Professional Documents
Culture Documents
Textbook A First Course in Predictive Control Second Edition Rossiter Ebook All Chapter PDF
Textbook A First Course in Predictive Control Second Edition Rossiter Ebook All Chapter PDF
https://textbookfull.com/product/a-first-course-in-machine-
learning-second-edition-mark-girolami/
https://textbookfull.com/product/renewable-energy-second-edition-
a-first-course-robert-ehrlich/
https://textbookfull.com/product/biota-grow-2c-gather-2c-cook-
loucas/
https://textbookfull.com/product/a-course-in-error-correcting-
codes-second-edition-jorn-justesen/
A course in categorical data analysis First Edition
Leonard
https://textbookfull.com/product/a-course-in-categorical-data-
analysis-first-edition-leonard/
https://textbookfull.com/product/a-first-course-in-sobolev-
spaces-giovanni-leoni/
https://textbookfull.com/product/a-first-course-in-statistics-
for-signal-analysis-wojbor-a-woyczynski/
https://textbookfull.com/product/a-second-course-in-topos-
quantum-theory-1st-edition-cecilia-flori-auth/
https://textbookfull.com/product/a-first-course-in-functional-
analysis-1st-edition-orr-moshe-shalit/
A First Course in
Predictive Control
Second Edition
A First Course in
Predictive Control
Second Edition
J.A. Rossiter
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2018 by Taylor & Francis Group, LLC
CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works
Printed on acid-free paper
Version Date: 20180322
International Standard Book Number-13: 978-1-138-09934-0 (Hardback)
This book contains information obtained from authentic and highly regarded sources. Reasonable
efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information
storage or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access
www.copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc.
(CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization
that provides licenses and registration for a variety of users. For organizations that have been granted
a photocopy license by the CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and
are used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
In the hope that students and lecturing staff will
find this easy to learn from and in the spirit of sharing,
creating a world that accepts and supports each other.
Contents
Acknowledgements xxiii
vii
viii Contents
2.9.1
The need for numerically robust prediction with open-loop
unstable plant . . . . . . . . . . . . . . . . . . . . . . . . 73
2.9.2 Pseudo-closed-loop prediction . . . . . . . . . . . . . . . . 74
2.9.3 Illustration of prediction structures with the OLP and CLP . 76
2.9.4 Basic CLP predictions for state space models . . . . . . . . 77
2.9.5 Unbiased closed-loop prediction with autonomous models . 78
2.9.6 CLP predictions with transfer function models . . . . . . . 79
2.10 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 81
2.11 Summary of MATLAB code supporting prediction . . . . . . . . . 82
4.5.6
Closed-loop transfer functions . . . . . . . . . . . . . . . . 136
4.5.7
GPC based on MFD models with a T-filter (GPCT) . . . . . 137
4.5.7.1 Why use a T-filter and conceptual thinking? . . . 137
4.5.7.2 Algebraic procedures with a T-filter . . . . . . . . 138
4.5.8 Sensitivity of GPC . . . . . . . . . . . . . . . . . . . . . . 141
4.5.8.1 Complementary sensitivity . . . . . . . . . . . . 142
4.5.8.2 Sensitivity to multiplicative uncertainty . . . . . . 142
4.5.8.3 Disturbance and noise rejection . . . . . . . . . . 143
4.5.8.4 Impact of a T-filter on sensitivity . . . . . . . . . 144
4.5.9 Analogies between PFC and GPC . . . . . . . . . . . . . . 145
4.6 GPC formulation for finite impulse response models and
Dynamic Matrix Control . . . . . . . . . . . . . . . . . . . . . . . 146
4.7 Formulation of GPC with an independent prediction model . . . . . 148
4.7.1 GPC algorithm with independent transfer function model . . 148
4.7.2 Closed-loop poles in the IM case with an MFD model . . . 151
4.8 GPC with a state space model . . . . . . . . . . . . . . . . . . . . 153
4.8.1 Simple state augmentation . . . . . . . . . . . . . . . . . . 153
4.8.1.1 Computing the predictive control law with an
augmented state space model . . . . . . . . . . . 154
4.8.1.2 Closed-loop equations and integral action . . . . 156
4.8.2 GPC using state space models with deviation variables . . . 156
4.8.2.1 GPC algorithm based on deviation variables . . . 157
4.8.2.2 Using an observer to estimate steady-state values
for the state and input . . . . . . . . . . . . . . . 160
4.8.3 Independent model GPC using a state space model . . . . . 161
4.9 Chapter summary and general comments on stability and tuning of
GPC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
4.10 Summary of MATLAB code supporting GPC simulation . . . . . . 163
4.10.1 MATLAB code to support GPC with a state space model and
a performance index based on deviation variables . . . . . . 163
4.10.2 MATLAB code to support GPC with an MFD or CARIMA
model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
4.10.3 MATLAB code to support GPC using an independent model
in MFD format. . . . . . . . . . . . . . . . . . . . . . . . . 165
4.10.4 MATLAB code to support GPC with an augmented state
space model. . . . . . . . . . . . . . . . . . . . . . . . . . 165
7.8.2 MATLAB code for supporting GPC based on an MFD model 281
7.8.3 MATLAB code for supporting GPC based on an independent
model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
9 Conclusions 329
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
9.2 Design choices . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
9.2.1 Scenario and funding . . . . . . . . . . . . . . . . . . . . . 330
9.2.2 Effective tuning . . . . . . . . . . . . . . . . . . . . . . . . 330
9.2.3 Constraint handling and feasibility . . . . . . . . . . . . . . 331
9.2.4 Coding complexity . . . . . . . . . . . . . . . . . . . . . . 332
Contents xvii
References 383
Index 397
Overview and guidance for use
The main aim is to create a focussed and affordable textbook which is suitable for a
single or beginners course on model predictive control (MPC). As a consequence:
1. In the early parts of each chapter, an attempt is made not to dwell too
much on the mathematical details and proofs of interest to researchers
but not to a typical student and instead focus as much as possible on the
concepts and understanding needed to apply the method effectively.
2. Many books and resources on MPC cover the basic concepts far too
briefly and focus on more advanced issues. This book does the opposite
and explains the basics slowly and with numerous examples.
3. Numerous illustrative examples are included throughout and also prob-
lems which encourage readers to engage with and understand the con-
cepts.
4. The topic coverage is deliberately limited to a range considered sufficient
for taught modules.
5. Chapters are supported by MATLAB R files which are available on the
open web (or with the book) so that readers can implement basic designs
on-linear models. Numerous student problems encourage readers to learn
by doing/designing.
6. In some cases short videos are available which support key parts of this
book. [http://controleducation.group.shef.ac.uk/indexwebbook.html]
7. Chapters include guidance for the lecturer to help highlight which sec-
tions can be used for a taught module and which could be considered
broadening but non-essential.
Insight is given in a non-theoretical way and there are a number of summary
boxes to give a quick picture of the key results without the need to read through the
detailed explanation. There is a strong focus on the philosophy of predictive control
answering the questions, ‘why?’ and ‘does it help me?’ The basic concepts are in-
troduced and then these are developed to fit different purposes: for instance, how to
model, to predict, to tune, to handle constraints, to ensure feasibility, to guarantee
stability and to consider what options there are with regard to models, algorithms,
complexity versus performance, and so forth.
Research students who want to study predictive control in more depth are advised
to make use of the research literature which is very extensive, but even for them I
hope they find the focus on concepts in this book will be an invaluable foundation.
xix
xx Overview and guidance for use
Websites:
http://controleducation.group.shef.ac.uk/indexwebbook.html
https://www.sheffield.ac.uk/acse/staff/jar
https://www.youtube.com/channel/UCMBXZxd-j6VqrynykO1dURw
Chapter 1: Gives basic motivation and background. This chapter explores, fairly
concisely, control problems where classical approaches are difficult to apply or
obviously suboptimal. It then gives some insight into how prediction forms a
logical foundation with which to solve many such problems and the key compo-
nents required to form a systematic control law.
Chapter 2: Considers how predictions are formed for a number of different model
types. Predictions are a main building block of MPC and thus this is a core skill
users will need. Also explains the concept of unbiased prediction and why this
is important.
Chapter 3: Introduces a very simple to implement and widely applied MPC law,
that is predictive functional control (PFC). Gives numerous illustrative examples
to help readers understand tuning, where PFC is a plausible solution and where
a more expensive MPC approach is more appropriate.
Chapter 4: Introduces the most common MPC performance index/optimisation
used in industrial applications and shows how these are combined with a pre-
diction model to form an effective control law.
Chapter 5: Gives insights and guidance on tuning finite horizon MPC laws. How
do I ensure that I get sensible answers and well-posed decision making? What
is the impact of uncertainty?
Chapter 6: Considers the stability of MPC and shows how a more systematic anal-
ysis of MPC suggests the use of so-called dual-mode predictions and infinite
horizons as these give much stronger performance and stability assurances. In-
troduces the algebra behind these types of MPC approaches.
Chapter 7: Considers constraint handling and why this is an important part of prac-
tical MPC algorithms. Shows how constraints are embedded systematically into
finite horizon MPC algorithms.
Chapter 8: Considers constraint handling with infinite horizon algorithms. It is
noted that while such approaches have more rigour, they are also more compli-
cated to implement and moreover are more likely to encounter feasibility chal-
lenges.
Chapter 9: Gives a concise conclusion and some indication of how a user might
both choose from and design with the variety of MPC approaches available.
xxi
xxii Book organisation
xxiii
1
Introduction and the industrial need for
predictive control
CONTENTS
1.1 Guidance for the lecturer/reader . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Motivation and introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Classical control assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3.1 PID compensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3.2 Lead and Lag compensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.3 Using PID and lead/lag for SISO control . . . . . . . . . . . . . . . . . . . 4
1.3.4 Classical control analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Examples of systems hard to control effectively with classical
methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4.1 Controlling systems with non-minimum phase zeros . . . . . . . 5
1.4.2 Controlling systems with significant delays . . . . . . . . . . . . . . . . 7
1.4.3 ILLUSTRATION: Impact of delay on margins and
closed-loop behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4.4 Controlling systems with constraints . . . . . . . . . . . . . . . . . . . . . . . 10
1.4.5 Controlling multivariable systems . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.4.6 Controlling open-loop unstable systems . . . . . . . . . . . . . . . . . . . 16
1.5 The potential value of prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.5.1 Why is predictive control logical? . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.5.2 Potential advantages of prediction . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.6 The main components of MPC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.6.1 Prediction and prediction horizon . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.6.2 Why is prediction important? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.6.3 Receding horizon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.6.4 Predictions are based on a model . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.6.5 Performance indices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.6.6 Degrees of freedom in the predictions or prediction class . . . 27
1.6.7 Tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.6.8 Constraint handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.6.9 Multivariable and interactive systems . . . . . . . . . . . . . . . . . . . . . . 30
1.6.10 Systematic use of future demands . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.7 MPC philosophy in summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.8 MATLAB files from this chapter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
1
2 A first course in predictive control: 2nd Edition
Consequently, although this book includes technical details, the underlying fo-
cus is on readers understanding the concepts and thus how to ensure the algorithm
and tuning they choose is likely to lead to an effective control law for their specific
context.
Section 1.3 will give a concise review of some classical control approaches be-
fore Section 1.4 demonstrates a number of scenarios where such approaches are dif-
ficult to tune effectively, or suboptimal. Section 1.5 then introduces arguments for a
predictive control approach, followed by Section 1.6 which uses analysis of human
behaviour to set out solid principles before we move into mathematical detail. The
overall MPC philosophy is summarised in Section 1.7 followed by Section 1.9 which
gives a concise summary of the structure of the book.
too fast it is highly likely that the control is too aggressive and a reduction
in input is needed to avoid overshoot and oscillation. However, the deriva-
tive is often selected to be zero as it has a high gain at high frequency and
thus can accentuate noise and lead to input chatter. Such discussions are
beyond the remit of this book.
Several tuning rules for PID are given in the literature but actually, for a typical
system, one could arrive at close to optimum values with very little trial and error
using a simulation package.
Terços esdruccioles.
BERARDO
Tauriso, el fresco viento, que
alegrándonos
murmura entre los árboles
altíssimos,
la vista y los oídos
deleitándonos;
Las chozas y sombríos
ameníssimos;
las cristalinas fuentes, que
abundancia
derraman de licores
sabrosíssimos;
La colorada flor, cuya
fragrancia
á despedir bastara la
tristicia,
que hace al corazón más
fiera instancia:
No vencen la braveza y la
malicia
del crudo rey, tan áspero y
mortífero,
cuyo castigo es pura sin
justicia.
Ningún remedio ha sido
salutífero
á mi dolor, pues siempre
enbraveciéndose
está el veneno y tóxico
pestífero.
TAURISO
Al que en amores anda
consumiéndose,
nada le alegrará: porque
fatígale
tal mal, que en el dolor vive
muriéndose.
Amor le da más penas, y
castígale,
cuando en deleites anda
recreándose,
porque él á suspirar contino
oblígale.
Las veces que está un ánima
alegrándose,
le ofresce allí un dolor, cuya
memoria
hace que luego vuelva á
estar quejándose.
Amor quiere gozar de su
victoria,
y al hombre que venció,
mátale ó préndele,
pensando en ello haber
famosa gloria.
El preso á la fortuna entrega, y
véndele
al gran dolor, que siempre
está matándole,
y al que arde en más
ardiente llama enciéndele.
BERARDO
El sano vuelve enfermo,
maltratándole,
y el corazón alegre hace
tristíssimo,
matando el vivo, el libre
captivándole.
Pues, alma, ya que sabes
cuán bravíssimo
es este niño Amor, sufre y
conténtate
con verte puesta en un lugar
altíssimo.
Rescibe los dolores, y
preséntate
al daño que estuviere
amenazándote,
goza del mal y en el dolor
susténtate.
Porque cuanto más fueres
procurándote
medio para salir de tu
miseria,
irás más en los lazos
enredándote.
TAURISO
En mí halla Cupido más
materia
para su honor, que en
cuantos lamentándose
guardan ganado en una y
otra Hesperia.
Siempre mis males andan
aumentándose,
de lágrimas derramo mayor
copia
que Biblis cuando en fuente
iba tornándose.
Extraño me es el bien, la pena
propia,
Diana, quiero ver, y en vella
muérome,
junto al tesoro estó, y muero
de inopia.
Si estoy delante della, peno y
quiérome
morir de sobresalto y de
cuidado,
y cuando estoy ausente,
desespérome.
BERARDO
Murmura el bosque y ríe el
verde prado,
y cantan los parleros
ruiseñores;
mas yo en dos mil tristezas
sepultado.
TAURISO
Espiran suave olor las tiernas
flores,
la hierba reverdesce al
campo ameno;
mas yo viviendo en ásperos
dolores.
BERARDO
El grave mal de mí me tiene
ajeno,
tanto que no soy bueno
para tener diez versos de
cabeza.
TAURISO
Mi lengua en el cantar siempre
tropieza,
por esso, amigo, empieza,
algún cantar de aquellos
escogidos,
los cuales estorbados con
gemidos,
con lloro entrerompidos,
te hicieron de pastores
alabado.
BERARDO
En el cantar contigo
acompañado,
iré muy descansado;
respóndeme. Mas no sé qué
me cante.
TAURISO
Di la que dice: Estrella
radiante,
ó la de: O triste amante,
ó aquella: No sé como se
decía,
que la cantaste un día
bailando con Diana en el
aldea.
BERARDO
No hay tigre ni leona que no
sea
á compassión movida
de mi fatiga extraña y
peligrosa;
mas no la fiera hermosa,
fiera devoradora de mi vida.
TAURISO
Fiera devoradora de mi vida,
¿quién si no tú estuviera
con la dureza igual á la
hermosura?
y en tanta desventura
¿cómo es possible, ay triste,
que no muera?
BERARDO
¿Cómo es possible, ay triste,
que no muera?
dos mil veces muriendo;
¿mas cómo he de morir
viendo á Diana?
El alma tengo insana:
cuanto más trato Amor,
menos le entiendo.
TAURISO
Cuanto más trato Amor,
menos le entiendo,
que al que le sirve mata,
y al que huyendo va de su
cadena,
con redoblada pena
las míseras entrañas le
maltrata.
BERARDO
Pastora, á quien el alto cielo
ha dado
beldad más que á las rosas
coloradas,
más linda que en Abril el
verde prado,
do están las florecidas
matizadas,
ansí prospere el cielo tu
ganado,
y tus ovejas crezcan á
manadas,
que á mí, que á causa tuya
gimo y muero,
no me muestres el gesto
airado y fiero.
TAURISO
Pastora soberana, que
mirando
los campos y florestas
asserenas,
la nieve en la blancura
aventajando
y en la beldad las frescas
azucenas,
ansí tus campos vayan
mejorando,
y dellos cojan fruto á manos
llenas,
que mires á un pastor, que
en solo verte
piensa alcanzar muy
venturosa suerte.