Modelling and Control:
Lecture Notes for the Project Oriented Course in Process Control

Mikael Sternad

Report UPTEC 94025K, 130 pp, March 1994.
Earlier version, in Swedish: Report UPTEC 89004K, January 1989.

From the Preface:

The present lecture notes have been developed for use in the project oriented course in process control, within the Engineering Physics program at Uppsala University. Alternative approaches to modelling, identification and controller design are presented briefly. The focus is almost exclusively on main results and on their interpretation and practical relevance. Theoretical derivations are mostly excluded, but references are given for the interested reader.

The course ``Processreglering'' on advanced control methods is project oriented. One of its aims is to fortify knowledge that has been introduced in previous courses. This is done by active learning in a goal-oriented project, which requires a major effort by all participants. A further goal is to cover important topics which are discussed only briefly, or not at all, in the basic courses on control and identification. This includes physical modelling, LQG control, direct and indirect adaptive control, decoupling of multivariable processes, implementation aspects, the interface between process and computer, the human-computer interface, and more.

Finally, the course provides the opportunity to master programming tools which are becoming increasingly widespread in industry; the control programs are written in Matlab under Linux. Real-time facilities are provided. The user interface utilizes Netscape Navigator and it is developed in HTML and Java. Indirect adaptive controllers will be implemented as compiles Matlab routines.

Last, but not least: this is (finally) a course in which you will have time to explore and to use your creativity.

The different project variants are all centered around the control of a single type of laboratory process: the Coupled Electric Drives by Tecquipment Ltd. Different groups use different methods for modelling and controlling this process. Some of the alternatives for obtaining process models are

• physical modelling
• frequency domain identification
• parametric time domain identification
• recursive identification, used in adaptive control.

Methods for obtaining models are discussed in Part I of the notes. Part II covers some strategies for controller design. Direct adaptive control algorithms are discussed in separate lecture notes.

Contents of Part I: Modelling and Identification

1. Introduction: the laboratory process

2. Modelling

3. Modelling of the laboratory process: Physical modelling

4. Modelling of the laboratory process: Frequency domain identification

5. System identification with prediction error methods

Contents of Part II: Methods for Regulator Design

1. Introduction

2. PID Control

3. State feedback with observers

4. Linear quadratic optimization: state space methods

5. Regulator and servo synthesis by polynomial methods

6. Linear quadratic optimization: polynomial methods

7. Robustness analysis and robust design

Book Chapter related to the themes of Chapter 5 and 6 of Part II.

Matlab Conference paper describing our computer system.