kalman filter (1)

9

Linear Kalman filter (2)

10

Linear Kalman filter (3)

11

Linear Kalman filter (4)

12

Linear Kalman filter (5)

13

Linear Kalman filter (6)

14

Some remarks for linear Kalman filtering (1)

15

Some remarks for linear Kalman filtering (2)

16

Emulator model / Control model concepts (1)

17

Emulator Model / Control Model Concepts (2)

18

Emulator Model / Control Model Concepts (3)

19

How do we obtain control models?

20

System identification

21

Lumped-parameter modelling (1)

22

Lumped-parameter modelling (2)

23

Grey-box modelling

24

Grey-box modelling example (1)

Example:

25

Grey-box modelling example (2)

26

Model order reduction (MOR) (1)

27

Model order reduction (MOR) (2)

28

Model order reduction (MOR) (3)

29

Model order reduction (MOR) (4)

30

Some remarks for MOR

31

Some references for POD-based MOR

32

A note on accurate control model development

33

Effect of prediction horizon length (N) on stability (1)

34

Effect of prediction horizon length (N) on stability (2)

35

Back to observer design: do we always need an observer? (1)

36

Back to observer design: do we always need an observer? (2)

37

Back to observer design: do we always need an observer? (3)

38

Back to observer design: do we always need an observer? (4)

References �(utilized for preparation of lecture notes or MATLAB code)

39

​

• “Model Predictive Control System Design and Implementation Using MATLAB”, Liuping Wang, Springer 2009.

​