Model Predictive Control: Theory, Computation,

and Design, 2nd Edition Rawlings James B. pdf

download

https://ebookmass.com/product/model-predictive-control-theory-
computation-and-design-2nd-edition-rawlings-james-b/

Explore and download more ebooks at ebookmass.com

We have selected some products that you may be interested in
Click the link to download now or visit ebookmass.com
for more options!.

Process Control: Modeling, Design, and Simulation 2nd

Edition B. Wayne Bequette

https://ebookmass.com/product/process-control-modeling-design-and-
simulation-2nd-edition-b-wayne-bequette/

Predictive Control Daniel Lequesne

https://ebookmass.com/product/predictive-control-daniel-lequesne/

Model Predictive Control for Doubly-Fed Induction

Generators and Three-Phase Power Converters Alfeu J.
Sguarezi Filho
https://ebookmass.com/product/model-predictive-control-for-doubly-fed-
induction-generators-and-three-phase-power-converters-alfeu-j-
sguarezi-filho/

Surface Area and Porosity Determinations by Physisorption:

Measurement, Classical Theories and Quantum Theory 2nd
Edition James B. Condon
https://ebookmass.com/product/surface-area-and-porosity-
determinations-by-physisorption-measurement-classical-theories-and-
quantum-theory-2nd-edition-james-b-condon/
Philosophy and Model Theory Tim Button

https://ebookmass.com/product/philosophy-and-model-theory-tim-button/

(eBook PDF) Data Mining and Predictive Analytics 2nd

Edition

https://ebookmass.com/product/ebook-pdf-data-mining-and-predictive-
analytics-2nd-edition/

Engineering Computation: An Introduction Using MATLAB and

Exce, 2nd Edition Musto

https://ebookmass.com/product/engineering-computation-an-introduction-
using-matlab-and-exce-2nd-edition-musto/

Tidal Energy Systems: Design, Optimization and Control 1st

Edition

https://ebookmass.com/product/tidal-energy-systems-design-
optimization-and-control-1st-edition/

Organization theory and design 13th Edition Daft

https://ebookmass.com/product/organization-theory-and-design-13th-
edition-daft/
Model Predictive Control:
Theory, Computation, and Design
2nd Edition

ISBN 9780975937730

9 780975 937730
 Model Predictive Control:
Theory, Computation, and Design
2nd Edition

James B. Rawlings
Department of Chemical Engineering
University of California
Santa Barbara, California, USA

David Q. Mayne
Department of Electrical and Electronic Engineering
Imperial College London
London, England

Moritz M. Diehl
Department of Microsystems Engineering and

D
Department of Mathematics
University of Freiburg
Freiburg, Germany

b Hi
No ll Publishing

Santa Barbara, California

This book was set in Lucida using LATEX, and printed and bound by
Worzalla. It was printed on Forest Stewardship CouncilÉ certified acid-
free recycled paper.

Cover design by Cheryl M. and James B. Rawlings.

Copyright c 2019 by Nob Hill Publishing, LLC

All rights reserved.

Nob Hill Publishing, LLC

Cheryl M. Rawlings, publisher
Santa Barbara, CA 93117
[email protected]
http://www.nobhillpublishing.com

No part of this book may be reproduced, in any form or by any means,

without permission in writing from the publisher.

Library of Congress Control Number: 2017909542

Printed in the United States of America.

First Edition
First Printing August 2009
Electronic Download November 2013
Electronic Download (2nd) April 2014
Electronic Download (3rd) July 2014
Electronic Download (4th) October 2014
Electronic Download (5th) February 2015
Second Edition
First Printing October 2017
Electronic Download October 2018
Electronic Download (2nd) February 2019
 To Cheryl, Josephine, and Stephanie,

for their love, encouragement, and patience.

Preface to the Second Edition

In the eight years since the publication of the first edition, the field
of model predictive control (MPC) has seen tremendous progress. First
and foremost, the algorithms and high-level software available for solv-
ing challenging nonlinear optimal control problems have advanced sig-
nificantly. For this reason, we have added a new chapter, Chapter 8,
“Numerical Optimal Control,” and coauthor, Professor Moritz M. Diehl.
This chapter gives an introduction into methods for the numerical so-
lution of the MPC optimization problem. Numerical optimal control
builds on two fields: simulation of differential equations, and numeri-
cal optimization. Simulation is often covered in undergraduate courses
and is therefore only briefly reviewed. Optimization is treated in much
more detail, covering topics such as derivative computations, Hessian
approximations, and handling inequalities. Most importantly, the chap-
ter presents some of the many ways that the specific structure of opti-
mal control problems arising in MPC can be exploited algorithmically.
We have also added a software release with the second edition of
the text. The software enables the solution of all of the examples and
exercises in the text requiring numerical calculation. The software is
based on the freely available CasADi language, and a high-level set of
Octave/MATLAB functions, MPCTools, to serve as an interface to CasADi.
These tools have been tested in several MPC short courses to audiences
composed of researchers and practitioners. The software can be down-
loaded from www.chemengr.ucsb.edu/~jbraw/mpc.
In Chapter 2, we have added sections covering the following topics:
• economic MPC
• MPC with discrete actuators
We also present a more recent form of suboptimal MPC that is prov-
ably robust as well as computationally tractable for online solution of
nonconvex MPC problems.
In Chapter 3, we have added a discussion of stochastic MPC, which
has received considerable recent research attention.
In Chapter 4, we have added a new treatment of state estimation
with persistent, bounded process and measurement disturbances. We
have also removed the discussion of particle filtering. There are two

vi
 vii

reasons for this removal; first, we wanted to maintain a manageable

total length of the text; second, all of the available sampling strate-
gies in particle filtering come up against the “curse of dimensionality,”
which renders the state estimates inaccurate for dimension higher than
about five. The material on particle filtering remains available on the
text website.
In Chapter 6, we have added a new section for distributed MPC of
nonlinear systems.
In Chapter 7, we have added the software to compute the critical
regions in explicit MPC.
Throughout the text, we support the stronger KL-definition of asymp-
totic stability, in place of the classical definition used in the first edition.
The most significant notational change is to denote a sequence with
(a, b, c, . . .) instead of with {a, b, c, . . .} as in the first edition.

JBR DQM MMD

Madison, Wis., USA London, England Freiburg, Germany
Preface

Our goal in this text is to provide a comprehensive and foundational

treatment of the theory and design of model predictive control (MPC).
By now several excellent monographs emphasizing various aspects of
MPC have appeared (a list appears at the beginning of Chapter 1), and
the reader may naturally wonder what is offered here that is new and
different. By providing a comprehensive treatment of the MPC foun-
dation, we hope that this text enables researchers to learn and teach
the fundamentals of MPC without continuously searching the diverse
control research literature for omitted arguments and requisite back-
ground material. When teaching the subject, it is essential to have a
collection of exercises that enables the students to assess their level of
comprehension and mastery of the topics. To support the teaching and
learning of MPC, we have included more than 200 end-of-chapter exer-
cises. A complete solution manual (more than 300 pages) is available
for course instructors.
Chapter 1 is introductory. It is intended for graduate students in en-
gineering who have not yet had a systems course. But it serves a second
purpose for those who have already taken the first graduate systems
course. It derives all the results of the linear quadratic regulator and
optimal Kalman filter using only those arguments that extend to the
nonlinear and constrained cases to be covered in the later chapters.
Instructors may find that this tailored treatment of the introductory
systems material serves both as a review and a preview of arguments
to come in the later chapters.
Chapters 2–4 are foundational and should probably be covered in
any graduate level MPC course. Chapter 2 covers regulation to the ori-
gin for nonlinear and constrained systems. This material presents in a
unified fashion many of the major research advances in MPC that took
place during the last 20 years. It also includes more recent topics such
as regulation to an unreachable setpoint that are only now appearing in
the research literature. Chapter 3 addresses MPC design for robustness,
with a focus on MPC using tubes or bundles of trajectories in place of
the single nominal trajectory. This chapter again unifies a large body of
research literature concerned with robust MPC. Chapter 4 covers state
estimation with an emphasis on moving horizon estimation, but also

viii
 ix

covers extended and unscented Kalman filtering, and particle filtering.

Chapters 5–7 present more specialized topics. Chapter 5 addresses
the special requirements of MPC based on output measurement instead
of state measurement. Chapter 6 discusses how to design distributed
MPC controllers for large-scale systems that are decomposed into many
smaller, interacting subsystems. Chapter 7 covers the explicit optimal
control of constrained linear systems. The choice of coverage of these
three chapters may vary depending on the instructor’s or student’s own
research interests.
Three appendices are included, again, so that the reader is not sent
off to search a large research literature for the fundamental arguments
used in the text. Appendix A covers the required mathematical back-
ground. Appendix B summarizes the results used for stability analysis
including the various types of stability and Lyapunov function theory.
Since MPC is an optimization-based controller, Appendix C covers the
relevant results from optimization theory. In order to reduce the size
and expense of the text, the three appendices are available on the web:
www.chemengr.ucsb.edu/~jbraw/mpc. Note, however, that all mate-
rial in the appendices is included in the book’s printed table of contents,
and subject and author indices. The website also includes sample ex-
ams, and homework assignments for a one-semester graduate course
in MPC. All of the examples and exercises in the text were solved with
Octave. Octave is freely available from www.octave.org.

JBR DQM
Madison, Wisconsin, USA London, England
Acknowledgments

Both authors would like to thank the Department of Chemical and Bio-
logical Engineering of the University of Wisconsin for hosting DQM’s
visits to Madison during the preparation of this monograph. Funding
from the Paul A. Elfers Professorship provided generous financial sup-
port.
JBR would like to acknowledge the graduate students with whom
he has had the privilege to work on model predictive control topics:
Rishi Amrit, Dennis Bonné, John Campbell, John Eaton, Peter Findeisen,
Rolf Findeisen, Eric Haseltine, John Jørgensen, Nabil Laachi, Scott Mead-
ows, Scott Middlebrooks, Steve Miller, Ken Muske, Brian Odelson, Mu-
rali Rajamani, Chris Rao, Brett Stewart, Kaushik Subramanian, Aswin
Venkat, and Jenny Wang. He would also like to thank many colleagues
with whom he has collaborated on this subject: Frank Allgöwer, Tom
Badgwell, Bhavik Bakshi, Don Bartusiak, Larry Biegler, Moritz Diehl,
Jim Downs, Tom Edgar, Brian Froisy, Ravi Gudi, Sten Bay Jørgensen,
Jay Lee, Fernando Lima, Wolfgang Marquardt, Gabriele Pannocchia, Joe
Qin, Harmon Ray, Pierre Scokaert, Sigurd Skogestad, Tyler Soderstrom,
Steve Wright, and Robert Young.
DQM would like to thank his colleagues at Imperial College, espe-
cially Richard Vinter and Martin Clark, for providing a stimulating and
congenial research environment. He is very grateful to Lucien Polak
and Graham Goodwin with whom he has collaborated extensively and
fruitfully over many years; he would also like to thank many other col-
leagues, especially Karl Åström, Roger Brockett, Larry Ho, Petar Koko-
tovic, and Art Krener, from whom he has learned much. He is grateful
to past students who have worked with him on model predictive con-
trol: Ioannis Chrysochoos, Wilbur Langson, Hannah Michalska, Sasa
Raković, and Warren Schroeder; Hannah Michalska and Sasa Raković,
in particular, contributed very substantially. He owes much to these
past students, now colleagues, as well as to Frank Allgöwer, Rolf Find-
eisen Eric Kerrigan, Konstantinos Kouramus, Chris Rao, Pierre Scokaert,
and Maria Seron for their collaborative research in MPC.
Both authors would especially like to thank Tom Badgwell, Bob Bird,
Eric Kerrigan, Ken Muske, Gabriele Pannocchia, and Maria Seron for
their careful and helpful reading of parts of the manuscript. John Eaton

x
Visit https://ebookmass.com today to explore
a vast collection of ebooks across various
genres, available in popular formats like
PDF, EPUB, and MOBI, fully compatible with
all devices. Enjoy a seamless reading
experience and effortlessly download high-
quality materials in just a few simple steps.
Plus, don’t miss out on exciting offers that
let you access a wealth of knowledge at the
best prices!
 xi

again deserves special mention for his invaluable technical support dur-
ing the entire preparation of the manuscript.
Added for the second edition. JBR would like to acknowledge the
most recent generation of graduate students with whom he has had the
privilege to work on model predictive control research topics: Doug Al-
lan, Travis Arnold, Cuyler Bates, Luo Ji, Nishith Patel, Michael Risbeck,
and Megan Zagrobelny.
In preparing the second edition, and, in particular, the software re-
lease, the current group of graduate students far exceeded expectations
to help finish the project. Quite simply, the project could not have been
completed in a timely fashion without their generosity, enthusiasm,
professionalism, and selfless contribution. Michael Risbeck deserves
special mention for creating the MPCTools interface to CasADi, and
updating and revising the tools used to create the website to distribute
the text- and software-supporting materials. He also wrote code to cal-
culate explicit MPC control laws in Chapter 7. Nishith Patel made a
major contribution to the subject index, and Doug Allan contributed
generously to the presentation of moving horizon estimation in Chap-
ter 4.
A research leave for JBR in Fall 2016, again funded by the Paul A.
Elfers Professorship, was instrumental in freeing up time to complete
the revision of the text and further develop computational exercises.
MMD wants to especially thank Jesus Lago Garcia, Jochem De Schut-
ter, Andrea Zanelli, Dimitris Kouzoupis, Joris Gillis, Joel Andersson,
and Robin Verschueren for help with the preparation of exercises and
examples in Chapter 8; and also wants to acknowledge the following
current and former team members that contributed to research and
teaching on optimal and model predictive control at the Universities of
Leuven and Freiburg: Adrian Bürger, Hans Joachim Ferreau, Jörg Fis-
cher, Janick Frasch, Gianluca Frison, Niels Haverbeke, Greg Horn, Boris
Houska, Jonas Koenemann, Attila Kozma, Vyacheslav Kungurtsev, Gio-
vanni Licitra, Rien Quirynen, Carlo Savorgnan, Quoc Tran-Dinh, Milan
Vukov, and Mario Zanon. MMD also wants to thank Frank Allgöwer, Al-
berto Bemporad, Rolf Findeisen, Larry Biegler, Hans Georg Bock, Stephen
Boyd, Sébastien Gros, Lars Grüne, Colin Jones, John Bagterp Jørgensen,
Christian Kirches, Daniel Leineweber, Katja Mombaur, Yurii Nesterov,
Toshiyuki Ohtsuka, Goele Pipeleers, Andreas Potschka, Sebastian Sager,
Johannes P. Schlöder, Volker Schulz, Marc Steinbach, Jan Swevers, Phil-
ippe Toint, Andrea Walther, Stephen Wright, Joos Vandewalle, and Ste-
fan Vandewalle for inspiring discussions on numerical optimal control
xii

methods and their presentation during the last 20 years.

All three authors would especially like to thank Joel Andersson and
Joris Gillis for having developed CasADi and for continuing its support,
and for having helped to improve some of the exercises in the text.
Contents

1 Getting Started with Model Predictive Control 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Models and Modeling . . . . . . . . . . . . . . . . . . . . . . 1
1.2.1 Linear Dynamic Models . . . . . . . . . . . . . . . . . 2
1.2.2 Input-Output Models . . . . . . . . . . . . . . . . . . 3
1.2.3 Distributed Models . . . . . . . . . . . . . . . . . . . 4
1.2.4 Discrete Time Models . . . . . . . . . . . . . . . . . . 5
1.2.5 Constraints . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.6 Deterministic and Stochastic . . . . . . . . . . . . . 9
1.3 Introductory MPC Regulator . . . . . . . . . . . . . . . . . . 11
1.3.1 Linear Quadratic Problem . . . . . . . . . . . . . . . 11
1.3.2 Optimizing Multistage Functions . . . . . . . . . . . 12
1.3.3 Dynamic Programming Solution . . . . . . . . . . . 18
1.3.4 The Infinite Horizon LQ Problem . . . . . . . . . . . 21
1.3.5 Controllability . . . . . . . . . . . . . . . . . . . . . . 23
1.3.6 Convergence of the Linear Quadratic Regulator . . 24
1.4 Introductory State Estimation . . . . . . . . . . . . . . . . . 26
1.4.1 Linear Systems and Normal Distributions . . . . . 27
1.4.2 Linear Optimal State Estimation . . . . . . . . . . . 29
1.4.3 Least Squares Estimation . . . . . . . . . . . . . . . 33
1.4.4 Moving Horizon Estimation . . . . . . . . . . . . . . 39
1.4.5 Observability . . . . . . . . . . . . . . . . . . . . . . . 41
1.4.6 Convergence of the State Estimator . . . . . . . . . 43
1.5 Tracking, Disturbances, and Zero Offset . . . . . . . . . . 46
1.5.1 Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . 46
1.5.2 Disturbances and Zero Offset . . . . . . . . . . . . . 49
1.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

2 Model Predictive Control—Regulation 89

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
2.2 Model Predictive Control . . . . . . . . . . . . . . . . . . . . 91
2.3 Dynamic Programming Solution . . . . . . . . . . . . . . . 107
2.4 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
2.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 112

xiii
xiv Contents

2.4.2 Stabilizing Conditions . . . . . . . . . . . . . . . . . 114

2.4.3 Exponential Stability . . . . . . . . . . . . . . . . . . 120
2.4.4 Controllability and Observability . . . . . . . . . . . 120
2.4.5 Time-Varying Systems . . . . . . . . . . . . . . . . . 123
2.5 Examples of MPC . . . . . . . . . . . . . . . . . . . . . . . . . 131
2.5.1 The Unconstrained Linear Quadratic Regulator . . 132
2.5.2 Unconstrained Linear Periodic Systems . . . . . . . 133
2.5.3 Stable Linear Systems with Control Constraints . 134
2.5.4 Linear Systems with Control and State Constraints 136
2.5.5 Constrained Nonlinear Systems . . . . . . . . . . . 139
2.5.6 Constrained Nonlinear Time-Varying Systems . . . 141
2.6 Is a Terminal Constraint Set Xf Necessary? . . . . . . . . 144
2.7 Suboptimal MPC . . . . . . . . . . . . . . . . . . . . . . . . . 147
2.7.1 Extended State . . . . . . . . . . . . . . . . . . . . . . 150
2.7.2 Asymptotic Stability of Difference Inclusions . . . 150
2.8 Economic Model Predictive Control . . . . . . . . . . . . . 153
2.8.1 Asymptotic Average Performance . . . . . . . . . . 155
2.8.2 Dissipativity and Asymptotic Stability . . . . . . . 156
2.9 Discrete Actuators . . . . . . . . . . . . . . . . . . . . . . . . 160
2.10 Concluding Comments . . . . . . . . . . . . . . . . . . . . . 163
2.11 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
2.12 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172

3 Robust and Stochastic Model Predictive Control 193

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
3.1.1 Types of Uncertainty . . . . . . . . . . . . . . . . . . 193
3.1.2 Feedback Versus Open-Loop Control . . . . . . . . 195
3.1.3 Robust and Stochastic MPC . . . . . . . . . . . . . . 200
3.1.4 Tubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
3.1.5 Difference Inclusion Description of Uncertain Sys-
tems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
3.2 Nominal (Inherent ) Robustness . . . . . . . . . . . . . . . . 204
3.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 204
3.2.2 Difference Inclusion Description of Discontinu-
ous Systems . . . . . . . . . . . . . . . . . . . . . . . 206
3.2.3 When Is Nominal MPC Robust? . . . . . . . . . . . . 207
3.2.4 Robustness of Nominal MPC . . . . . . . . . . . . . 209
3.3 Min-Max Optimal Control: Dynamic Programming Solution 214
3.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 214
3.3.2 Properties of the Dynamic Programming Solution 216
Contents xv

3.4 Robust Min-Max MPC . . . . . . . . . . . . . . . . . . . . . . 220

3.5 Tube-Based Robust MPC . . . . . . . . . . . . . . . . . . . . 223
3.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 223
3.5.2 Outer-Bounding Tube for a Linear System with
Additive Disturbance . . . . . . . . . . . . . . . . . . 224
3.5.3 Tube-Based MPC of Linear Systems with Additive
Disturbances . . . . . . . . . . . . . . . . . . . . . . . 228
3.5.4 Improved Tube-Based MPC of Linear Systems with
Additive Disturbances . . . . . . . . . . . . . . . . . 234
3.6 Tube-Based MPC of Nonlinear Systems . . . . . . . . . . . 236
3.6.1 The Nominal Trajectory . . . . . . . . . . . . . . . . 238
3.6.2 Model Predictive Controller . . . . . . . . . . . . . . 238
3.6.3 Choosing the Nominal Constraint Sets Ū and X̄ . . 242
3.7 Stochastic MPC . . . . . . . . . . . . . . . . . . . . . . . . . . 246
3.7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 246
3.7.2 Stabilizing Conditions . . . . . . . . . . . . . . . . . 248
3.7.3 Stochastic Optimization . . . . . . . . . . . . . . . . 248
3.7.4 Tube-Based Stochastic MPC for Linear Constrained
Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 249
3.8 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
3.9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262

4 State Estimation 269

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
4.2 Full Information Estimation . . . . . . . . . . . . . . . . . . 269
4.2.1 State Estimation as Optimal Control of Estimate
Error . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
4.2.2 Duality of Linear Estimation and Regulation . . . . 281
4.3 Moving Horizon Estimation . . . . . . . . . . . . . . . . . . 283
4.3.1 Zero Prior Weighting . . . . . . . . . . . . . . . . . . 283
4.3.2 Nonzero Prior Weighting . . . . . . . . . . . . . . . . 287
4.3.3 Constrained Estimation . . . . . . . . . . . . . . . . 294
4.3.4 Smoothing and Filtering Update . . . . . . . . . . . 295
4.4 Bounded Disturbances . . . . . . . . . . . . . . . . . . . . . 300
4.5 Other Nonlinear State Estimators . . . . . . . . . . . . . . . 308
4.5.1 Particle Filtering . . . . . . . . . . . . . . . . . . . . . 308
4.5.2 Extended Kalman Filtering . . . . . . . . . . . . . . . 309
4.5.3 Unscented Kalman Filtering . . . . . . . . . . . . . . 310
4.5.4 EKF, UKF, and MHE Comparison . . . . . . . . . . . 312
4.6 On combining MHE and MPC . . . . . . . . . . . . . . . . . 318
xvi Contents

4.7 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325

4.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327

5 Output Model Predictive Control 339

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
5.2 A Method for Output MPC . . . . . . . . . . . . . . . . . . . 341
5.3 Linear Constrained Systems: Time-Invariant Case . . . . 344
5.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 344
5.3.2 State Estimator . . . . . . . . . . . . . . . . . . . . . . 344
5.3.3 Controlling x̂ . . . . . . . . . . . . . . . . . . . . . . . 346
5.3.4 Output MPC . . . . . . . . . . . . . . . . . . . . . . . . 348
5.3.5 Computing the Tightened Constraints . . . . . . . 352
5.4 Linear Constrained Systems: Time-Varying Case . . . . . 353
5.5 Offset-Free MPC . . . . . . . . . . . . . . . . . . . . . . . . . 353
5.5.1 Estimation . . . . . . . . . . . . . . . . . . . . . . . . . 355
5.5.2 Control . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
5.5.3 Convergence Analysis . . . . . . . . . . . . . . . . . 360
5.6 Nonlinear Constrained Systems . . . . . . . . . . . . . . . . 363
5.7 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
5.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366

6 Distributed Model Predictive Control 369

6.1 Introduction and Preliminary Results . . . . . . . . . . . . 369
6.1.1 Least Squares Solution . . . . . . . . . . . . . . . . . 370
6.1.2 Stability of Suboptimal MPC . . . . . . . . . . . . . . 375
6.2 Unconstrained Two-Player Game . . . . . . . . . . . . . . . 380
6.2.1 Centralized Control . . . . . . . . . . . . . . . . . . . 382
6.2.2 Decentralized Control . . . . . . . . . . . . . . . . . 383
6.2.3 Noncooperative Game . . . . . . . . . . . . . . . . . 384
6.2.4 Cooperative Game . . . . . . . . . . . . . . . . . . . . 392
6.2.5 Tracking Nonzero Setpoints . . . . . . . . . . . . . . 398
6.2.6 State Estimation . . . . . . . . . . . . . . . . . . . . . 405
6.3 Constrained Two-Player Game . . . . . . . . . . . . . . . . 406
6.3.1 Uncoupled Input Constraints . . . . . . . . . . . . . 408
6.3.2 Coupled Input Constraints . . . . . . . . . . . . . . 411
6.3.3 Exponential Convergence with Estimate Error . . . 413
6.3.4 Disturbance Models and Zero Offset . . . . . . . . 415
6.4 Constrained M-Player Game . . . . . . . . . . . . . . . . . . 419
6.5 Nonlinear Distributed MPC . . . . . . . . . . . . . . . . . . . 421
6.5.1 Nonconvexity . . . . . . . . . . . . . . . . . . . . . . . 421
6.5.2 Distributed Algorithm for Nonconvex Functions . 423
Contents xvii

6.5.3 Distributed Nonlinear Cooperative Control . . . . 425

6.5.4 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . 428
6.6 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430
6.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435

7 Explicit Control Laws for Constrained Linear Systems 451

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
7.2 Parametric Programming . . . . . . . . . . . . . . . . . . . . 452
7.3 Parametric Quadratic Programming . . . . . . . . . . . . . 457
7.3.1 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . 457
7.3.2 Preview . . . . . . . . . . . . . . . . . . . . . . . . . . 458
7.3.3 Optimality Condition for a Convex Program . . . . 459
7.3.4 Solution of the Parametric Quadratic Program . . 462
7.3.5 Continuity of V 0 (·) and u0 (·) . . . . . . . . . . . . 466
7.4 Constrained Linear Quadratic Control . . . . . . . . . . . . 467
7.5 Parametric Piecewise Quadratic Programming . . . . . . . 469
7.6 DP Solution of the Constrained LQ Control Problem . . . 475
7.7 Parametric Linear Programming . . . . . . . . . . . . . . . 476
7.7.1 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . 476
7.7.2 Minimizer u0 (x) is Unique for all x ∈ X . . . . . . 478
7.8 Constrained Linear Control . . . . . . . . . . . . . . . . . . 481
7.9 Computation . . . . . . . . . . . . . . . . . . . . . . . . . . . 482
7.10 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
7.11 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484

8 Numerical Optimal Control 491

8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491
8.1.1 Discrete Time Optimal Control Problem . . . . . . 492
8.1.2 Convex Versus Nonconvex Optimization . . . . . . 493
8.1.3 Simultaneous Versus Sequential Optimal Control 496
8.1.4 Continuous Time Optimal Control Problem . . . . 498
8.2 Numerical Simulation . . . . . . . . . . . . . . . . . . . . . . 501
8.2.1 Explicit Runge-Kutta Methods . . . . . . . . . . . . . 502
8.2.2 Stiff Equations and Implicit Integrators . . . . . . . 506
8.2.3 Implicit Runge-Kutta and Collocation Methods . . 507
8.2.4 Differential Algebraic Equations . . . . . . . . . . . 511
8.2.5 Integrator Adaptivity . . . . . . . . . . . . . . . . . . 513
8.3 Solving Nonlinear Equation Systems . . . . . . . . . . . . . 513
8.3.1 Linear Systems . . . . . . . . . . . . . . . . . . . . . . 513
8.3.2 Nonlinear Root-Finding Problems . . . . . . . . . . 514
8.3.3 Local Convergence of Newton-Type Methods . . . 517
xviii Contents

8.3.4 Affine Invariance . . . . . . . . . . . . . . . . . . . . . 519

8.3.5 Globalization for Newton-Type Methods . . . . . . 519
8.4 Computing Derivatives . . . . . . . . . . . . . . . . . . . . . 520
8.4.1 Numerical Differentiation . . . . . . . . . . . . . . . 521
8.4.2 Algorithmic Differentiation . . . . . . . . . . . . . . 522
8.4.3 Implicit Function Interpretation . . . . . . . . . . . 523
8.4.4 Algorithmic Differentiation in Forward Mode . . . 526
8.4.5 Algorithmic Differentiation in Reverse Mode . . . 528
8.4.6 Differentiation of Simulation Routines . . . . . . . 531
8.4.7 Algorithmic and Symbolic Differentiation Software 533
8.4.8 CasADi for Optimization . . . . . . . . . . . . . . . . 533
8.5 Direct Optimal Control Parameterizations . . . . . . . . . 536
8.5.1 Direct Single Shooting . . . . . . . . . . . . . . . . . 538
8.5.2 Direct Multiple Shooting . . . . . . . . . . . . . . . . 540
8.5.3 Direct Transcription and Collocation Methods . . 544
8.6 Nonlinear Optimization . . . . . . . . . . . . . . . . . . . . . 548
8.6.1 Optimality Conditions and Perturbation Analysis 549
8.6.2 Nonlinear Optimization with Equalities . . . . . . . 552
8.6.3 Hessian Approximations . . . . . . . . . . . . . . . . 553
8.7 Newton-Type Optimization with Inequalities . . . . . . . 556
8.7.1 Sequential Quadratic Programming . . . . . . . . . 557
8.7.2 Nonlinear Interior Point Methods . . . . . . . . . . 558
8.7.3 Comparison of SQP and Nonlinear IP Methods . . 560
8.8 Structure in Discrete Time Optimal Control . . . . . . . . 561
8.8.1 Simultaneous Approach . . . . . . . . . . . . . . . . 562
8.8.2 Linear Quadratic Problems (LQP) . . . . . . . . . . . 564
8.8.3 LQP Solution by Riccati Recursion . . . . . . . . . . 564
8.8.4 LQP Solution by Condensing . . . . . . . . . . . . . 566
8.8.5 Sequential Approaches and Sparsity Exploitation 568
8.8.6 Differential Dynamic Programming . . . . . . . . . 570
8.8.7 Additional Constraints in Optimal Control . . . . . 572
8.9 Online Optimization Algorithms . . . . . . . . . . . . . . . 573
8.9.1 General Algorithmic Considerations . . . . . . . . . 574
8.9.2 Continuation Methods and Real-Time Iterations . 577
8.10 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 580
8.11 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583

Author Index 601

Citation Index 608

Contents xix

Subject Index 614

A Mathematical Background 624

A.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 624
A.2 Vector Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . 624
A.3 Range and Nullspace of Matrices . . . . . . . . . . . . . . . 624
A.4 Linear Equations — Existence and Uniqueness . . . . . . 625
A.5 Pseudo-Inverse . . . . . . . . . . . . . . . . . . . . . . . . . . 625
A.6 Partitioned Matrix Inversion Theorem . . . . . . . . . . . . 628
A.7 Quadratic Forms . . . . . . . . . . . . . . . . . . . . . . . . . 629
A.8 Norms in Rn . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
A.9 Sets in Rn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
A.10Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
A.11Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 633
A.12Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 636
A.13Convex Sets and Functions . . . . . . . . . . . . . . . . . . . 641
A.13.1 Convex Sets . . . . . . . . . . . . . . . . . . . . . . . . 641
A.13.2 Convex Functions . . . . . . . . . . . . . . . . . . . . 646
A.14Differential Equations . . . . . . . . . . . . . . . . . . . . . . 648
A.15Random Variables and the Probability Density . . . . . . 654
A.16Multivariate Density Functions . . . . . . . . . . . . . . . . 659
A.16.1 Statistical Independence and Correlation . . . . . . 668
A.17Conditional Probability and Bayes’s Theorem . . . . . . . 672
A.18Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 678

B Stability Theory 693

B.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 693
B.2 Stability and Asymptotic Stability . . . . . . . . . . . . . . 696
B.3 Lyapunov Stability Theory . . . . . . . . . . . . . . . . . . . 700
B.3.1 Time-Invariant Systems . . . . . . . . . . . . . . . . 700
B.3.2 Time-Varying, Constrained Systems . . . . . . . . . 707
B.3.3 Upper bounding K functions . . . . . . . . . . . . . 709
B.4 Robust Stability . . . . . . . . . . . . . . . . . . . . . . . . . 709
B.4.1 Nominal Robustness . . . . . . . . . . . . . . . . . . 709
B.4.2 Robustness . . . . . . . . . . . . . . . . . . . . . . . . 711
B.5 Control Lyapunov Functions . . . . . . . . . . . . . . . . . . 713
B.6 Input-to-State Stability . . . . . . . . . . . . . . . . . . . . . 717
B.7 Output-to-State Stability and Detectability . . . . . . . . . 719
B.8 Input/Output-to-State Stability . . . . . . . . . . . . . . . . 720
B.9 Incremental-Input/Output-to-State Stability . . . . . . . . 722
B.10 Observability . . . . . . . . . . . . . . . . . . . . . . . . . . . 722
xx Contents

B.11 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 724

C Optimization 729
C.1 Dynamic Programming . . . . . . . . . . . . . . . . . . . . . 729
C.1.1 Optimal Control Problem . . . . . . . . . . . . . . . 731
C.1.2 Dynamic Programming . . . . . . . . . . . . . . . . . 733
C.2 Optimality Conditions . . . . . . . . . . . . . . . . . . . . . 737
C.2.1 Tangent and Normal Cones . . . . . . . . . . . . . . 737
C.2.2 Convex Optimization Problems . . . . . . . . . . . . 741
C.2.3 Convex Problems: Polyhedral Constraint Set . . . 743
C.2.4 Nonconvex Problems . . . . . . . . . . . . . . . . . . 745
C.2.5 Tangent and Normal Cones . . . . . . . . . . . . . . 746
C.2.6 Constraint Set Defined by Inequalities . . . . . . . 750
C.2.7 Constraint Set; Equalities and Inequalities . . . . . 753
C.3 Set-Valued Functions and Continuity of Value Function . 755
C.3.1 Outer and Inner Semicontinuity . . . . . . . . . . . 757
C.3.2 Continuity of the Value Function . . . . . . . . . . . 759
C.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 767
Visit https://ebookmass.com today to explore
a vast collection of ebooks across various
genres, available in popular formats like
PDF, EPUB, and MOBI, fully compatible with
all devices. Enjoy a seamless reading
experience and effortlessly download high-
quality materials in just a few simple steps.
Plus, don’t miss out on exciting offers that
let you access a wealth of knowledge at the
best prices!
Exploring the Variety of Random
Documents with Different Content
 Temperament Credulity,
Enthusiasm,
Superstition,
Timidity,
Imagination,
Poetic frenzy.

Excitement Sympathy,
Exalted joy,
Deep grief,
Love,
Hatred,
Protracted anxiety,
Delirium of fever,
Delirium of alcohol,
Delirium of narcotics,
Exhaustion,
Disease of the brain.
The second class, which are spectres or ghosts of the eye, may
be scientifically explained by the laws which govern the material
world. These are the only substantial ghosts which I can grant to my
friend. The objects themselves exist, and are exactly as they appear.
The philosopher regards them as interesting exceptions to general
rules, from peculiar combinations of natural causes. The unlearned
will term them preternatural phenomena, simply because they are of
uncommon occurrence. But which among the works of divine
creation is not a phenomenon? We may think we know a law of
nature, but can we analyse it? Novelty and magnitude astonish, but
that which is familiar excites not our surprise. We gaze with delight
on the progress of an eclipse; we watch with wonder the eccentric
course of the comet; but we look on the sun in its meridian glory
with a cold and apathetic indifference. Yet do they all alike display
Divine Omnipotence, and the expansion of a vegetable germ, the
bursting of a flower, is as great a miracle as the overwhelming of a
deluge, the annihilation of a mighty world.
To discriminate between these classes is not difficult: we may
prove their nature by simple experiment. Optical illusions will be
doubled by a straining or altering of the axes of the eyes; and, by
turning round, as they are removed from the axis of vision, they will
disappear.
So, indeed, will those of the second class, which are real objects
converted into phantoms by mental excitement or disorder.
But in the purely metaphysical ghost or phantom, the change of
position or locality will not essentially dispel the illusion, (the
spectrum following, as it were, the motion of the eye;) because it
exists in the mind itself, either as a faint or transient idea, or a mere
outline, fading perhaps in a brighter light, or as the more permanent
and confirmed impression of insanity, (unchanged even by “brilliant
glare,”) or from the day-dream of the castle-builder, to the deep and
dreadful delusion of the maniac.
Among the mute productions of nature, there are eccentricities
and rarities, which, in default of analysis or explanation, would not
fail of being referred to some supernatural agency: as Leo Afer,
according to Burton, accounts for the swarms of locusts once
descending at Fez, in Barbary, and at Arles, in France, in 1553. “It
could not be from natural causes; they cannot imagine whence they
come, but from heaven. Are these and such creatures, corn, wood,
stones, worms, wool, blood, &c. lifted up into the middle region by
the sunbeams, as Baracellus the physician disputes, and thence let
fall with showers, or there engendered? Cornelius Gemma is of that
opinion, they are there conceived by celestial influences: others
suppose they are immediately from God, or prodigies raised by arts
and illusions of spirits which are princes of the ayre.”
Over Languedoc there once burst an awful and supernatural
cloud, from which fell immense snow-flakes like glittering stars.
There is nothing strange in this, for the shape of the snow-flake is
ever that of an asteroïd. But then there came pouring down gigantic
hail-stones, with their glassy surface impressed with the figures of
helmets, and swords, and scutcheons. This too may be the effect of
very sudden and irregular congelation; but this law was not known,
and therefore its result was a mystery.
Among the wonders seen by the great traveller, Pietro della Valla,
was the bleeding cypress-tree, which shadows the tomb of Cyrus, in
Italy. Under the hollow of its boughs, in his day, it was lighted with
lamps and was consecrated as an oratory. To this shrine resorted
many a devout pilgrim, impressed with a holy belief in the miracle.
And what was this but the glutinous crimson fluid, exuding from the
diseased alburnum of a tree, which the woodmen indeed term
bleeding, but which the ancient Turks affirmed, or believed, to be
converted on every Friday into drops of real blood?
The red snow, which is not uncommon in the arctic regions, is
thus tinted by very minute cryptogamic plants; and the fairy ring is
but a circle of herbage poisoned by a fungus.
In Denbighshire (I may add) the prevalent belief is, that the
shivering of the aspen is from sympathy with that tree in Palestine,
which was hewn into the true cross.
The simple stratification of vapours, especially during sudden
transitions of temperature, may produce very interesting optical
phenomena; not by refraction or reflection, but merely by partial
obscuration of an object. We have examples of these illusive spectra
in the gigantic icebergs seen by Captain Scoresby, and other arctic
voyagers, which assumed the shape of towers, and spires, and
cathedrals, and obelisks, that were constantly displacing each other
in whimsical confusion and endless variety, like the figures of a
kaleidoscope. Phipps thus describes their majestic beauty: “The ice
that had parted from the main body, they had now time to admire,
as it no longer obstructed their course; the various shapes in which
the broken fragments appeared were indeed very curious and
amusing. One remarkable piece described a magnificent arch, so
large and completely formed, that a sloop of considerable burden
might have sailed through it without lowering her masts. Another
represented a church, with windows, pillars, and domes.”
We may scarcely wonder at the mystifications of nature, when
she assumes these gorgeous eccentricities, as have been witnessed
also in the barren steppes of the Caraccas, on the Orinoko, where
the palm-groves appear to be cut asunder; in the Llanos, where
chains of hills appear suspended in the air, and rivers and lakes to
flow on arid sand; in the lake of the Gazelles, seen by the Arabs and
the African traveller; and the lakes seen by Captain Munday, during
his tour in India.
The very clearness of the atmosphere, like that which floats
around the Rhine, renders distance especially distinct; but
mountainous regions, from the attraction of electric clouds, afford
the highest examples of atmospheric beauty and effect. London and
other cities, however crowded with lofty buildings, are not deficient
in these aërial illusions. Even from the bridge of Blackfriars I have
seen a cumulo stratus cloud so strangely intersect the steeples and
the giant chimneys of London, as distinctly to represent a sea-port,
with its vessels and distant mountains.
We have among us several minor illusions, which are only less
imposing because more familiar; and though often occurring, few
are recorded with scientific accuracy. The phosphorescence of the
marshes, the ignis fatuus, Will o’Wisp, Jack o’ the Lanthorn, or Friar
Rush, and the corpse-candles, are mere luminous exhalations,
strained into the marvellous by the vulgar, and thus set down as
heralds of mortality. The dancing-light of luminous flies has been
termed the green light of death; and, if you wish for more,
Astrophel, read the “Armorican Magazine” of John Wesley, or the
quaint volume of Burton, and thereabouts where he writes in this
fashion: “The thickness of the aire may cause such effects, or any
object not well discerned in the dark, fear and phantasie will suspect
to be a ghost or devil. Glowwormes, firedrakes, meteors, ignis
fatuus, which Plinius calls Castor and Pollux, with many such that
appear in moorish grounds, about church-yards, moist valleys, or
where battles have been fought, the causes of which read in
Goclenius, Velcurius, Finkius, &c.”
The Parhelia, or mock suns, are produced by the reflection of the
sun’s light on a frozen cloud. How readily these phenomena are
magnified you may learn from ancient and modern records. In 1223
four suns were seen of crimson, inclosed in a wide circle of crystal
colour. This is natural: but then comes the miracle. In the same year
two giant dragons were seen in the air, flapping their monstrous
wings and engaging in single combat, until they both fell into the
sea and were drowned! Then, in 1104, there were seen four white
circles rolled around the sun: and in 1688, two suns and a reversed
rainbow appeared at Bishop’s Lavington, in Wiltshire: and in
February, 1647, there is an account and sketch of three suns, and an
inverted rainbow, which Baxter terms “Binorum Pareliorum
Φαινομενον.” And because there were two lunar and one solar
eclipses in 1652, it was called, as Lily records, “Annus tenebrum,” or
“the dark year.”
The Corona, or halo around the sun, moon, and stars, is easily
illustrated by the zone formed by placing, during a frost, a lighted
candle in a cloud of steam or vapour.
The Aurora Borealis is arctic electricity, and is beautifully imitated
by the passage of an electric flash through an exhausted glass
cylinder.
The rainbow is a combination of natural prisms breaking the light
into colours; and it may be seen in the cloud, or in the spray of the
ocean, or in the beautiful cascades of Schaffhausen, Niagara, or
Terni, or indeed in any foaming spray on which the meridian
sunbeams fall, or even in the dewy grass, lying, as it were, on the
ground.
When the sun shines on a cloud, there is always a bow produced
visible to all who are placed at the proper angle. The lunar rainbow
is achromatic, or destitute of colour, because reflected light is not
easily refracted into colour. In a brilliant sunset the floods of light
around him often indicate the gradation of prismatic colouring.
Cast. In some waterfalls I have seen the Iris form a complete
circle; as in the Velino at Terni, and in others, especially in Ionia and
Italy. A perfect illusion is produced, for the bow seems to approach
the spectator and then recede, as if Juno were sending her
messenger on some special mission. There are many minds which
would yield with delight to this conviction, and such probably was
the illusion of Benvenuto Cellini—was it not? “This resplendent light
is to be seen over my shadow till two o’clock in the afternoon, and it
appears to the greatest advantage when the grass is moist with
dew. It is likewise visible in the evening at sunset. This phenomenon
I took notice of when I was at Paris, because the air is exceedingly
clear in that climate, so that I could distinguish it there much plainer
than in Italy, where the moists are much more frequent, &c.” A
consciousness of superior talent, and probably the homage which
was paid him even by the members of the holy conclave, were the
springs of this flattering vision.
Ida. The beauty of these must light up even the fancy of a child,
yet a holier feeling will ever inspire a Christian philosopher, when the
bow is seen in the cloud, for it was the sign of the covenant. There
is, indeed, something in the glories of the firmament which never
fails to elevate my own thoughts, and I can readily sympathise with
the Spanish religionists of the fifteenth century, and with the North
Americans, who gaze upon the beautiful constellation of the
“Southern Cross,” insulated as it is from all other stars in its own
dark space; in solemn belief that it is the great symbolical banner
held out by the Deity in approval of their faith.
Ev. The “Fata Morgana,” in the straits of Reggio, presents a
perfect scene of enchantment; when the shouts of “Morgana,
Morgana,” echo from rock and mountain, as the wondering people
flock in crowds to the shore. During this splendid illusion, gigantic
columns, and cloud-capt towers, and gorgeous palaces, and solemn
temples, are floating on the verge of the horizon, and sometimes
beneath this picture of a city, on the very bosom of the water, a
fainter spectrum may be seen, which is a reflected image of the
other. These spectra are usually colourless, but if certain watery
vapours are floating in the air, they are beautifully fringed with the
three primitive colours of the prism. Such also is the illusion of the
Calenture, or sylvan scenes of the ocean.
Cast. Let us seek these wonders of the waters, Astrophel;
perchance we might, in some enchanted hour, see even beneath yon
Severn flood the grotto of Sabrina, with its green and silver weeds,
its purple shells and arborescent corallines; and, if we dive into the
depths of the sea, might we not light on the palace of Amphitrite,
and, while the Nereids and Tritons were mourning over the
desolation of a shipwreck, hear the echo of some Ariel’s song “full
fathom five,” undulating through the water; or realise the
overwhelming of Maha-Velipoor, in the curse of Kehama:
 “Their golden summits in the noontide ray,
Shone o’er the dark green deep, that roll’d between;
For domes, and pinnacles, and spires were seen
Peering above the sea.”

Or the legend of Thierna Na Oge, in Lough Neagh, in Ireland; for

Moore has sung —

“On Lough Neagh’s banks, when the fisherman strays,

He sees the round towers of other days;”

and why may not we?

Who that has wandered among the dark mountains of Brecon,
remembers not the blue pool of Lynsavaddon, and has not listened
to the tales of the mountaineers, of the city over which to this day
its waves are rolling? and in the beautiful vale of Eidournion, in
Merioneth—but listen to a fragment of a romance of this valley,
which from memory I quote: —
There was a proud and wealthy prince in Gwyneth, when the
beautiful isle was under the rule of the Cymri. At his palace gate a
voice was once heard echoing among the mountains these words:
‘Edivar a ddau’—Repentance will come. The prince demanded
‘When?’ and in the rolling thunder the voice was again heard, ‘At the
third generation.’
Nothing daunted, the wicked lord lived on, committing plunder
and all evil excesses, and laughing to scorn the holy hymns in the
churches. A son and heir was born to him, and there was a gorgeous
assemblage in the hall of beautiful ladies and high-born nobles, to
celebrate the festival of his birth.
It was midnight, when in the ear of an old harper, a shrill voice
whispered, ‘Edivar, Edivar;’ and a little bird hovered over him, and
flew out of the palace in the pale moonshine: and the harper and
the little bird went together into the mountains. The bird flitted
before him in the centre of the moon’s disc, and warbled its
mournful cry of ‘Edivar’ so plaintively, that the old man thought of
the shriek of his little child Gwenhwyvar, as she sunk beneath the
waters of Glaslyn.
On the top of the mountain he sank down with weariness, and
the little bird was not with him; all was silent, save the cataract and
the sheep-bells on the mountain side. In alarm at the wild solitude
around him, he turned towards the castle, but its lordly towers had
vanished, and in the place of its woods and turrets there was a
waste of rolling waters—with his lone harp floating on their surface.
Ev. I am unwilling to check your flight, fair Castaly, but my
illustrations are not yet exhausted.
The “Spectre of the Brocken” is a mere shadow of the spectator
on a gigantic scale. This phantom, the “Schattenmann,” according to
vulgar tradition, haunts the lofty range of the Hartz mountains, in
Hanover. It is usually observed when the sun’s rays are thrown
horizontally on thin fleecy clouds, or vapour of highly reflective
power, assuming the shape of a gigantic shade on the cloud.
The romantic region of the Hartz was the grand temple of Saxon
idolatry, the very hot-bed of terrible shadows; the first of May
especially being the grand annual rendezvous of unearthly forms.
Even now, it is affirmed, Woden, known in Brunswick as the Hunter
of Hackelburgh, (whose sepulchre, an immense rough stone, is
shown to the traveller,) is still influential in the Oden Wald and
among the ruins of Rodenstein: even as in our own Lancashire, a
dark gigantic horseman rushes on a giant steed in stormy nights,
over “Horrock Moor;” indeed, a spot or tomb is still shown where he
used to disappear.
Thus are the “Spectres of the Brocken” invested with
supernatural dignity, in the minds of credulity and ignorance. And no
wonder, for, although the discoverer of this gigantic illusion, Mr.
Jordan, might convince the Germans of the nature of this shadow,
how could the credulous believe, when they beheld a second figure,
a faint refracted spectrum of the shadow, that it was any other than
the shadow king of the Brocken himself, frowning defiance on
intruders.
And this reminds me of the confession of Gaffarel, in his
“Unheard of Curiosities” of the seventeenth century; in his quaint
chapter on the “readynge of the cloudes and whatever else is seene
in the air, and of hieroglyphicks in the cloudes.”
Among other miraculous illusions, as recorded by Cardanus, “An
angel once wafted on the cloudes above Millane, and great was the
consternation at its appearance, until Pellicanus, a philosopher, made
it plainly appear, that this angel was nothing else but the reflection
of an image of stone, that was on the top of the church of Saint
Godart, which was represented in the thick cloudes as in a looking-
glasse.”
While I was in South Wales, in 1836, I conversed with a labourer
in the Cyfarthfa works at Merthyr Tydvil, an illiterate seer, who saw,
three times appearing before him, an unsubstantial tram-road; and
on it a train drawn by a horse, and in this, the dead body of a man.
Twice this shadow emerged from the earth, and on the third ascent
he looked on it, and recognized the well-known face of a comrade.
The man was horror struck, but his friend lived to laugh at him.
When my friend, Mr. David Taylor, ascended the mountain that
rises over Chamouni, on the opposite side of the valley to Mont
Blanc, his magnified shadow was distinctly seen by him on the
vapoury cloud that floated between these giant rocks.
In February 1837, two gentlemen, on whom I confidently trust,
were standing on Calton Hill while a murky cloud hung over
Edinburgh. Above this veil Arthur’s Seat peeped out like a rocky
island beneath two white arches, like the lunar bows; and on the
cloud itself, each gentleman saw the shadow of his companion
magnified to gigantic proportions.
The aeronaut, among other glories of his ascent, may by chance
be gratified by the shadow of his balloon on the face of a cumulus
cloud; thus did the Duke of Brunswick, who ascended with Mrs.
Graham, in August 1836. And this is the analogous recital of Prince
Puckler Muskau, in his “Tutti Frutti.”
“We dipped insensibly into the sea of clouds which enveloped us
like a thick veil, and through which the sun appeared like the moon
in Ossian. This illumination produced a singular effect, and continued
for some time, till the clouds separated, and we remained swimming
about beneath the once more clear azure heavens. Shortly after we
beheld, to our great astonishment, a species of ‘Fata Morgana,’
seated upon an immense mountain of clouds the colossal picture of
the balloon and ourselves surrounded by myriads of variegated
rainbow tints. A full half hour the spectral reflected picture hovered
constantly by our side. Each slender thread of the network appeared
distended to the size of a ship’s cable, and we ourselves two
tremendous giants enthroned on the clouds.”
The phantom, which rode side by side with Turpin, might be a
mere reflected shadow in the mist; indeed, Burton writes that
“Vitellio hath such another instance of a familiar acquaintance of his,
that after the want of three or four nights’ sleep, as he was riding by
a river-side, saw another riding with him, and using all such gestures
as he did, but when more light appeared, it vanished.”
The principles of refraction are the sources of many an illusion,
which is startling even to those who are aware of them. The sea, the
vessels floating on its surface, the rocks and buildings on its shores,
often appear elevated far beyond their usual position: things are
thus presented to the eye which, in the direct course of the rays,
would be completely out of sight; and the praises bestowed on the
Irish telescope may not have been a bull, although we are assured
that we may see through it round the corner.
Baron Humboldt, Mr. Huddart, Professor Vince, Captain Scoresby
and others, will entertain you with these natural eccentricities, if you
read the learned letter of Sir David Brewster, on “Natural Magic;”
and he will teach you how easy is the solution of all these marvels,
on the principles of atmospheric reflection. Yet how many are there
who are not contented with the light of our philosophy, though it
may fall like a sunbeam on the mind. Like the recorder of the
“Unheard-of Curiosities,” they, at one time, confess the optical
illusion, as when the Romans “saw their navy in the clouds;” at
another, as when Constantine professed to see the “Crosse shining
most gloriously in the aire,” marked with the motto, “In hoc signo
vinces,”—philosophy was silent, and they believed it might be divine.
But a mind in its state of nature cannot know all this. If a savage
looked on the two white horses cut on the chalk hills of Berkshire
and of Wiltshire, on the white cross of the Saxons on the Bledlow
Ridge in Buckinghamshire, and on the white-leaf cross near Princes
Risboro’,—would he not deem them deities, or the work of a
magician or a devil?
When the sailors of Lord Nelson saw the bloated corpse of the
murdered Prince Caraccioli floating erect in the water directly
towards their ship, can we wonder they should deem it a
supernatural visitation?
When Franklin set his bells a ringing, by drawing down the
electric fluid from the thunder-cloud, and when Columbus foretold to
the hour the sun’s eclipse;—can we wonder that the transatlantic
Indians listened, as to one endued with preternatural knowledge, or
that the other might be thought superhuman? And when the king of
Siam was assured that water could be congealed into ice, on which
the sounding skate could glide,—can we wonder that he smiled in
absolute disbelief of such a change, and called the tale a lie.
Thus, when the peasants of Cardigan, who were not versed in
Pontine architecture, looked on the bridge which the monks of
Yspitty C’en Vaen had thrown across the torrent of the Monach, they
could not believe it a work of human, but of infernal, hands, and
called it the “Devil’s Bridge.”
On my ascent of the Vann mountain in Brecon, there often came
a mass of limestone rolling down the precipice. “Ah sure,” said the
old shepherd, who was watching his fold on the mountain-side, “the
fairies are at their gambols, master, for they sometimes do play at
bowls with these chalk stones.” Such was his explanation; but, on
my gaining another ridge of the Brecon Beacon, I startled a whole
herd of these fairies, who scudded off as fast as their legs could
carry them, having first changed themselves into a flock of sheep.
There was once a caravan journeying from Nubia to Cairo, which
met the Savans attending on the expedition of Napoleon into Egypt,
among whom was Rigo, the painter. Struck with the deep character
of expression in the face of one of the Nubians, Rigo induced him,
with gold, to sit for his portrait. The African sat calmly perusing its
progress until the laying on of the colours, when, with a cry of terror,
he rushed from the house, and, to his awe-struck companions,
affirmed that his head and half his body had been cut off by an
enchanter. And this impression was not solitary, for an assemblage
of the Nubians were equally terror-struck, and (somewhat like those
monomaniacs who refuse to drink water which reflects their faces,
believing that they are swallowing their friends,) could never be
dispossessed of the notion that the picture was formed of the
loppings and toppings of the human frame.
We believe these influences the more, because we see that, even
to some few men wiser than they, a leaning to superstition will warp
a simple fact into a wonder; and that mere sensitiveness of mind
may work as great a fear.
Suetonius tells us that Caligula and Augustus were the most
abject cowards in a thunder-storm; and the bishop of Langres
D’Escaro fell in a fainting-fit whenever an eclipse took place,—a
weakness which at length proved his death.
There was an old house in Angoulême, the “Chateau du Diable,”
on the spot where the sable fiend was wont to repair to enjoy his
moonlight walk. The house was never finished, for the devil, jealous
of this usurpation, like Michael Scott’s spirit, destroyed every night
the walls which had been erected during the day. At length the men
abandoned their work in despair. On the twenty-fifth night in May
(1840), the ruined windows seemed on an instant in brilliant
illumination, which struck the inhabitants of the little village of “Petit-
Rochford” with wonder and dismay. Some dauntless heroes,
however, sallied forth with weapons to storm the enchanted castle.
In an upper room, lighted by eight blood-red wax candles, they
discovered a man of a strange and melancholy aspect, tracing
cabalistic figures on the sanded floor. He was conveyed to the maire,
and was proved to be a poor sawyer, named Favreau, who, bound
by a superstitious oath, self-administered, had thus created a
sensation of terror throughout a whole community.
In the records of the Harleian Miscellany, the curious reader may
discover one which might impress his mind with some terrific ideas
of the natural history of the south of England in the seventeenth and
eighteenth centuries. It is styled, “The True and Wonderful.” The
portion of the MSS. to which I allude is the “Legend of the Serpent
of St. Leonard’s Forest.” This terrific legend of my own native town
was a favourite of my boyish days; it has moulted some feather of
its once awful interest, and is now but the shadow of a memory; and
those who were once converts to its reality, now laugh the legend to
scorn.
 ILLUSIONS OF ART.

“If in Naples
I should report this now, would they believe me?”
Tempest.

Ev. The science of chemistry has unfolded most of the secrets of

material miracles, as Psychology those of the intellect and senses.
Not that I would attempt thus to explain your wonders of
Palingenesy, Astrophel; I will rather favour you with another batch,
for I was once fond of unkennelling these sly foxes.
It is solemnly attested by the noble secretary of a Duke of Guise,
that, in company with many scientific men, he saw the face of a
person in his blood, which had been given by a bishop, for
experiment, to La Pierre, the chemist, of Le Temple, near Paris.
There is an old book of one Dr. Garmann, “De Miraculis
Mortuorum,” and thus he writes:—“When human salt, extracted and
depurated from the skull of a man, was placed in a water dish, there
appeared next morning in the mass, figures of men fixed to a cross;”
and “when human skulls, on which mosses had vegetated, were
pounded, the family of the apothecary who pounded them were
alarmed in the night by strange and terrific noises from the
chamber.”
The body of the Cid, Ruy Diaz, as we read in Heywood’s
“Hierarchie,” sat in state at the altar of the cathedral at Toledo for
ten years. A Jew one day attempted, in derision, to pull him by the
beard; but on the first touch, the Cid started up, and in high
resentment scared the Israelite away by the unsheathing of his
mighty sword. And Master Planche has brought you legends from
the church of Maria Taferl, in Lower Austria, and other noted spots
on the Danube.
When Bernini’s bust of Charles I. was being conveyed in a barge
on the Thames, from a strange bird there descended a drop of blood
on the bust, which could never be effaced.
This is nothing but a fact in nature mystified, and (like the
growth of the Christmas flowering thorn of Glastonbury, from the
walking-staff of Joseph of Arimathæa) is too glaring to be
misconstrued.
Other of these blood miracles are still more easy of solution. The
blood spots from David Rizzio are shown to this day in Holyrood: and
it was believed, that after the Irish massacre the blood of the victims
then slain on Portnedown Bridge, has indelibly stained its
battlements. But these spots are nothing but the brown vegetative
stains which geology has discovered on many fossils.
Now listen to Father Gregory of Tours. “A thief was committing
sacrilege at the tomb of Saint Helius, when the saint caught him by
the skirt, and held him fast.” Probably his garment hitched on a nail.
Another old man, while removing a stone from the grave of a saint,
was in a moment struck blind, dumb, and deaf. Probably the
mephitic gases exhaling from the tomb were the source of all this
mystery.
Then, as to the impositions of the priesthood. In Naples was the
blood of Saint Januarius concealed in a phial, and on certain solemn
days this so called blood really became liquified; but it was effected
secretly, by chemical means; and I remember, the archbishop who
confessed the secret to the French general Championet, was exiled
by the Vatican.
In the reign of Henry VIII. too (I quote from Hume), other
bloody secrets of this sort were unfolded. “At Hales, in the county of
Gloucester, there had been shown during several ages the blood of
Christ brought from Jerusalem; and it is easy to imagine the
veneration with which such a relic was regarded. A miraculous
circumstance also attended this relic. The sacred blood was not
visible to any one in mortal sin, even when set before him; and, till
he had performed good works sufficient for his absolution, it would
not deign to discover itself to him. At the dissolution of the
monastery the whole contrivance was detected. Two of the monks,
who were let into the secret, had taken the blood of a duck, which
they renewed every week; they put it in a vial, one side of which
consisted of thin and transparent crystal, the other of thick and
opaque. When any rich pilgrim arrived, they were sure to show him
the dark side of the vial till masses and offerings had expiated his
offences, and then, finding his money, or patience, or faith, nearly
exhausted, they made him happy by turning the vial.”
But there is no end to relics in Italy. Even two hundred years ago,
John Evelyn makes out this catalogue of those he saw in St. Mark’s,
at Venice.
“Divers heads of saints, inchased in gold; a small ampulla, or
glass, with our Saviour’s blood; a great morsel of the real cross; one
of the nails; a thorn; a fragment of the column to which our Lord
was bound when scourged; a piece of St. Luke’s arm; a rib of St.
Stephen; and a finger of Mary Magdalene!”
Among the more innocent illusions of art, I may remind you of
concave and cylindrical mirrors and lenses, the magic lanthorn, “les
ombres chinoises,”and the phantasmagoria of Cagliastro, by which
daggers appear to strike the breast of the spectator, and images of
objects in other rooms are thrown on the walls of that in which we
are sitting. A mirror, thus accidentally placed, has afforded the
evidence of murder within our own time.
The duration of impressions on the eye, is another source of
illusion. An image remains on the retina, I believe, about the eighth
of a second; as it departs, if another object supplies its place in
quick succession, the two images form, as it were, a union, and
become blended. A knowledge of this law, in the ages of blind
superstition, would have placed an overwhelming weapon in the
hands of priestcraft; in our day, it is the source of rational and
innocent pleasure, by the invention of optical toys.
The whisking of an ignited stick produces a fiery circle—why?
Because from excessive rapidity the rays from one point remain
impressed on the retina, until the revolution completes the circle.
The Thaumatrope, or wonder-turner, and the Phantasmascope,
are ingenious illustrations of this law of impression; so also is the
whirling machine, which so beautifully evinces the fact of white
being compounded of all the prismatic colours, blended in certain
proportions. The prismatic Iris is painted on a revolving circle; by
excessive rapidity of revolution, the colours are actually blended (as
if mixed in a vessel) on the retina, and the surface of the machine is
white to the eye.
To these may be added the combustion of phosphorus and other
substances, in oxygen: red, green, and blue lights, which change the
angel face of beauty into the visage of a demon; and the inhalation
of noxious fumes and gases, creating altogether a new train of
phantoms in the world of experimental magic, and developing the
formerly occult mysteries of the art of incantation.
Chance may also involve a seeming mystery of very awful import.
Some years ago the town of Reading was thus bewildered. On the
loaves were seen the most mysterious signs. On one, a skeleton’s
head and cross-bones; on another, the word “resurgam;” on another,
a date of death was marked in deep impressions. The loaves of
course were, by some mysterious influence, the vehicles of solemn
warning from the Deity.
The baker was churchwarden of St. Giles’s; his oven needed
flooring, and, winking at the sacrilege, he stole the flat inscribed
tombstones from the church-yard, and therewith floored his oven.
From the inscriptions of these stones the loaves took their mystic
impressions.
In the reign of Edward the Martyr, during one of the synods
assembled by Dunstan, the floor of the chamber suddenly gave way,
involving the death of many of its members. It chanced that Dunstan
had on that day warned the king not to attend the synod, and the
only beam which did not give way was that on which his own chair
was placed. This might be coincidence merely, although I believe it
was discovered that it was a concerted trick; but the preservation of
the king and the priest were, of course, attributed to special
interference of the Deity.
But there is one phenomenon in animal chemistry so rare, and
indeed so wonderful, that there are few even among philosophers
who can give it credence. This is “spontaneous combustion,” the
result of an evolution of phosphorated hydrogen from the blood; the
remote cause of which may be traced in some cases to the free use
of alcohol. The records of these cases are very circumstantial,
especially the two most remarkable—that of the Contessa Cornelia
Bandi, of Cerena; and of Don Bertholi, an ecclesiastic of Mount
Valerius. But I check my wanderings into this maze of mystery, in
pity to your patience, fair ladies; for I perceive Astrophel is again out
of our sphere, and, enveloped in the cloud of his own mystic
meditations, will not know that this spontaneous combustion is
almost as wondrous a tale as his “Lady of the Ashes.”
 ILLUSTRATION OF MYSTERIOUS SOUNDS.

“The isle is full of noises,

Sounds and sweet airs, that give delight and hurt not.
Sometimes a thousand twangling instruments
Will hum about mine ears: and sometimes voices.”
Tempest.

Ev. So, you see, the effect of novelty is never more powerfully
displayed than by unusual impressions on the finer senses; that
appearances which the eye perceives, and which the mind cannot
explain, become phantoms, involving some special motive of wonder
or dismay.
So eccentric impressions on the mechanism of the internal ear
may be equally illusive. We have ghosts of the ear as well as of the
eye.
As ignorance has often warped the optical phenomena which
certain atmospheric changes may produce, so peculiar and unusual
sounds may be accounted for on equally erroneous principles,
especially if they chance to resemble sounds which are the effects of
daily or common causes.
As the Hebrew bards hung their harps by the waters of Babylon,
the Irish were wont, during their mourning for the death of a chief,
to loosen their harp-strings, and hang them on the trees; and while
the wind swept the strings, they ever believed that the harp itself
sympathized in their sorrow.
Thus, when the lament, or “ullaloo,” of these wild Milesians
boomed along the mountain glens, mingled with the coione, or
funeral song, and the poetical cadence blended with the winds, how
easy to impart to it a more than human source; and thus the dismal
coronach among the Scottish Highlands may be mystified into the
“boding scream of the Banshee.”
 It is a classical question whether the rebel giant, Typhœus, was
crushed by Jupiter beneath the island of Inarime, or Mount Ætna;
but it might readily be believed by the Sicilian, who had read this
mythological tale, that the volcanic convulsions arose from the vain
struggles for freedom of this monster, who sent forth flames from his
mouth and eyes.
Within a mountain of Stony Arabia, to the north of Tor, very
strange noises are often heard as of the striking of an harmonic
hammer, or the sound of a humming-top, which completely infuriate
the camels on the mountain when they hear it. The Arabs believe
these sounds to proceed from a subterranean convent of monks, the
priest of which, to assemble them to prayer, strikes with a hammer
on the nakous, a metallic rod suspended in the air. M. Teetzen, who
visited the spot, assures us that the cause of all this is the mere
rolling of volumes of sand from the summit and sides of the
mountain.
In the last century, I remember there was a legend current in the
west of England, of the “Bucca,” a demon whose howling was heard
amid the blast which swept along the shore. It was a sure
foreboding of shipwreck. The prophecy was often but too fully
verified, but the voice of the demon was merely the premonitory
gale from one certain quarter, which is always the avant-courier of a
tempest.
I remember, when I was a child, the prevalent belief in Horsham,
that, at a certain hour of the night, the ghost of Mrs. Hamel was
heard groaning in her vault, beneath the great eastern window, and
it required some self-possession to walk, at midnight, around this
haunted tomb; for few would believe that the noises were nothing
more than the wind sweeping along the vaulted aisles of the church.
Those very extraordinary impositions on the sense of hearing at
Woodstock, in the truth of which, Astrophel, your faith was so firm,
were resorted to to create terror, and effect a political purpose. In
“the genuine History of the good Devil of Woodstock,” written in
1649, we are told of the pealing of cannon, the barking of dogs, and
neighing of horses, and other mysterious sounds, which certainly
created the greatest wonder and anxiety, until “funny Joe Collins”
Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.

More than just a book-buying platform, we strive to be a bridge

connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.

Join us on a journey of knowledge exploration, passion nurturing, and

personal growth every day!

ebookmasss.com