You are on page 1of 20

Pergamon

00051098(95)00101-8

Control Systems Engineering


N. A. KHEIR,P

Auromarrca, Vol. 32. No. 2, pp. 147-166, IYYh


Copyright 0 1996 Elsevier Science Ltd
Prmted in Great Britain. All rights reserved
ouo5-109X/96 $1.5.00 + 0.00

Education*

K. J. ASTROM,* D. AuSLANDER,~
K. c. CHECK,?
M. MASTEN
and M. RABINStt

0. F. FRANKLIN,~~

This article reflects on the origins of control engineering education (CEE),


assesses its maturing development and looks into what the future might
bring. Curricular, laboratory and software issues are discussed, academic
programs are surveyed, and recent textbooks are presented. This work is
not only an overview but also a clear indication of how CEE continues to
contribute to the preparation of modern engineers for the betterment of
humanity.
Key Words-Education:
programs.

analysis and design; software: experimentation

Abstract-This
comprehensive
article deals with the
important field of control systems engineering education.
Efforts have been made to present some historical
perspectives, major concepts and thoughts on a practical
curriculum when this field is viewed as a discipline. Also
discussed are curricular issues including typical laboratory
systems with emphasis on the role of simulation, logic and
sequencing, and real-time simulation. An elaborate section is
devoted to CACE software, its role in teaching and learning,
potential shortfalls, and trends in software development and
use. Views from industry are sought in terms of desirable
skills in the practicing engineer and continuing education
needs. A survey of a few academic programs and a complete
list of textbooks in control over the past three decades
appear in the appendices.

and laboratories:

academic

CEE has become an important and integral part


of engineering curricula in electrical, mechanical,
chemical and aeronautical disciplines. On the
other hand, industry continues to introduce more
sophisticated control applications in its products.
Society at large also enjoys the benefits of
applying controls in non-technical areas (including, for example, medical, economical
and
management systems applications).
This status report on CEE is addressed to
educators and interested industrialists in developed
and, more importantly,
developing
countries around the world. In this regard,
special focus has been placed on curricular issues
(including te xt b oo k s, laboratory
experimentation, and physical systems versus computerbased analysis, design, evaluation and visualization) and an assessment of the qualities needed
by industry in control engineers.
Specifically, Section 2 discusses the concepts of
dynamic systems, stability, feedback, dynamic
compensation and robustness, and is followed by
some historical perspectives with a few early
examples from various disciplines. The dilemma
facing control engineering educators, namely the
balance between excessive theoretical
proofs
versus emphasis on physical intuition
and
understanding
of the physical systems, is
discussed. Industrys views on the skills needed
of control systems engineers are presented in
Section 3. The curriculum is the focus of Section
4 and Appendix A, where 12 academic programs
from Asia, Europe and the United States are
presented
and analyzed.
The diversity
in
coverage, depth and philosophy of the programs
surveyed is discussed. Section 5 highlights the
significance of laboratory experimentation
and

I. INTRODUCTION

this article we reflect on the origins of control


engineering education (CEE), assess its maturing
development
and look into what the future
might bring. Undoubtedly, the importance and
stature of the field of control are evidenced by
the number of annual national and international
meetings and conferences, publications (including textbooks) and, more importantly perhaps,
its impact on industrial applications that touch
the lives of everyone. In the academic arena,
In

*Received
16 June 1994: revised 16 February 1995:
received in final form I June 199.5. This paper was not
presented at any IFAC meeting. This paper was recommended for publication in revised form by Editor A. P. Sage.
Corresponding author Professor N. A. Kheir. Tel. +I 810
370 2177: Fax +I 810 370 4261.
t Department of Electrical and Systems Engineering,
Oakland University, Rochester, MI 48309, U.S.A.
$ Department of Automatic Control, Lund Institute of
Technology, Lund, Sweden.
?jMechanical Engineering
Department,
University of
California at Berkeley, CA 94720, U.S.A.
11
Department
of
Electrical
Engineering,
Stanford
University, CA 94305, U.S.A.
n Texas Instruments, Plano, TX 75075, U.S.A.
tt Department of Mechanical Engineering, Texas A&M
University, College Station, TX 77845, U.S.A.
147

N. A. Kheir et al.

148

how it fits into the curriculum. Answers to what


constitutes a good laboratory experiment are
suggested; several papers addressing specific
advances in laboratory experimentation
as well
as a few curricular issues, including the role of
simulation in teaching, are found in Furuta
(1994) and Kheir et al, (1992). Section 6 is
concerned
with CACE
software.
Teaching
control with CACE software is not, however,
without shortfalls, one of which is the potential
de-emphasis of practical hands-on laboratory
experience and the possible lack of appreciation
of the difference between the imperfect world of
reality
and the perfect
world of CACE
simulation; trends in CACE including animated
visualization are also addressed. Some concluding remarks are presented in Section 7.
Educators and industrialists have authored
more than 100 textbooks in the past 20 years,
and Appendix B presents this survey, categorized by topical specializations; the growth in
computer-based
control textbooks in the past
three to five years is especially noted.
It is our hope that the reader will see through
this article not only a valuable historic overview
but also a clear appreciation
of how CEE
continues to contribute to the preparation of
engineers
for the betterment
of
modern
mankind.
2. CONTROL

SYSTEMS
ENGINEERING
DISCIPLINE

AS A

2.1. Major concepts


In order to describe control education, it is
useful to identify the basic concepts understood
by all control engineers and that form the
common foundation of any control curriculum.
The first concept common to all of control is
that of a dynamic system. The processes and
plants that are controlled have responses that
evolve in time with memory of past response.
The most common mathematical tool used to
describe
dynamic
systems is the ordinary
differential equation (ODE). Furthermore,
by
considering small signals near an equilibrium for
to
a short time, it is typically possible
approximate the model equations as linear and
time-invariant. Control engineering begins with a
thorough understanding of the dynamic response
and structure of linear, time-invariant ordinary
equations. Extensions can be made from this
foundation to system models that are nonlinear,
distributedsampled-data,
time-varying,
parameter, and so on. Included in the concept of
dynamic systems is the idea of writing equations
of motion based on the physics and chemistry of
the situation
as well as that of system
identification based on experimental data. Also,

part of understanding
the dynamical system
requires understanding the performance limitations and expectations of the system. Typically,
engineers from the different classical disciplines
such as electrical, mechanical, aeronautical and
chemical engineering
approach
the field of
control differently, based on the science and
technology
of their discipline. The correct
formulation of the model of the plant to be
controlled and a thorough understanding of its
properties is so very important, in fact, that a
good case can be made that control engineering
education should begin with mastery of one of
these traditional disciplines.
The second basic concept of control is
stability. Without stability, failure is guaranteed;
with stability, the system can at least be used.
The so-called classical control design methods
are based on a stability test. The root locus of
Evans traces the closed-loop poles in the s plane,
where the right half signals disaster. Bodes
frequency response method and the Nyquist
stability criterion focus on stability margins in
gain and phase. The introduction of optimal
especially
linear-quadratic
Gaussian
control,
(LQG) control, was always haunted by the fact
that the method did not include a guarantee of a
margin of stability (except in the very uncommon case of full-state feedback in the continuous
case); the same limitation applies to the less
sophisticated method of arbitrary pole placement. Many ad hoc schemes and guidelines for
the designer have been developed to go with
these so-called modern control methods to
mitigate the deficiency. The theory and techniques of robust design have been developed as
alternatives to LQG, and are methods that
optimize performance while assuring a margin of
stability robustness. In the realm of nonlinear
control, including adaptive control, it is common
practice to base the design on a Lyapunov
function in order to be able to guarantee stability
of the final result. In every case, stability as a
concept, and as a system requirement,
is a
central part of control engineering education.
The third basic concept, and the one most
distinctive of control, is feedback. While many
open-loop devices such as programmable logic
controllers (PLC) are in use, their design and
use are not part of control engineering.t One of
-.t Having

said this, it is just as well to point out that as


control is applied to manufacturing
processes
the engineer
frequently
finds that a logic controller,
or its equivalent,
is
contained
in the feedback
loop and must be treated. Such
combinations
of logic and differential
systems
are called
hybrids, and in many cases their evolution can be modeled as
discrete event systems (DES). Study of DES is increasingly
common
and, while on the periphery
today, may in the
future assume a more central place in control engineering.

Control systems engineering


the most important and early examples of the
use of feedback was in the work of James Watt
on steam engine design. James Clerk Maxwell is
considered by many to have begun the field with
his analysis in 1868 of feedback based on the
flying ball governor using ordinary differential
equations and considering the location of the
roots of the characteristic equation to determine
stability.
The introduction of feedback brings costs as
well as benefits. Among the costs are the need
for both actuators
and sensors, especially
sensors. Actuators define the control authority
and set the limits of speed in dynamic response.
The sensors, via their inevitable noise, limit the
ultimate accuracy of the control. Within these
limits, feedback affords the benefits of improved
dynamic response and stability margins, improved disturbance
rejection,
and improved
robustness to parameter variability. The tradeoff between costs and benefits of feedback is at
the center of control design, and should be at the
center of control engineering education.
The fourth and final basic concept of control
engineering is that of dynamic compensation.
This is the idea that by adding additional
dynamics to the system in a feedback loop, we
nearly
realize
the
limits
of
can
more
performance set by the physical properties of the
components. In the beginning there was PID
conpensation,
and the three-term
controller
remains a widely used element of control,
especially
in the process industries.
Other
compensatory
approaches are the lead-and-lag
networks and the observer-based compensators,
which include pole-placement approaches as well
as the LQG designs. Of increasing interest are
designs capable of including trade-off among
stability, dynamic response,
and parameter
robustness. Interesting controller structures inadaptive
the
Q
parameterization,
clude
and
schemes, such as self-tuning regulators
neural-network-based
controllers.
How the control engineering discipline developed with time and how the basic concepts
came to be introduced can be appreciated by
tracing the history of the field to locate the
origins of the important
concepts described
above. Use of feedback, for example, is to be
found in very ancient practices.
2.2. Historical perspectives
The roots of automatic control practice can be
traced to antiquity and medieval times (Mayr,
1970) in such devices as liquid level controls for
water clocks and irrigation trenches, position and
speed controls for windmills, and temperature
controls for chicken egg hatching incubators.

education

149

The early practitioners of controls had to have


been trained as apprentices and by observation.
Most of these early control manifestations
appear as simple on-off (bang-bang) controllers
with empirical settings much dependent upon
experience. One early exception to this pattern
grew out of flyball governors to regulate the
turning speed of windmill threshing stones.
James Watts observations of windmill speed
undoubtedly
strongly
incontrol
practices
fluenced his contributions to the flyball governor
control of steam engines. Airys analysis in 1840
of the speed control of a rotating telescope to
observe a fixed star for extended periods is
another example of early controls mathematical
analysis (Fuller, 1976).
The following half-century
witnessed the
establishment
of the mathematical
basis for
much of what we now teach in automatic control
courses, including such advances as Rouths and
Hurwitz stability analysis (1877), Lyapunovs
state model and nonlinear
stability criteria
(1890), Sperrys early work on the gyroscope and
autopilot (1910), and Sikorskys work on ship
steering (1923). All of these early efforts
exhibited a probing investigation into previously
uncharted mathematical waters, and also exhibited the absence of a framework of study.
Without a discipline of established theory and
practice to guide them, all of these early
researchers were problem-driven
and had to
invent
methodology
as they went along.
Differential equations, Heaviside operators and
Laplace transforms were their tools.
This largely changed in the late 1920s and
1930s with Blacks development of the feedback
electronic amplifier, Bushs differential analyzer,
Nyquists stability criterion and Bodes frequency response methods. It is no surprise that
the earliest college courses on automatic control
appeared in electrical engineering curricula in
the 1940s. The electrical engineering problems
faced usually had very complex, albeit mostly
linear, models, and had arbitrary and wideranging dynamics. The early telecommunication
and control advances at Bell Laboratories were
combined with British research on radar to
heavily contribute to weaponry control needs in
World War II. As important advances were
declassified
following the war, many new
concepts found their way into electrical engineering classrooms, including Nichols charts
and Wiener-Hopf filter theory.
Simultaneously, process control in chemical
engineering was being developed to control fluid
flow and level, temperature,
pressure
and
composition. Most of the processes controlled
were complex and highly nonlinear. but usually

N. A. Kheir et al.
had relatively
docile dynamics. One major
outcome
of this type of work was the
Ziegler-Nichols PID three-term controller with
experimental tuning based upon step-response
and self-oscillation data. The pneumatic realization of this control approach is still in use today,
worldwide, with relatively minor modifications
and upgrades (including sampled data PID
controllers
with feedforward
control,
antiintegrator-windup
controllers
and fuzzy logic
implementations).
Naturally, this material now
appears in most chemical and mechanical
engineering control courses.
The application of controls in mechanical
engineering dealt mostly in the beginning with
mechanism controls such as servomechanisms,
governors and robots. The problems tended to
be well modeled with good equations of motion
that were often highly non-linear.
Typical
dynamic responses were lightly damped with
low-frequency oscillations. Some early methodooutcomes
were
the
Oldenburgerlogical
Kochenburger
describing function method of
equivalent
linearization,
and minimum-time,
bang-bang control. The application areas of
mechanical controls in recent years has tended to
be broader than just mechanism control. Some
typical control application areas now include
manufacturing process controls, vehicle dynamics and safety controls, biomedical devices and
genetic process research.
In aeronautical engineering, flight control of
airplanes and helicopters has been the driving
force behind control courses in that discipline.
The problems were generally
a hybrid of
well-modeled mechanics plus marginally understood fluid dynamics. The models were often
and the dynamics were
weakly nonlinear,
sometimes unstable. Major contributions to the
framework of controls as discipline were Evans
root locus (1948) and gain-scheduling. The latter
to fly controlnecessary
was absolutely
configured aircraft that were otherwise unstable.
Additional major contributions to the growth
of the discipline of control over the last 30-40
to be independent
of
years have tended
Examples
include
disciplines.
traditional
principle
(1956),
maximum
Pontryagins
programming
(1957),
dynamic
Bellmans
Kalmans optimal estimation (1960) and the
recent advances in robust control. All of these
contributions
share a heavy emphasis
on
mathematical methodology and the capability of
being applied to any traditional discipline based
problem.
The next section addresses the question of
what elements of control practice now appear in
typical control curriculum. Prioritization
and

hierarchical ranking of different subject matter


are considered, as are some prognostications on
where we might go from here.
2.3. Abstract thoughts on a practical curriculum
It has been argued earlier that control
engineering education must include the ideas of
dynamic system, stability, feedback and compensation. Within these basic concepts, there is a
wealth of details and subtopics from which the
educator must select with care to compose a
control curriculum. The above historical review
has traced the origins of control to the solution
of practical problems such as control of a steam
engine or airplane, and to the study of such
problems by theoreticians. The possibilities for
topics to teach are sufficiently great that if one
tries to provide a global perspective, there will
not likely be time to discuss the theory on which
the results are based. On the other hand, if one
tries to present proofs of all theoretical results,
one is in danger of giving the students many
mathematical details with little physical intuition
or appreciation for the purposes for which the
system is designed. The pedagogical problem
arises from the fact that control, more than
almost any other discipline, is based on two
distinct streams of thought. One stream is
physical and discipline-based, because one must
always be controlling some thing. The other
stream is mathematics-based,
because the basic
and
feedback
are
concepts
of stability
fundamentally abstract concepts best expressed
mathematically. This duality has raised, over the
complaints
about the gap
years, regular
between theory and practice. One argument is
that the theory doesnt solve the real problem;
the other argument is that these practitioners
do not understand available theory well enough
to realize that using it would make a big
improvement
in sytem performance.
In our
view, this is rather like complaining about the
gap between the left and the right horse in a
wagon team. In fact, the distinction-the
gap-is
inherent in the nature of control, and both
theory and experience are required to pull the
load. Control education must recognize this fact
from the start. Both theory and practice need to
be covered-the
question is how?
A rather standard compromise is to assume
students coming to the control course have
already had simple linear systems, including
Fourier and Laplace transforms. Also assumed is
understanding
of electric circuits and simple
rigid-body dynamics. The control curriculum
typically begins with one or two courses designed
to present an overview of control based on
linear, constant, ODE models, s-plane and

Control systems engineering


Nyquists stability ideas, single-input-singleoutput
feedback,
and PID, lead-lag
and
pole-placement
compensation.
These introductory courses can then be followed by courses in
linear systems theory, digital control, optimal
control, advanced
theory of feedback,
and
system identification,
as well as applicationspecific courses in such areas as autopilot design
and case studies in process control. (Appendix A
outlines samples of undergraduate and graduate
control curricula from around the world).
Because control design is always rooted in
some real-world
problem,
it is especially
important to include practical experience and
real-world open-ended
project assignments in
the curriculum. As part of control education,
laboratory experience is extremely important.
Experiment is particularly important in the case
of control, because of the interdisciplinary
nature of the field. Students in a first course in
control may be majoring in electrical, mechanical, aeronautical or chemical engineering. And it
is increasingly common to find students from
operations research and biology who find a need
to understand the principles of control.
Constructing practical experiments for such a
diverse audience is especially challenging. It is
common to devise distinct experiments for the
chemical engineers,
whose previous courses
typically include little work on linear systems
and almost none on rigid-body
dynamics.
Experiments
(servos)
that use mechanical
plants, electromechanical
actuators and sensors
and electrical compensation are common for all
but the chemical engineers. In such an example,
students from the several disciplines are usually
more familiar with one part of the system than
another, and the experience of working in a
small team with students from another discipline
can be very interesting and educational. Also
important and increasingly used is computerbased simulation and design. As computers and
control design software become more readily
available and more powerful, these tools will be
ever more widely used to give a sense of more
and more realistic experiments to each student.
The result will be better prepared
control
engineers (Sections 5 and 6 respectively address
control
laboratories
and software
in more
detail).
We have not discussed the level of control
courses, whether graduate or undergraduate, nor
has the possiblity of a distinct department or
program in control been mentioned. In fact,
there are almost as many ways of organizing the
presentation of control as there are topics to be
covered. In truth, there is probably no best way
but rather many equally effective ways to solve

education

151

this particular aspect of control education. Each


institution needs to evaluate its own mission and
the resources available and reach an independent conclusion. However, having said that, it
seems worthwhile to point out that, since control
has frequently evolved in classical disciplines, the
possibility arises that several parallel courses,
taught in different departments on essentially the
same material, will be introduced,
and the
richness
of interdisciplinary
teamwork
so
effective in this field (particularly at the graduate
level) will not be realized. We believe that
efforts to avoid this situation will be well
rewarded.
To conclude this discussion of control as a
discipline, we offer a statement of goals for
control education that might be kept in mind
regardless of the specific choice of material or
the structure of the course presentation. These
are
(I) to provide the basis for a lifetime of learning
to deal with ever-new control problems;
(2) to establish and maintain high standards of
excellence for the experience of learning the
foundations/concepts
of control.
In the next section we shall consider the
question of how graduating engineers, especially
those specializing in controls, fare in the real
world.
3. THE REAL WORLD

In the earliest societies, learning and education consisted of observing older members of the
group as they hunted, cooked, gathered food and
conducted other tasks necessary for survival. In
the middle ages education and training was
gained through apprentice
assignments with
master craftsmen. However, with the advent of
the Industrial Revolution,
workers were no
longer able to learn by observation
and
one-on-one
apprenticeship-there
were
no
masters with the new inventions.
Suddenly
everyone needed new training. New education
disciplines, as well as new modes of education,
were required because the industrial world had
changed!
Industrys needs and expectations have now
become even more important drivers in the
development
of technical
education.
This
progression
is not surprising,
since most
engineers earn their livelihood in industry. Since
industry
is critically
dependent
upon the
capabilities of its workforce, both industry and
its engineers have a keen interest in education.
Meanwhile, advances in technology are expected and often demanded. Numerous articles

152

N. A. Kheir et al.

have been written regarding the advancement of


technology and its impact on society. Such
advances also impact upon engineers,
who
themselves are developing the technology.
3.1. The world of the practicing engineer
As technology
advances, the culture of
industry also evolves, and such changes impact
upon industrys needs for educated workforce.
There are several aspects of this cultural shift.
Marketplace.

There was a time when companies focused almost entirely on domestic


markets within their own country. Today, the
marketplace is truly global.
Company focus. It is still necessary for a
company to make profits in order to exist.
However, in a real sense, the primary focus
has shifted away from profits to customers. It
is recognized that if a companys focus is not
directed toward its customers then profits will
eventually suffer. Todays customers expectations drive companies
to introduce
more
sophisticated control in their products.
Quality.

The key ingredient


for todays
products and services is quality and reliability.
Whereas quality was sometimes traded off
against affordability in prior times, today there
are no compromises to quality.
Time versus money. In previous generations,
companies who best controlled their cost were
usually the most successful. Cost control is still
important, but today product life cycles for
many products are measured in months, not
years. Short time-to-market and early market
capture are essential for success.
Production.

Mass production,
where large
production lots were generated by assembly
lines, is now giving way to smaller production
runs that require flexible factories, just-intime inventory
processes
and concurrent
engineering.
These fundamental cultural shifts within industry
have filtered down to become dramatic changes
for the engineers who design, develop and build
industrys products.
l

Teamwork.
In previous generations,
each
engineer generally performed his/her work as
an individual. However, the trend of the
future is toward teams with broad-based
work objectives. These concurrent engineering
teams are multifunctional, with representation
from all of the relative disciplines necessary
for an overall job.

Cultural diversity. Engineering workforces of


the future will have a greater ethnic and
gender mix. In addition to diversity within a
given work location, there will also be more
geographic dispersion of a companys total
workforce.
Worker expectations.

In previous generations,
one of the key expectations of workers was job
security, in which most engineers devoted
their entire career to only a few employers.
Today, many workers change companies every
four to five years. In addition, government
figures for the U.S.A. indicate that, by the
end of the decade, 43% of the workforce will
be in new work arrangements
such as
part-time, work at home, self-employed, or
multiple jobs-up from 29% in 1970.
Productivity. In recent times manufacturing
productivity has increased dramatically (through use of computers,
automation
and
manufacturing
process redesign). The next
surge in productivity will likely come in the
engineering function; some experts believe
the increase will be 100-200% in the next five
years. Product development life cycle times
(from concept to design to production)
is
expected to be reduced by a factor of five.

3.2. Control system skills sought by industry


The next generation of engineering can be
described by a single word; change. World events
have stimulated fundamental changes for many
industries. New ways of doing business are now
becoming familiar, and the impact of these
changes is dramatic.
Industry wants educators to provide technical
foundations that will enable engineers to remain
current and competitive in an ever changing,
advancing, global marketplace.
Numerous industry surveys indicate that todays college
graduates are of very high quality, but they lack
hands-on experience, interpersonal
skills, and
basic understanding and appreciation of business
perspectives beyond their technical specialties.
wants educators
to improve
the
Industry
practical as well as theoretical skills of engineers.
Although there are many work skills and
attitudes that benefit every engineer, there are
several unique talents of control system engineers.
Control systems is an interdisciplinary technology that uses practical aspects of many fields.
The applications of control theory are likewise
diverse; vehicle control, manufacturing, automation, environmental control, economic systems,
medicine, resource control allocation and management systems. Control system engineers are
well equipped for the new world where diverse

Control systems engineering


skills are needed-control
engineers are not
narrow specialists.
We have recently surveyed key individuals at
several companies who employ a fairly large
number of control system engineers. Our specific
focus was to generate a report card regarding
recent university graduates. Table 1 summarizes
our results. New control system engineers are
usually assigned to analysis and simulation tasks.
Indeed, analytical and computer skills are among
the greatest strengths of recent control system
graduates. Although general industry surveys
indicate that new graduates often lack practical,
hands-on experience, this is not necessarily the
case with control system engineers. Owing to
co-op or laboratory experience, many already
have these skills; even those without the
experience generally catch on quickly. Likeneed additional
wise, many new graduates
interpersonal
skills. The acquired knowledge
level within control systems per se, is much
but industry
still wishes more
improved,
attention was given to applications and knowledge of related fields. On-the-job training is
required for virtually all control system engineers, because university
curricula simply
cannot cover all the diverse application fields of
control systems.
In previous surveys there has usually been a
wide diversity over the usefulness of modern
versus classical control techniques. Our recent
survey indicates
that classical control
still
dominates in industry, since these techniques are
well tested, easy to use and generally more
understandable than modern approaches. However, a continual, though moderately paced, shift
toward modern techniques is expected. One of

Table
Attribute
Control

system

knowledge

systems

153

the unanimous opinions from our survey was


that university coverage of both classical and
Industrys
modern techniques
is acceptable.
predictions are that the most important fields in
the next century will be intelligent control,
identification/estimation,
learning
techniques,
new CAD
techniques,
and computers
in
real-time control.
3.3. Continuing education
The relationship
between
university
and
industry
does not end when the student
graduates.
It is generally
recognized
that
additional training and continuous on-the-job
learning will be expected
from tomorrows
engineers.
Although
most large companies
develop some of their own training materials,
universities still continue as a primary supplier of
continuing education. Focused, lock step graduate degree/certificate
programs offered on site
by neighboring universities seems to be gaining
popularity.
Practicing engineers today recognize the need
to continue their education in order to retain and
expand their technical capabilties. Likewise,
company leaders and managers recognize that
the skills of their workforce significantly impact
upon their future competitiveness. Even though
both company managers and the individual
engineers recognize the necessity for continuing
education, there is really no satisfactory solution
yet to this challenge. In 1988, the National
Academy of Engineering stated that individual
professional growth and productivity are the
personal responsibility
of the individual engineer; thus, unless a future response to this
continuing education challenge is developed, the

report

card from industry

Grade

Comment

A+

Much improved: maybe PhDs are too specialized without adequate


knowledge
of the whole
field. Need more knowledge
of related fields (dynamics, communications,
information
theory, computer
science).
On-the-job
training is still required in our specialized applications.
frankly, theres no
university course for many applications.
Good coverage of both classical and modern techniques;
70-90% of industry work is
classical (well tested easier to use). Modern methods used more in R&D tasks due to
flexibility and adaptability
for computer
implementation.
Shift will continue to modern,
but only at a moderate
pace. Both techniques
are essential.
Need more laboratory
experience.
professors
need more practical exposure: need to
address real problems.
Most students know how to use industrys analysis tools.
Some new hires are excellent: some inadequate.
however. most catch on very quickly.
Wide mixture; some understand
the need for teamwork..
. but many do not!
Most new graduates
recognize need for continuing
education
and a majority are active in
this pursuit. Half-life of control system education
is 5-12 years.
Analytical
skills, computer
skills, control theory. overall enthusiasm
and interest.
Very few understand
industrys expectations,
many graduates
are disappointed
regarding
lower impact of theory than expected.
Implementation,
signal processing,
simulation,
statistics, identification,
intelligent
control. . . we need smart engineers who do control: not control engineers.

Job preparation

A-

Curriculum

Laboratories

B-

Hands-on
experience
Interpersonal
skills
Career development

BC+
B

Greatest strengths
Biggest disappointments

Most important

topics

1. A control

education

154

N. A. Kheir et al.

immediate task, while students are still at the


universities, is for their educators to instill an
understanding that they must take responsibility
upon themselves to maintain (and advance) their
engineering
skills. During the IFAC 1994
Symposium on Advances in Control Education,
two sessions were devoted to new approaches to
continuing education in industry.
4. A SNAPSHOT

OF THE CURRICULUM

From a historic perspective, a snapshot of the


curriculum in control systems engineering is
offered. We have contacted personal colleagues
from around the world and 12 universities have
been surveyed for their undergraduate
and
graduate curricula in control (seven are in the
United States and five from non-U.S. institutions) . From the international
community,
Tokyo Institute of Technology, the Swiss Federal
Institute of Technology (ETH) in Zurich, the
Polytechnic Federal School in Lausanne, Switzerland, the Technical University of Delft at the
and the Tokyo
Metropolitan
Netherlands,
University are surveyed; it is recognized that this
sample may not be representative
of non-U.S.
universities. From the U.S.A. programs from
Purdue University,
University
of Illinois at
Urbana,
University
of California
at Santa
Polytechnic
Institute,
Rensselaer
Barbara,
Oakland
University
and the University
of
Maryland are included; the details of these
programs are presented in Appendix A.
The information
request included undergraduate as well as graduate curricula, a brief
statement on the graduate degree requirements
and titles of textbooks used in each course
whenever available. The information received
reflects not only the diversity of the curricula in
controls in the 1990s but also shows significant

differences in the areas of emphasis among the


surveyed programs. Several models of where
control is being administered within institutions
of higher education become apparent from the
material in Appendix A.
Table 2 presents a summary of the main
control courses, and the number of programs
surveyed that offer them, along with information
on textbooks reported in use. Table 3 is a brief
list of less frequent specialized/experimental
courses; this list should be valuable for future
expansion of existing control curricula. An
earlier documentation was presented by Dorato
(1990).
A few additional points can be made from
carefully examining the sample surveyed in
Appendix A.
The curricula vary significantly from one
institution to the other in terms of structure,
flexibility, and breadth.
Less coursework accompanied
by developmental learning through research projects are
more popular and widely used in institutions
outside the United States.
The degree of sophistication and usage of
computer-aided
control engineering (CACE)
software among the surveyed programs are
not clear from the current survey.
Courses in microprocessor-based
control systems, robust control, manufacturing systems,
modeling and simulation of discrete-event
systems, computational methods, fuzzy control
and neural networks are not common to the
majority of the programs surveyed.
Owing to limitations of space, no effort is
made to document the graduate degrees
requirements.

Table 2. Main control courses

Title
Introduction to Control
Lumped Systems Theory
Nonlinear Control
Optimal Control

Offerings
(out of 12)
11
7
8
7

Adaptive Control
Robot Control
Digital Control

5
7
7

Modeling and Simulation


Advanced Theory
Stochastic Processes
Large Scale Multivariable Systems
Manufacturing Systems
Fuzzy Logic/Neural Networks

6
4
4
3
3
3

Textbooks reported in use


Ogata (1990), Franklin et al. (1994)
DeCarIo (1989), Vidyasagar (1978), Ogata (1992)
Khalil (1992), Slotine and Li (1991)
Bagar and Bernhard (1992), Luenberger (1984). Sage and White (1977),
Lewis (1986b)
Goodwin and Sin (1984)
Koivo (1989), Fu et al. (1987), Spong (1991)
Franklin and Powell (1990), Astrom and Wittenmark (1990), Kuo (1980)
Phillips and Nagle (1990)
Kheir (1995), Close and Fredrick (1993)
Lewis (1986a, b)
Franklin er al. (1990), Bertsekas (1987)
Maciejowski (1989), Wonham (1979) Friedland (1986)
Wright and Bourne (1988), Scheer (1994)
Klir and Folger (1988)

155

Control systems engineering education


Table 3. Specialized/experimental
courses and frequency of
their offering in the surveyed programs (out of 12)
Intelligent Control
Mechatronics and Control
Application of Artificial Intelligence
Simulation and Optimization of Large Scale Systems
Robust Control
System Identification
Nonlinear System Identification
Microcomputer-based Control Systems
Discrete-event Systems
Parallel and Distributed Computation
Power Systems Control
Power Systems Dynamics and Stability
Numerical Optimization Methods
Numerical Systems Theory
Active Control of Mechanical Vibration

2
1
1

1
1
2
1
2
1
1
1
1
1
1
1

Appendix
B lists about
100 textbooks
authored during the last two decades; this is a
strong indicator of the wealth of talent and
commitment of many educators and industrialists
to the development
of the field of control
systems engineering. The textbooks are listed in
13 groups according to areas of interest and
specialization. It is noted that during the past
three to five years, a larger number
of
computer-based books has been written (several
authors include a student-version diskette with
the text).
Now that the specifics of control engineering
curricula have been discussed, we devote the
next two sections to a discussion of the
important laboratory
experience and modern
computer-aided
control engineering
(CACE)
software. It should be emphasized that the
degree of students exposure to rich (interdisciplinary) experiences
in the laboratory
varies
significantly from program to pogram and from
one institution to another. Typical laboratory
experiments with the characteristics of quality
lab experiments are also introduced.

and numerical) are useful in the sense that they


can accurately predict what will happen. The
laboratory should also stimulate physical intuition and give an opportunity
to exercise
engineering judgment.?
Because of the rich variety and complexity of
automatic control, there is no possibility to
develop laboratories that can give a comprehensive view of all application fields. It is however,
important for the laboratory to indicate some of
the typical hierarchies found in real control
systems. Typical tasks were simple feedback and
feedforward loops. The laboratory should also
illustrate that many different technologies are
used in control systems and that interfaces and
components are important. It is also relevant to
at least indicate issues concerning supervisory
loops and the problem of humans in control
systems, including man-machine
interfaces and
humans in-the-loop in real time.
5.1. Typical laboratory systems

A control laboratory should illustrate many


aspects of control engineering, i.e. modeling,
identification, simulation, analysis, design and
implementation.
Twenty
years ago it was
necessary to use specialized equipment to cover
these tasks. The availability of microprocessors
and personal computers has made it much
simpler to develop cheap and reliable laboratory
processes. A personal computer can be used to
implement the control algorithm, for graphics,
man-machine
interfacing
and for computer
aided instruction. On the other hand, it is no
simple task to develop good processes and
experiments.
A good laboratory experiment should (Balthen et al., 1981)
demonstrate

important theoretical

ideas;

reflect important real-life problems;


5. LABORATORIES
In this section the control laboratory,
its
objectives and organization along with how it fits
into the curriculum are discussed. As stated
earlier, one difficulty in teaching automatic
control is to provide a balance between theory
and practice.
A control
laboratory
is an
indispensable tool for this purpose. There are
many requirements for a control laboratory. Its
main purpose is to provide the connection
between abstract control theory and the real
world. Therefore it should give an indication of
how control theory can be applied and also an
indication of some of its limitations. In doing
this, the laboratory experience should reinforce
the concept that analytical approaches (theory

give visual and acoustic sensation;


have a suitable timescale;
be nonhazardous;
be inexpensive;
be easy to understand
A tall order indeed!
A number
of

and use.

different

processes

have

*Many papers have been written on philosophy and


construction
of control laboratories.
See for example.
Balchen et al. (1981), Mansour and Schaufelberger, (1981).
Edston et al. (1981), McDermott et al. (1988) Bonner et al.
(1988), Kulakowski (1988) Milek (1990) and Astrom and
Lundh (1992).

N. A. Kheir et al.
traditionally been used in control laboratories all
over the world. Many of them are designs that
have been developed in university laboratories.
Some of them have also been commercialized.
Processes originating at Imperial College and
UMIST (in the U.K.) have been commercialized
by Feedback Ltd and TecQuipment
Ltd. A
number of experiments
have recently been
marketed by Quanser and Camo; some typical
processes are discussed.
51.1. Mechatronics. The term mechatronics
traditionally meant a combination of mechanical
and electromechanical
systems. Position and
velocity control have been used successfully in
control labs for a long time. The basic system
consists of a motor with devices for measuring
angular orientation and rotation speed. Some
systems have two motors on the same shaft so
that disturbance torques can be added. Several
systems have a variable backlash and friction.
There are also more elaborate systems with
several masses connected
by springs. Such
systems are good for illustration of many aspects
of linear and nonlinear control. Many control
schemes can be applied to them, including
simple
lead-lag
compensation
and
statefeedback observers.
To do this well, it is
important that the systems are designed in such a
way that the linear behavior dominates. The
servo systems can also be used to illustrate
backlash, and friction can make application of
theory very complicated.
The inverted pendulum has always been a
popular experiment. A very elegant implementation of the inverted pendulum has been
developed
by Furuta at Tokyo Institute of
Technology. This allows demonstration
of the
difficulties in controlling an inherently unstable
system. With a proper design, an inverted
pendulum on a cart can also be used to illustrate
a gantry crane. The key problem here is to
introduce damping in an almost-unstable system.
Inverted double and triple pendulums have also
been developed (ETH Zurich, Furuta). These
systems are interesting because they are good
devices to illustrate the inherent limitations in
feedback; also see Kanzaki et al. (1994).
The ball-and beam process is another
mechatronics
device that illustrates
several
interesting measurement and control techniques.
Apart from positioning the ball on the beam, it
can also be used to throw the ball into a basket
(Astrom and Lundh, 1992). The hovering ball is
another popular example; systems with multiple
axes of motion have also been developed for
illustration.
5.1.2. Process control. There are many control
devices that are inspired by process control.

Typical examples are tanks with level control


(Astrbm and &tberg, 1986) and temperature
control. Such devices are easy to make and
maintain. A very practical implementation
of
temperature
control can be made by using
Peltier effect elements. It is also possible to have
multivariable systems by controlling level and
temperature.
The fan-and-plate system developed at Chalmers in Sweden (Schmidtbauer, 1990) is a useful
process. A small motor drives a propeller that
blows air on a hinged plate. The system has
highly oscillatory dynamics, which can easily be
changed by moving the propeller.
Another popular process is a hairdryer with a
long pipe having thermistors that control the
energy to the dryer. With a long pipe, this is an
excellent way to illustrate the effect of dead
time. A commercial version is available from
Feedback Ltd.
5.1.3. Manufacturing systems. Manufacturing
systems contain not only control systems, but
also a large amount of logic and sequencing that
is typically implemented in a programmable logic
controller (PLC). It is possible to build up small
models of manufacturing
systems by using
advanced toy kits from Fisher Technique and
Lego. Such systems are very useful to illustrate
many issues in manufacturing systems. Several
laboratories have also built more elaborate and
expensive
test
facilities
for
much
more
manufacturing.
Other types of manufacturing systems can be
built up around standard
components
like
electric teapots, valves and pumps.
Small robots are other useful devices. They
are now available in many shapes over a wide
price range. Some of them have real control
systems in the joints while others have stepping
motors.
5.1.4. Autonomous vehicles. The subject of
small autonomous vehicles is another category of
laboratory systems. Small robots are available as
kits. Remote control can easily be arranged
using standard equipment for control of model
airplanes. Position sensing can be done in a
variety of ways. One interesting possibility is to
use a CCD camera. Inexpensive cards for such
cameras
are available both for PCs and
workstations. An interesting experiment is to
reverse a truck with two or more trailers.
5.2. Curriculum issues
There are many curriculum issues involved in
the laboratory. The standard way of using a
scheme is that all students are exposed to the
practice of automatic control, and even canned
laboratories
can have enormous pedagogical

Control systems engineering


benefits when they are integrated into ordinary
courses. An example of this is in Lund (Astrom,
1992), where integrated laboratories are used
for classes with several hundred students. In this
scheme a laboratory is set up and an experiment
is run with all students for about two weeks. The
advantage of such a scheme is that all students
are exposed to the practice of automatic control
and that the students perform each experiment
at more or less the same time. Another
advantage is that it is possible to refer to the
laboratory in lectures with all students doing the
same experiments. The best benefits are clearly
obtained from open-ended
laboratories where
students set up their own experiments. Such
laboratories
do, however, require substantial
time and substantial supervision.
52.1. The role of simulation. Laboratory
equipment is expensive to buy and costly to run
and maintain.
Why not substitute
it with
simulation, which is much cheaper? Simulation is
certainly useful, but it cannot replace real
physical equipment. It is important that students
be exposed to the vagaries and perversities of
the real-world early-they
need to deal with
measurement
noise, friction, saturating valves
and motors. It is also very useful for them to
have some experience in making a real system
work (see also Section 6).
5.2.2. Logic and sequencing. Logic and Sequencing are important aspects of real automation
systems that a control engineer should master.
They are often not part of the ordinary control
curriculum, but definitely should be. In manufacturing systems, sequencing of parts is often
essential. In chemical process control they are
often taught in connection with batch control. In
Europe it is becoming increasingly popular to
base courses on Grafset, which is a modified
version of Petri nets.
5.2.3. Real-time simulation. A simulator with
real-time facilities has many advantages. It is
then possible to start with as simulation of a
process and controller. Once satisfactory simulations have been obtained, the simulated process
can then be replaced with the real process simply
by changing some connections. One simulator
that admits this is Simnon. This also has an
additional feature in terms of a code generator
that generates real-time code in Modula II with
calls to a real-time kernel (Dahl, 1989). With this
system, it is possible to get better user interfaces
than with the real-time simulator. It can also be
used in systems where complicated processes like
boilers and distillation columns are simulated on
one computer and a separate computer is used to
control them. With this scheme, it is also
possible to change the timescale so that the

education

157

experiments do not take too long. We should,


however, be reluctant to put such a system into
the hands of a student who has not had some
previous laboratory experience.
6. COMPUTER-AIDED
(CACE)

CONTROL
SOFTWARE

ENGINEERING

One of the recent technological breakthroughs in computer-aided


engineering
is the
continuing
development
of
sophisticated
software tools that allow control engineers to
design and analyze control systems in a friendly,
interactive environment. The underlying philosophy of the software is to minimize programming effort on the part of the users and allow
them to concentrate on applications.
Computer-aided
engineering
tools are an
important aspect of teaching, since they make it
possible to significantly increase the personal
efficiency in problem solving. They also make it
possible to deal with realistic problems, which
contributes to making teaching more interesting
and more realistic. Different aspects of CACE
are found in Atherton (1982) Schaufelberger
(1987, 1988, 1990) and Maier and Schaufelberger
(1990). The efforts by Schaufelberger
are
particularly innovative and noteworthy.
A control laboratory should have software for
computer-aided
control engineering.
This includes numerical
linear algebra,
computer
algebra, and data analysis and report writing.
There should also be tools for modeling, system
identification and control system design simulation. It is also advantageous if there are facilities
for hardware-in-the-loop
simulation as discussed above. Most of the desired functions can
be put together with standard software available
today.
In this section the presentation will focus on
the CACE software environment, the role of
software in teaching and learning, potential
shortfalls of CACE software, and trends in
CACE software.
6.1. CACE software environment
An essential feature for successful CACE
software is a smart technical computing environment architecture where a user can interact with
the computer and build his or her applications
in a convenient way. For instance, a user should
be able to express technical concepts, design
algorithms etc. to the computer using simple
intuitive natural language, graphical user interface (GUI) and icon manipulation, instead of
programming in scripted codes. The computer
displays the results or response with color
graphics, animation, 3D visualization, text and

158

N. A. Kheir et al.

sound. At the kernel of CACE software is


normally a collection of callable mathematical,
scientific and engineering functions for delivering
high-performance
numerical
computation
required in control applications. It should also
have an open architecture that easily allows a
user to modify existing functions as well as add
his or her own.
Such a philosophy,
which evolves with
computer
technology,
has been adopted by
several software vendors, who often introduce
and advertise their software at conferences and
in magazines relating to control engineering. A
partial list of the available CACE software
MATLAB,
includes
MATRIXx,
ACSL,
Control-C and Program CC. Of the above
software, MATLAB appears to be the most
popular among academic institutions (see Atherton ef al., 1994). This can be attributed to its
pricing through a university support program
and its operability
across PC, Macintosh,
workstations, minicomputers
and supercomputers. Numerous control textbooks have been
written based on the use of MATLAB (Franklin
et al. 1994; Shahian and Hassul, 1992, 1993;
Saadat, 1993; Strum and Kirk, 1994; Ogata,
1994a, b). The software is a general-purpose
high-performance numerical computation engine
with over 500 callable kernel functions to
support numerical and graphics applications
across a wide range of disciplines. Its functionality is extended
by toolboxes
for signal
processing, system identification, optimization,
control, robust control, p analysis and synthesis,
neural network, symbolic math, image processing, statistics, etc. Its open architecture allows a
third party to create his or her own toolbox
where necessary
(see Schaufelberger,
1987,
1988). Aside from these toolboxes, MATLAB
supports two other important salient features,
namely, the Simulink and 3-D visualization
graphics (MathWorks, 1993). Since it is impractical to discuss all aspects of the software, we
shall briefly highlight only some of the powerful
functions in the control toolboxes and the use of
Simulink.
Simulink
is an advanced
modeling
and
simulation feature of MATLAB that allows a
user to build block diagrams of dynamic systems
through interactive GUI and icon manipulation.
For example, a user can simply drag and drop
icons for a transfer function summation, signal
generator
and oscilloscope
into a working
Simulink window. Using a few intuitive actions
with the mouse, one can point, click and connect
them to form a block diagram for a closed-loop
control system whose input is driven by the
signal generator and output is observed by the

scope. After setting up the necessary parameters


via GUI dialog windows, the model is complete
and a simulation can be run with a click of the
mouse button. A click on the oscilloscope icon
will display and update the results of the
simulated output as the simulation crunches on.
The parameters (such as coefficients) and input
functions
(such as sine or square wave,
frequency and magnitude) for simulation can be
varied on the fly while the simulation is running,
The modeling and simulation of a simple control
block diagram can be achieved within a matter
of minutes. Students often find Simulink an
awesome introduction to the world of dynamic
systems simulation.
There is a very short
learning curve to overcome, since the block
diagram icons and graphics are similar to
textbook standards. Once they have their model
checked out, students usually find themselves
exploring
new variations
and designs, and
rediscovering or confirming the principles of
feedback and control system (see Jorgl, 1994).
Simulink has over 200 built-in block types of
simple and complex subsystem icons, and allows
a user to add personal custom blocks using the
MATLAB language and tool boxes. Simulink is
indeed a powerful interactive workbench for
dynamic system simulation for undergraduate
and graduate courses as well as for research and
development projects.
Control toolboxes are for the benefits of
control engineers as well as engineering students.
The control toolbox is a basic collection of
MATLAB functions that perform block diagram
manipulation, system analysis and design based
on classical and modern control methodologies,
It covers Laplace-transform,
frequency-domain
and state-space methods in the continuous-time
and discrete-time
domains rather extensively.
Some of the control design and analysis tools
include root locus, Bode diagram, Nyquist plots,
Nichols chart, pole placements, linear quadratic
optimal control and estimators, etc. Students
usually combine the calls to the toolbox with a
call to the Simulink to evaluate their design,
from within a main program.
The beauty of employing
an easy-to-use
powerful CACE software with intuitive command language and GUI is the short learning
phase necessary before one is familiar with the
CACE environment
and capability. As an
example, a student only requires about half an
hour of introduction by the instructor to be
familiar with the MATLAB
file setup and
programming philosophy. A student with a basic
mathematics background in matrix and vector
algebra should be able to become familiar with
the majority of the MATLAB commands and its

Control systems engineering education


potential in a two-hour self-tutorial session with
the CACE software. After another one-hour
self-tutorial session with Simulink, a student with
dynamic systems background will be up to speed
with the simulation capability of the software.
6.2. Role of software in teaching and learning
Computer laboratory assignments for control
curricula can be carefully designed around
CACE software to highlight the essential aspects
of dynamic systems and control. MATLAB
assignments to students may consist of simulation analysis of system stability, effect of
parameter
variation,
evaluation
of feedback
control system, transient and steady-state analysis, frequency response analysis, etc. Laboratories should be designed to challenge students
to correlate observations from the simulation
with the theory presented in the class (see
Kheir, 1984, 1987, 1990, 1993; Gillet et al., 1994).
From the instructors perspective,
teaching
control engineering
classes has never been
more exciting now that one has the use of CACE
software as a teaching companion. With a few
lessons on the basic necessary commands and
using the on-line help for the CACE software,
students can progress on their own through the
tutorials that accompany the software and the
examples given in the class. In a short time, an
instructor can begin to assign challenging CACE
problems for the topics covered in the class. For
instance, to build students confidence,
an
instructor
may ask them to first verify a
pencil-and-paper
example or a simple CACE
solution presented in the classroom. Students
may then be challenged with problems that
highlight the theory they have learned in class.
They should be encouraged to experiment with
different
designs
and
explore
alternative
approaches to solve the problems. The numerous
design failures and frustration experienced by
students are actually subliminal valuable lessons
taught by the companion. Control applications
for illustrating and exposing students to different
aspects of control theory are only limited by the
imagination of the instructors, scope of the
course and time factor.
Personal computers are a learning aid or
catalyst to students, whether one realizes it or
not. The problems posed by laboratory assignments occupy their thought and challenge them
to prove their ability to solve control engineering
problems
and demonstrate
them using the
CACE software. The motivation and determination to succeed often find students working
persistently on the computer, and learning more
and more about the subject in their own time.
Students usually rediscover the theory taught
AUTO 32-2-B

159

and discover the strong link between theory


and application.
CACE software also allows a user to conduct
exploratory simulation of new ideas. Because of
its ease of use, the software can accelerate the
pace of discovery, evaluation and development
of new control methods. Faculty and graduate
students have found CACE software to be of
tremendous research and creative value.

6.3. Potential shortfalls of CACE software


The shortfalls for relying heavily on CACE
software are the potential ignorance of hidden
numerical computation that powers the software,
the deficiency of realizing the limitation of the
actual physical system being modeled
and
simulated, and the lack of practical hands-on
experience in dealing with actual physical control
systems.
6.3.1. Numerical
results uersus
theoretical
validation. Students using CACE software may
not need to know the details of a numerical
integration algorithm, as long as the software has
a smart routine with variable integration step
size. The truth, however, is that users are still
the final people who must use their theoretical
grounding and control background to confirm
the validity of the simulation, analysis and design
of CACE results.
6.3.2. Behavioral versus physical component
mode&. The ability to successfully simulate a
control system provides a designer with confidence in his or her design and useful insights
into the problem
and its potentials.
The
confidence and insights, however, are limited by
the validity of the system model, perhaps in
terms of parameter estimates and omission of
rather significant nonlinearities. A main reason
for this shortfall is that control engineering
curricula tend to simplify representations
of
systems using generic
behavioral
type of
representation
rather than physical component
type representation.
The following is a case in
point. A control scientist often views a DC motor
with voltage input and position output as a
third-order system, ignoring many details of the
internal circuitry and mechanism. On the other
hand, a product deuelop~ent
and evalaatioff
control engineer, who has to examine the actual
circuitry and mechanism, is confronted
with
questions such as Coulomb friction, voltage and
current saturation, nonlinearities of amplifiers,
backlashes, and limit stops that have not been
carefully considered by the control scientist. The
limitation of the behavioral models used in
CACE simulation
must be emphasized
to
students.

N. A. Kheir et al.
6.3.3. Practical
hands-on
laboratory
experience. As described in Section 5, a laboratory
control experiment or project is one of the most
exciting, revealing and confidence-building
engineering experiences
to undergraduate
and
graduate students alike. Students derive satisfaction in seeing working control systems that they
have just successfully implemented. A control
system project requires them to pull together
much of their previously acquired engineering
knowledge, including system theory, computer
programming,
electronics,
mechanics, sensors
and instrumentation,
to successfully complete
the control application. More importantly, the
hands-on laboratory experience is the reckoning
where control engineering students learn to
appreciate the difference between the perfect
world of CACE simulation and the imperfect
world of reality.
6.4. Trends in CACE software
Over just a span of several years, computer
vendors have made tremendous breakthrough in
the areas of simulation, graphics and visualization technologies (IEEE, 1993; Kheir, 1995).
Virtual prototyping, which consists of interactive visualization for physical component functionality and virtual mockup model, is the current
thrust for many applications
that involve
multidiscipline engineering. The ultimate goal in a
virtual prototyping application is to build a computer model that behaves just as a real system
would, and lets a user manipulate to explore and
interact with the prototype model. CACE will
be a part of such virtual prototyping simulation.
6.4.1. Computer-aided
engineering
(CA E)
software for modeling and simulation of physical
components. As mentioned earlier, while empiri-

cal behavioral models are sufficient to represent


the dynamics of a system, they lack the ability to
predict the more realistic behavior of an actual
system. CAE software for capturing detailed
models of physical components has recently been
developed in the field of electrical, electronics,
mechanical, thermal pneumatic and hydraulics
systems, with minimal consultation with the field
of control. One such software that is widely used
in the industry is the Saber Simulation Software
(a product of Analogy), which uses the concept
of component templates for various engineering
parts. For example, its library consists of
comprehensive sets of templates for off-the-shelf
electronic parts (transistors, ICs etc.), generic
mechanical components
(motors, gears, mass,
spring, dampers etc.), electrical circuits (cables,
The
thermal
properties
etc.) and others.
the
software
allows a user to construct
component diagram for a control application

using the templates of the tentative physical


component.
The simulation can be used to
evaluate the suitability and statistical tolerance
of the components for the proposed design. The
use of such physical (object) oriented modelling
and simulation should be an excellent and
natural extension for many control engineering
curricula.
6.4.2. Animated
visualization
of dynamic
system motion.
Computer
simulation
of a
dynamic system traditionally
produces some
form of time-response
plots that show how the
system would respond to external excitation.
Trained analytical skills are often required to
interpret these graphical plots and extract useful
information about the systems characteristics
and performance. Therefore graphical plots are
not necessarily a natural means of conveying the
information: even though they are an excellent
means of representing data.
Computer visualization, on the other hand,
can enhance the perception and understanding
of a system being simulated many fold, while, at
the same time, it brings the simulation to life
and portrays a sense of reality. Visualization
often confirms the more abstract or hidden
information that would otherwise be missed or
misinterpreted.
Animation is often the most
natural form of visualization for simulating the
motion of a physical system. Our experience
shows that students conceptual understanding of
the simulation results and control concepts can
be significantly
enhanced
by having them
interact with real-time simulation and animation
software (Cheok et al., 1991, 1993; Cheok and
Huang, 1992; Cheok and Kheir, 1993).
The field of visualization covers a vast area of
applications, with varying breadth and depth
(IEEE,
1993). A representative
trend of
visualization
in the area where
graphics
animation is used to depict systems behavior is
found in Kheir (1995). High-performance
and
high-speed interactive video graphics technology
is available today, and it will only be a matter of
time before the technology is utilized more
extensively in CACE.
6.4.3. Real-time code generation and controller
hardware. Although computer simulation and
visualization provide a convenient means for
verifying control concepts under investigation,
actual implementation of laboratory experiments
offer the opportunity of validating the results.
Implementation
of a real-time controller normally requires skillful hardware design and
software programming. as well as careful project
scheduling and management. This is the reason
why certain pieces of CACE software, such as
Matlab/Simulinks
Real-Time
Workshop
and

Control systems engineering


MatrixJSystemBuilds
AutoCode-100, now support automatic generation of real-time code for
implementing dynamic function of the system or
subsystem block developed in the simulation
diagram. An automatically generated C-language
source code for a controller block, for example,
can be combined with interface (A/D, D/A,
DIO) device drivers, compiled and run off a
personal
computer
to control a laboratory
experiment.
Although some experience in C
programming
may be required,
tremendous
development time and effort can be saved by
eliminating the need to program and debug the
software, or the use of low-level assembly
language. Special digital signal processor-based
such as dSPACE
DSP, is also
hardware,
supported by these real-time code generators, to
enable real-time implementation
of high-order
and complex dynamical control systems. Such
real-time software and hardware CACE support,
which offer a quick mechanism to transfer
theoretical simulation into actual implementation, is an excellent tool for introducing students
to the world of computer-controlled
systems.
In summary, CACE software is the integration
of numbers, text. mathematical models, control
knowledge and graphics to communicate information about control engineering to a user in an
effective manner. It is a powerful computer tool
that will play an even more useful role in the
control engineering education. The future trend
of CACE should capitalize
on the rapid
advances of computer technology toward interactive virtual environment for analysis, design
and prototyping for CAE instruction, research,
development and engineering of control systems.
7. CONCLUSIONS

To summarize, an attempt has been made to


examine control systems engineering
as an
interdisciplinary field. In this context, focus has
been broadly placed on the control curriculum
(with sample programs from Asia, Europe and
the United States), the significance of the
laboratory
experience,
and the increasingly
important role of CACE software in teaching
and learning. How the educational experience
shapes the control engineer is discussed with
views from industry highlighting the desirable
skills and talents of a productive
control
engineer in the 1990s and beyond. Attention has
also been paid to not only the origins of control
systems engineering, but also to the future,
where it is expected that more emphasis on good
control laboratories
and use of sophisticated
CACE software tools will have perhaps the
utmost impact on the quality and level of

161

education

preparedness of graduates. The importance of


continuing education
and recently published
textbooks is well documented.
We note that
practical hands-on laboratory experience can be
complemented
but cannot be replaced by the
exact world of CACE simulation environment.
In conclusion, the future of control education
will undoubtedly experience
more real-world laboratory
novel applications;

experiments

increased reliance on computer-aided


engineering (CACE);

with

control

broader education of control engineers and


more emphasis on interdisciplinary teamwork
activities;
closer ties between
particularly
with
education;

industry and academia,


regard
to
continuing

continued novelty in how textbooks in control


engineering blend theory and applications for
a computer-based instructional environment;
more discussion among educators on
pedagogical aspects of teaching controls;

the

continued
sophistication
in the laboratory
(computer visualization, interactive and virtual) environment
as technology keeps advancing rapidly;
more programs expected to present newly
developed concepts of control in the form of
specialized courses.

REFERENCES
Akashi, H. (1984). Trends in automatic
control education
in
Japan. In Preprims 9th IFAC World Congress, Budapest,
Hungary,
Vol. IV, pp. 121-126.
Anderson,
B. D. 0. and J. B. Moore
(1990). Optimal
Control: Linear Quadratic Methods. Prentice-Hall,
Englewood Cliffs, NJ.
Aris, A. (1979). Mathemutical Modeling Techniques. Pitman,
London.
Astram, K. J. (1968). Reglerteori (Confrol Theory). Almqvist
& Wiksell. Unusalal (in Swedish).
Astram,
K. J. and M. Lundh,
M. (1992). Lund control
program combines theory with hands-on experience.
IEEE
Control Systems, 12, 22-30.
AstrBm, K. J. and B. Wittenmark
(1989). Aduprive ConrroL
Addison-Wesley,
Reading, MA.
AstrGm,
K. J. and B. Wittenmark
(1990).
Comparer
Controlled
Systems,
Theory
und
Design,
2nd
ed.
Prentice-Hall,
Englewood
Cliffs, NJ.
Atherton,
D. P. (1981). The role of CAD in education
and
research.
In Proc. 8th IFAC Trienniul World Congress,
Kyoto. Janan. DD. 339553400.
Atherton,
D. P.: 0. B. Sorenson
and A. Goucem t 1994).
Teaching
control
engineering
using implementation
of
MATLAB.
In Preprims of 94 IFAC/ACE,
Tokyo, pp.
291-294.

162

N. A. Kheir

Auslander,
D. and R. Horowitz (1988). Mechanical
engineering control system curriculum. In Proc. American
Control Co@., Atlanta, GA, pp. 115-1156.
Bagchi, A. (1993). Optimal Control Stochastic Systems.
Prentice-Hall, Englewood Cliffs, NJ.
Balchen, J. G., M. Handlykken and A. Tyss (1981). The need
for better laboratory experiments in control engineering
education. In Proc. 8th IFAC Triennial World Congress,
Kyoto, Japan.
Bagar, T. and P. Bernhard (1991). H, Optimal Control and
Related Minimax Design
Approach. Springer-Verlag,

Problems:

Dynamic

Care

Berlin.
Bertekas,
D.
P.
(1987).
Dynamic
Programming:
Deterministic and Stochastic Models. Prentice-Hall, Englewood Cliffs, NJ.
Biston, J., J. DuFour and G. Gilles (1981). An example of
computer control systems practical teaching. In Proc. 8th
IFAC
Triennial World Congress,
Kyoto, Japan, pp.
3419-3424.
Bitmead,
R. R. (1990). Adaptive
Optimal
Control.
Prentice-Hall, Englewood Cliffs. NJ.
Bohlin, S. T. (1991). Interactive System IdentiJication,
Prospects and Pitfalls. Springer-Verlag, Berlin.
Bonner. K., R. Fischl and P. Kalata (1988). A generic.
hands-on control system laboratory. In Proc. American
Control Conf, Atlanta, GA, pp. 1600-1605.
Boyd. S. P. (1991). Linear Controller Design: Limits of
Performance. Prentice-Hall. Enelewood Cliffs. NJ.
Brogan, W. L. (1991). Modern -Control Theory. 3rd edn.
Prentice-Hall, Englewood Cliffs, NJ.
Butler, H. (1992): Model Reference Adaptive Control.
Prentice-Hall. Enelewood Cliffs. NJ.
Callier, F. M. and C. A. Desoer (1982). Multivariable
Feedback Systems. Springer-Verlag, Berlin.
Callier, F. M. and C. A. Desoer (1991). Linear Systems
Theory, Springer-Verlag, Berlin.
Chalam, V. V. (1987). Adaptive Control Systems, Techniques
and Applications. Marcel Dekker, New York.
Chen, C. T. (1984). Linear Systems Theory and Design. Holt,
Rinehart & Winston, New York.
Chen, C. T. (1987). Control System Design: Conventional,
Algebraic, and Optimal Methods. Pond Woods Press,
Stony Brook, NY.
Cheok, K. C. and N. J. Huang (1992). Real-time simulation
and animation for dynamic control systems. Simulation, 59,
246-263.

Cheok, K. C. and N. A. Kheir (1993). Computer


visualization teachware for evaluating performance of
control systems. In Proc. American Control Conf, San
Francisco, CA, pp. 1244-1245.
Cheok, K. C., N. A. Kheir and N. J. Huang (1991). A
computer simulation and animation control education
environment.
In Proc. IFAC Symp. on Advances in
Control Education, Boston, MA, pp. 56-61.
Cheok. K. C., N. J. Huang, T. Horner and T. Settle (1993).
Concurrent real-time simulation and animation of an
active suspension using TMS320C30 and graphics hardware. Simulation, 61,405-416.
Chiang, R. Y. and M. G. Safonov (1988). Robust-Confrol
Tooihox. The MathWorks, Natick, MA.
Close, C. M. and D. F. Frederick (1993). Modeling and
Analysis of Dynamic Systems, 2nd ed. Houghton Mifflin,
New York.
Cochin. I. and H. J. Plass Jr (1990). Analysis and Design of
Dvnamic Svstems. 2nd ed. Harrier & Row, New York.
Dahl, 0. (1989). Generation of-structured Modula-2 Code
from a Simnon system. TFRT-7416, Department
of
Automatic Control, Lund Institute of Technology.
DAzzo, J. J. and C. H. Houpis (1988). Linear Control
System Analysis and Design. McGraw-Hill, New York.
DeCarlo, R. A. (1989). Linear Systems: A State Variable
Approach with Numerical Implemeniation. Prentice-Hall,
Englewood Cliffs, NJ.
Desrochers, A. and G. Saridis (1984). Educational trends in
automation in the United States. In Preprints 9th IFAC
World Congress,
Budapest, Hungary, Vol. IV. pp.
116-120.

et al.
B. W. (1991). Systems Analysis, Design
Computation. Prentice-Hall, Englewood Cliffs, NJ.
Dorato, P. (1987). Robust Control. IEEE, New York.

Dickenson,

and

Dorato, P. (1990). A survey of control systems education in


the United States. IEEE Trans. Educ., E-33, 306-310.
Dorato. P. and C. Abdallah (1993). A survey of engineering
education outside the United States: implications for the
ideal engineering program. ASEE J. Engng Educ., 84,
212-215.

Dorf, R. C. (1992). Modern Conrrol Systems, 6th ed.


Addison-Wesley, Reading, MA.
Dorny, C. W. (1993). Understanding Dynamic Systems:
Approaches to Modeling, Analysis and Design. PrenticeHall, Englewood Cliffs, NJ.
DSouza, A. F. (1988). Design of Control Systems.
Prentice-Hall, Englewood Cliffs, NJ.
Elmavist, H., K. J. Astrom, T. Schoenthal
and B.
Wittenmark
(1990). Simnon
Users
Guide.
SSPa,
Goteborg.
Eyman. E. (1988). Modeling, Simulation and Control. West
Publishing Company, St. Paul, MN.
Fiekema, H. (1987). Can the efficiency of (control)
engineering education be improved? An answer from the
field of holistic education and possibilities of a Japanese
method. In Preprints 10th World Congress on Automatic
Control, Munich, Germany, Vol. 5. pp. 285-290.
Francis, B. A. (1986). A Course in H, Control Theory.
Springer-Verlag, Berlin.
Franklin, G. F. and J. D. Powell (1988). Digital Control at
Stanford. In Proc. American Control Conf, Atlanta. GA,
pp. 1146-1150.
Franklin, G. F. and J. D. Powell (1990). Digital Control of
Dynamic Systems. Addison-Wesley. Reading, MA.
Franklin, G. F., J. D. Powell and A. Emani-Naeini (1994).
Feedback Control of Dynamic Systems, 3rd ed. AddisonWesley, Reading, MA.
Frederick. D. K. and A. B. Carlson (1971). Linear Systems in
Communication and Control. Wiley, New York.
Friedland, B. (1986). Control System Design: Introduction to
State-Space Methods. McGraw-Hill, New York.
Fu. K. S.. R. C. Gonzalez and C. S. G. Lee (1987). Robotics:
Control, Sensing, Vision and Intelligence. McGraw-Hill,
New York.
Fuller, A. T. (1976). The early development of control
theory. Trans. ASME: .I. Dyn. Syst. Meas. Control, 38,
109-l 18, 224-235.

Furuta, K. (1994). In Preprints 3rd IFAC Symp. on Advances


in Control Education, Tokyo, Japan, pp. 223-226.
Furuta, K., A. Sano and D. Atherton (1988). State Variable
Methods in Automatic Control. Wiley, New York.
Gajic. Z. and X. Shen (1993). Parallel Algorithms for
Optimal Control of Large Scale Linear Systems. SpringerVerlag, Berlin.
Gillet, D., R. Longchamp, D. Bonvin and G. F. Franklin
(1994). Introduction to automatic control via an integrated
instruction approach. In Preprints 1994 IFACIACE Symp.,
Tokyo, Japan, pp. 83-86.
Golten, J. and A. Verwer (1991). Control Systems Design
and Simulation. McGraw-Hill, New York.
Goodwin, G. C. and K. S. Sin (1984). Adaptive Filtering,
Prediction, and Control. Prentice-Hall, Englewood Cliffs,
NJ.
Groover, M. P., M. Weiss, R. N. Nagel and N. G. Odrey
(1986). Industrial Robotics: Technology, Programming and
Applications. McGraw-Hill, New York.
Gustafsson, K., M. Lundh and M. Lilja (1591). A set of
Matlab routines for control system analysis and design. In
Proc.
1991 IFAC Symp. on Advances in Control
Education, Boston, MA, pp. 86-91.
Hale. F. J. (1988). Introduction to Control System Analysis
and Design. Prentice-Hall, Englewood Cliffs, NJ.
Hostetter, G. H. (1988). Digital Control System Design. Holt,
Rinehart & Winston, New York.
Hostetter, G. H., C. J. Savant and R. T. Stefani (1982).
Design of Feedback Control Systems, Holt, Rinehart &
Winston, New York.
Houpis, C. H. and G. B. Lamont (1992). Digital Control

Control systems engineering


Systems.

Theory,
Hardware
and Software,
2nd ed.
McGraw-Hill, New York.
IEEE (1993). IEEE Spectrum (November).
Isermann, R. (1992). Adaptive Control Systems. PrenticeHall, Englewood Cliffs, NJ.
Isidori, A. (1989). Nonlinear Control Systems. SpringerVerlag, New York.
Jacquot, R. (1981). Modern Digital Control Systems. Marcel
Dekker, New York.
Jacquot, R. G. (1988). Introduction to Engineering Systems.
Allyn and Bacon, New York.
Jamshidi, M. (1992). Computer Aided Analysis and Design of
Linear Control. Prentice-Hall, Englewood Cliffs, NJ.
Jamshidi, M. (1983). Large-scale Systems: Modeling and
Control. North-Holland, Amsterdam.
Jorgl. M. P. (1994). A combined simulation and lab
experiment using Simulink. In Preprints of IFAC/ACE
Symp., Tokyo, Japan, pp. 259-262.
Kailath, T. (1980). Linear Systems. Prentice Hall, Englewood
Cliffs, NJ.
Kalouptsidis, N. (1993). Adaptive System Identification.
Prentice-Hall, Englewood Cliffs, NJ.
Kamen, E. W. (1987). Introduction to Signals and Systems,
Macmillan, New York.
Kanzaki, K., H. Kawasaki and 0. Kaneko (1994).
Experimental
systems for mechatronics
and control
education. In Preprints of IFACIACE
Symp., Tokyo,
Japan, pp. 17-30.
Kashyap, R. L. (1976). Dynamic Stochastic Models from
Empirical Data. Academic Press, New York.
Katz,. P. (1981). Digifal Control Using Microprocessors.
Prentice-Hall. Enelewood Cliffs. NJ.
Khalil, H. (1992). Ninlinear Systems. Macmillan, New York.
Kheir, N. A. (1984, 1987, 1990 and 1993). Moderator and
Organizer of Panel Discussions on Advances in Software
Development for Control Education and Research at
IFAC Congresses held in Budapest, Munich, Tallin and
Sydney.
Kheir, N. A. (1988). Systems Modeling and Computer
Simulation. Marcel Dekker, New York.
Kheir, N. A. (1995). Systems Modeling and Computer
Simulation, 2nd ed. Marcel Dekker, New York.
Kheir. N. A., G. F. Franklin and M. J. Rabins (Eds) (1992).
Advances

in

Control

Education:

Proc.

IFAC

Symp.,

.
Boston, MA, June 1991. Pergamon Press, Oxford.
Klein, R. E. (1988). Novel systems and dynamics teaching
technique using bicycles. In Proc. American Control Conf,
Atlanta, GA, pp. 1157-1160.
Klir. G. J. and T. A. Folaer (1988). Fuzzv Sets. Uncertaintv

and Information. Prentiie-Hall, Englewood Ciiffs, NJ.


Koivo, A. J. (1989). Fundamentals for Control of Robotic
Manipulators. Wiley, New York.
Kopacek, R., R. Genser, I. Troth and A. Weinmann (1984).
Automatic control education in Austria. In Preprints 9th
IFAC World Congress, Budapest, Hungary, Vol. IV, pp.
112-11s.
Kalakowski, B. T. (1988). Automatic control laboratory: new
approach to classical problems. In Proc. American Control
Cotzf, Atlanta, GA, pp. 1612-1616.
Kumar, P. R. and P. Varaiya (1987). Stochastic Systems:
Estimation. Identification and Adaptive Control. PrenticeHall. Englewood Cliffs, NJ.
Kuo. B. C. (1980). Digital Control Systems. Holt, Rinehart &
Winston, New York.
Kuo, B. C. (1991). Automatic Control Systems, 6th ed.
Prentice-Hall, Englewood Cliffs, NJ.
Landau, Y. D. (1979). Adaptive Control: The Model
Reference Approach. Marcel Dekker, New York.
Lawrence. P. and K. Mauch (1987). Real-time Micro
Computer Systems Design: An Introduction. McGraw-Hill,
New York.
Leigh, J. R. (1992). Applied Digital Control. Prentice-Hall,
Englewood Cliffs, NJ.
Leonard, N. W. and W. S. Levine (1992). Using MATLAE to
Analyze
and
Design
Conrrol
Systems.
Benjamin/
Cummings, Reading, MA.
Lewis, F. L. (1986a). Optimal Estimation: With an

education

163

Introduction to Stochastic Control Theory. Wiley, New


York.
Lewis, F. L. (1986b). Optimal Control. Wiley, New York.
Limebeer, D. J. N. and M. Green (1994). Robust Control
Systems: Design and Synthesis. Prentice-Hall, Englewood
Cliffs, NJ.
Little, J. and C. Model (1985). PC-Matlab Users Guide. The
MathWorks, Portola Valley, CA.
Luenberger, D. G. (1969). Optimization by Vector Space
Methods. Wiley, New York.
Luenberger, D. -(1978). Introduction to Dynamic Systems:
Theorv, Models, and ADDbCatiOnS.
Wilev. New York.
Luenberger, D. (1984). Linear and Nonlinear Programming.
Addison-Wesley, Reading, MA.
Lundh, M. (1991). A package for laboratory experiments in
discrete time control. In Proc. 1991 IFAC Symp. on
Advances in Control Education, Boston, MA, pp. 263-268.
Lunze, J. (1992). Feedback Control of Large Scale Systems.
Prentice-Hall, Englewood Cliffs, NJ.
Lunze, J. (1988). Robust Multivariable Feedback Control.
Prentice-Hall, Englewood Cliffs, NJ.
McDermott,
M., 0. Masory and B. Mooring (1988).
Laboratory for a first course in modeling and automatic
control. In Proc. American Control Conf.. Atlanta, GA,
pp. 1140-l 145.
Maciejowski, J. M. (1989). Multiuariable Feedback Design.
Addison-Wesley, Reading, MA.
Maier, G. E. and W. Schaufelberger (1990). Simulation and
implementation
of discrete-time
control systems on
IBM-compatible PCs bv FPU. In Preprints llrh IFAC
World Congress, Tallinn: Estonia, Vol. 12, pp. 178-183.
(1981). Digital
Mansour, M. and W. Schaufelberger
computer control experiments in the control group of
ETH-Zurich. In Proc. 8th IFAC Triennial World Congress,
Kyoto, Japan, pp. 3375-3381.
Mansour, M. and W. Schaufelberger (1989). Software and
laboratory
experiments
using computers
in control
education. IEEE Control Syst. Mag., 9(3), 19-24.
Mansour, M., W. Schaufelberger, F. E. Cellier, G. E. Maier
and M. Rimwall (1984). The use of computers in the
education of control engineers at ETH Zurich. Eur. J.
Engng Educ., 9, 135-1.51.

Masten. M. K. (1991). An industrial challenge to control


system educators. In Proc. 1991 IFAC Symp. on Advances
in Control Education, pp. 1-6.
MathWorks (1993). The Matlab Expo. The MathWorks, Inc.
Natick, MA.
Matko, D. (1992). Simulation and Modeling of Continuous
Sysrems. Prentice-Hall, Englewood Cliffs, NJ.
Mayhan, R. J. (1984). Discrete-time and Continuous-time
Linear Systems. Addison-Wesley, Reading, MA.
Mayr, 0. (1970). The Origins of Feedback Control. MIT
Press, Cambridge, MA.
Mendel, J. (1987). Lessons in Digital Estimation Theory.
Prentice-Hall, Englewood Cliffs, NJ.
Michel, A. N. and R. K. Miller (1977). Qualitative Analysis of
Large Scale Dynamical Systems. Academic Press, New York.
Middleton, R. H. and G. C. Goodwin (1990). Digital Control
and Estimation: A Unified Approach.
Prentice-Hall,
Englewood Cliffs, NJ.
Milek, J. J. (1990). Microcomputer-based
digital control
laboratory exercises. In Preprints 11th IFAC World
Congress, Tallinn, Estonia, Vol. 12, pp. 172-177.
Morari, M. and E. Zfiriov (1989). Robust Process Conrrol.
Prentice-Hall, Englewood Cliffs, NJ.
Narendra, K. S. and A. M. Annaswany (1989). Srable
Adaptive Systems, Prentice Hall, Englewood Cliffs, NJ.
Norton, J. P. (1986). An Introduction to Identification.
Academic Press, New York.
Ogata, K. (1987). Discrete-time Control Sysrems. PrenticeHall. Enelewood Cliffs. NJ.
Ogata, K.-(1990). Modern Control Engineering, 2nd ed.
Prentice-Hall, Englewood Cliffs, NJ.
Ogata, K. (1992). Systems Dynamics, 2nd ed. Prentice-Hall,
Englewood Cliffs, NJ.
Ogata, K. (1994a). Designing Linear Control Systems with
MA TLAB. Prentice-Hall, Englewood Cliffs, NJ.

164

N. A. Kheir et al.

Ogata,
K. (1994b). Solving Control Engineering Problems
with MATLAB. Prentice-Hall,
Englewood
Cliffs, NJ.
Owens, D. H. (1981). Multivariable and Optimal Systems.
Academic
Press, New York.
Ozguner,
U. (1988). A three course
control
laboratory
sequence.
In Proc. American Control Conf., Atlanta,
GA,
pp. 1617-1622.
Phillips, C. L. and R. D. Harbor
(1991). Basic Feedback
Control Systems. 2nd ed. Prentice-Hall,
Englewood
Cliffs,
NJ.
Phillips,
C. L. and H. T. Nagle (1990). Digital Control
Systems, Analysis and Design, 2nd ed. Prentice-Hall,
Englewood
Cliffs, NJ.
Pontryagin,
L. S., V. G. Boltyanskii,
R. V. Gamkrelidze
and
E. F. Mischenko
(1964).
The Mathematical Theory of
Optimal Processes. Pergamon
Press, Oxford.
Rowland,
J. R. (1986). Linear Control Systems: Modeling,
Analysis and Design. Wiley, New York
Rugh, W. J. (1993). Lineur System Theory. Prentice-Hall,
Englewood
Cliffs, NJ.
Saadat, H. (1993). Compututionul Aids in Control Systems
Using MATLAB. McGraw-Hill,
New York.
Sage, A. P. and C. C. White III (1977). Optimal Systems
Control, 2nd ed. Prentice-Hall,
Englewood
Cliffs, NJ.
Sastry, S. (1989). Adaptive Control Stability Convergence.
Prentice-Hall,
Englewood
Cliffs, NJ.
Schaufelberger,
W. (1983). Experiences
with project-oriented
teaching
courses
in a graduate
control
program.
Engng
Educ., 8, 79-87.
Schaufelberger,
W. (1987).
Teachware
for control.
In
Preprints of 10th World Congress on Automatic Control,
Munich, Vol. 5, pp. 267-272.
Schaufelberger,
W. (1988). Teachware
for control. In Proc.
American Control Conf, Atlanta GA, pp. 1135-1139.
Schaufelberger,
W.
(1990).
Educating
future
control
engineering.
In Preprints 1 lth LFAC World Congress,
Tallinn, Estonia, Vol. 1, pp. 82-93.
A.
W.
(1994).
CIM:
Computer
Integrated
Scheer,
Manufacturing:
Towards the Factory of the Future.
Springer-Verlag,
Berlin.
Schmidcbauer,
ST (1990). A new first course in analog and
dieital control.
In Preprints 1 lth LFAC World Congress.
Tallinn, Estonia, Vol. i2, pp. 161-165.
Seborg, D. E., T. F. Edgar and D. A. Mellichamp
(1989).
Process Dynamics and Control. Wiley, New York.
Shahian,
B. and M. Hassul (1992). Control System Design
wing MATRIX,.
Prentice-Hall,
Englewood
Cliffs, NJ.
Shahian,
B. and M. Hassul (1993). Control System Design
Using MATLAB. Prentice-Hall.
Enelewood
Cliffs. NJ.
Sheare;,
J. L. and B. T. Kulakowski
(1990). Dynamic
Modeling and Control of Engineering Systems. Macmillan,
New York.
Sinha, P. K. (1984). Multivariable Control: An Introduction.
Marcel Dekker, New York.
Sinha, N. K. (1986). Control Systems. Holt, Rinehart
&
Winston, New York.
Skelton,
R. E. (1988). Dynamic Systems Control: Linear
Systems Analysis and Synthesis. Wiley, New York.
Slotine, J.-J. and E. W. Li (1991). Applied Nonlinear Control.
Prentice-Hall.
Enalewood
Cliffs. NJ.
Sorenson,
H. W. 71980).
Parameter Estimation. Marcell
Dekker, New York.
Spong. M. W. (1991). Robot Dynamics and Control. Wiley,
New York.
Stengal, R. (1986). Stochastic Optimal Control: Theory and
Applications. Wiley, New York.
Strum. R. B. and D. E. Kirk (1994). Contemoorurv Linear
Systems Using MA TLAB. PWS-Kent,
Boston.
.
Swisher,
G. M. (1986). Introduction to Linear Systems.
Matrix Publishing Company.
Van de Vegte, J. (1994). Feedback Control Systems, 3rd ed.
Prentice-Hall,
Englewood
Cliffs, NJ.
Van den Bosch, P. P. J. and M. H. Verhaegen
(1989).
Modeling, Systems Identification and Simulation. Technical
University
of Delft, the Netherlands.
Vidyasagar,
M. (1978). Nonlinear Systems Analysis. PrenticeHall, Englewood
Cliffs, NJ.

Vidyasagar,
M. (1985).
Control
System Synthesis:
A
Factorization Approach. MIT Press, Cambridge,
MA.
Voland, C. (1986). Control Systems Modeling und Analysis.
Prentice-Hall,
Englewood
Cliffs, NJ.
Vukobratovic,
M. and
D. Stokic
(1982).
Control of
Munipalution Robots: Theory and Application. SpringerVerlag, Berlin.
Watanabe,
K. (1991). Adaptive Estimation and Control
Partitioning. Prentice-Hall.
Enelewood
Cliffs. NJ.
Williamson,
D. (1991). Digital Control and Implementation.
Prentice-Hall,
Englewood
Cliffs, NJ.
Wolovich,
W. A. (1987).
Robotics: Basic Anulysis und
Design. Holt, Rinehart & Winston, New York.
Wonham,
W. M. (1979). Linear Mtrltivuriuble Control: A
Geometric Approach. Springer-Verlag.
Berlin.
Wright,
P. K. and Bourne,
D. A. (1988). Munttfactaring
Intelligence. Addison-Wesley.
Reading, MA.

APPENDIX

A-SAMPLE
OF ACADEMIC
PROGRAMS

Programs in the United States


(a) Purdue University. The School of Electrical
ing at Purdue

offers the following

courses

Engineerfor EE majors:

Signals and Systems:


Systems Simulation
and Control Laboratory:
Feedback
System Analysis and Design (Ogata, 1990):
Electromechanical
Motion Devices and Control:
Electromechanical
Motion Control:
Modern Filter Design;
Digital Control Systems Analysis and Design (Kuo, 1980):
Microprocessor
Based Control Systems Laboratory:
Sensor-based
Robot Control Laboratory:
Introduction
to Robotic Systems (Koivo, 1980):
Optimization
Methods for Systems and Control (Luenberger, 1969):
Real-time Robot Control Laboratory
(Fu et ul., 1987):
State Estimation
and Parameter
Identification
of Stochastic Systems:
Lumped System Theory (DeCarlo.
1989):
Introduction
to Analysis of Nonlinear
Systems:
Introduction
to Modern Control Theory (Lewis, 1986a):
Control of Robot Manipulators
(Koivo, 1989);
Computational
Robotics;
Linear Multivariable
Control (Wonham.
1979):
Digital Process Control
and Mathematical Modeling of
Industrial Systems:
Topics in Advanced
Deterministic
Control:
Self Organizing
Control
(Adaptive
Control)
(Goodwin
and Sin, 1984):
Introduction
to Decision and Control under Uncertainty
(Kashyap,
1976):
Experimental
Courses:
Discrete-Event
Systems:
Parallel
and Distributed
Computation
and Control:
Fuzzy Logic Control
and Decision
Folger, 1988).

for

Optimization

Systems

(Klir

and

(b) University of Illinois at Urbana-Champaign.


The
department
of Electrical
and Computer
Engineering
in
conjunction
with the Decision and Control Laboratory
offer
one basic senior-level
control course: The Control Systems
course
is enhanced
with a set of laboratory
projects,
including
MATLAB
and digital simulation
using TUTSIM.
On the graduate
level, there are two groups: control courses
are listed first followed by closely related courses.
I. Control System Theory and Design:
Nonlinear
and Adaptive Control:
Optimum
Control Systems:
Sampled-Data
Control Systems:
Control of Stochastic Systems.

165

Control systems engineering education


II. Courses on Opfimization:

Introduction to Optimization;
Analysis of Nonlinear Systems;
Optimization by Vector Space Methods.
Courses on Power Systems Applications:

Modeling and Control of Electromechanical Systems


Power Systems Control
Power Systems Dynamics and Stability
It is interesting to note that control and related courses are
also offered in the Departments of Mechanical and Industrial
Engineering, General Engineering, Aeronautical and Astronautical Engineering, and Nuclear Engineering. Fields of
concentration at the graduate level are in Adaptive Control
and Identification; Dynamic Games and Decision Theory;
Digital Control Systems or Computer Control Systems;
Large-Scale Multivariable Systems; Nonlinear Systems and
Control: Stochastic Control; and Manufacturing Systems.
(c) University of California, Santa Barbara. The graduate
program in the Electrical and Computer Engineering
Department, in cooperation with the Center for Control
Engineering and Computation, offers the following options
and courses.
Linear Systems and Robust Control:

Linear Systems I and II:


Introductory Robust Control with Applications;
Robust Control Theory;
Analysis of Linear Models;
Automatic Feedback Control.
Nonlinear and Adaptive Control:

Analysis and Design of NL Control Systems;


Nonlinear Control Design;
Systems Identification:
Adaptive Control Systems.
Dynamical Systems and Robotics:

Advanced Dynamics I;
Advanced Dynamics II;
Dynamical Systems and Bifurcation Theory;
Robot Motion.
Filtering and Estimation:

Theory of Information;
Stochastic Processes in Engineering;
Optimal Estimation and Filtering;
Detection Theory;
Adaptive Filter Theory;
Neural Network Theory.
Optimization and Simulation:

Numerical Simulation;
Principles of Optimization;
Numerical Optimization Methods;
Dynamic Optimization.

(d) Rensselaers School of Engineering (Electrical,


puter and Systems Engineering Department).
Undergraduate

Com-

Courses:

Lumped Parameter Systems;


Linear Systems;
Control Systems Engineering;
Discrete Time Systems.
Graduate Courses:

System Analysis Techniques: Robotics and Advanced


Automation;
Nonlinear Control Systems;
Optimization Methods I;
Optimal Control Theory;
Multivariable Control Systems:
Adaptive Systems:
State Estimation and Stochastic Control:
Large Scale Systems.
(e) Oakland University, Department of Electrical and
Systems Engineering.
Undergraduate Courses:

Lumped Parameter Linear Systems;


Automatic Control Systems;
Modern Control System Design:
Robotic Systems;
Microcomputer-based Control Systems.
Flexible Manufacturing Systems
Graduate Offerings and Courses:

Three options in Systems Engineering allow the student to


focus on one of Dynamic Systems and Control, Robotics
and Intelligent Systems, or Manufacturing Systems;
Signal and System Analysis;
Dynamics and Control of Robot Manipulators;
Analysis of Nonlinear Control Systems;
Foundation of Computer-aided Design:
Computer Simulation in Engineering;
Digital Control Systems;
Optimal Control Theory;
Estimation and Control Theory:
Computer-Integrated
Manufacturing;
Large-scale Systems;
Linear Multi-variable Systems;
Stochastic Optimal Control and Estimation Theory;
Adaptive Control Systems:
Intelligent Control Systems.
(f) University
Engineering:

o,f Maryland,

Department

of

Electrical

Control Systems;
Control Systems Laboratory;
Systems, Control and Computation;
Control Seminar;
Modern Control System Design Methods;
Nonlinear Control Systems;
Systems Theory;
Optimal Control;
Advanced Topics in Control Theory.
(g) University of Maryland, Department

of

Chemical

Engineering:
Undergraduate required course:

Chemical Engineering System Analysis;


Graduate Courses:

Computational Methods:

Numerical System Theory;


Numerical System Applications;
Algebraic Eigenvalue Problems;
Topics in Scientific Computation.
Process Control:

Advanced Process Dynamics;


Advanced Process Control:

Advanced supporting courses in mathematics


are also
available: it is interesting to note dual listing of courses with
Computer Science, Mechanical Engineering, and Chemical
Engineering.

Chemical Process Dynamics (includes robust and model


predictive control);
Advanced Topics.

Programs in Europe
(h) The Swiss Federal Institute of Technology (ETH),
Institute of Automatic Control. Courses are offered in

Linear Control Systems (5th semester);


Digital Control Systems (6th semester);
Industrial Control Systems (6th semester):
Control of Nonlinear Systems (7th semester);
Optimal Control of Nonlinear Systems (8th semester);
Automation Techniques (7th semester);
Methods for Presentation
of Sequence Control (8th
semester):
Simulation Techniques (8th semester).

166

N. A. Kheir et al.

(i) The Polytechnic Federal School of Lamanne,


of Automoric Control. Courses are offered in

lnstitlrte

Dynamic Systems (4th semester);


Control System Analysis and Design (5th semester);
Digital Control (6th semester):
Process Dynamics and Control (5th semester):
Adaptive Control (7th semester):
State Space Methods (8th semester):
Modeling and Simulation of Dynamic Systems (7th
semester):
Modeling and Simulation of Discrete-event
Dynamic
System (8th semester).
(j) Technical Uniuersiry at Delft. the Netherlands. The
Control Engineering Laboratory in the Department
of
Electrical Engineering offers the following eight courses:
Systems and Control Engineering:
Sampled Data and Digital Control Systems;
Systems Theory:
Systems Modeling and Simulation:
Process Computer Systems:
Recent Trends in Control and System Theory I and II:
Robotic Systems.

Programs in Asia
(k) Tokyo Institute of Technology, Department of Control
Engineering. The curriculum focuses on the following areas:

Automatic Control Fundamentals; System Dynamics, Machinery for Measurement and Control: Industrial Measurement:
and Process Control. The courses offered include
Automatic Control:
Control Engineering Field Practice:
Control Engineering Laboratory:
Control Engineering Seminar:
Creation and Design in Control Engineering:
Elementary Automatic Control:
Feedback Control,
Instrument and Control System Analysis:
Intelligent Control:
Mechatronics and Control:
Modern Control Theory:
Physics for Control Engineers:
Robotics:
Signal Analysis in Control Engineering.
Graduate offerings include
System Theory:
Introduction to Detection and Estimation Theory:
Estimation Theory and Stochastic Control:
Identification for Control;
Robust and Adaptive Control.
(I) Tokyo
Metropolitan
University,
Japan.
Systems
Control Engineering within the Department of Electrical
Engineering emphasizes the following areas: fuzzy control,
neural networks, control, simulation and optimization of
large scale systems.
In the Department of Precision Engineering, analysis and
design of control systems are taught, with emphasis on
nonlinear
systems identification,
nonlinear
control by
geometric approaches, computer-aided modeling, application
active control of mechanical
of artificial intelligence,
vibration. and precision positioning.

In Computer-Aided Design and Manufacturing, product


modeling, virtual reality environment, intelligent information
processing, environment recognition, robotics, communication networks, computer-aided instruction are provided.
APPENDIX B-CONTROL
ENGINEERING
TEXTBOOKS: 1973-1993 SURVEY

Linear systems
Rugh (1993) Dorny (1993), Brogan (1991), Callier and
Desoer (1991). DeCarlo (1989), Skelton (1988), Furuta et al.
(1988), Chen (1984), Kailath (1980), Luenberger (1978).

Nonlinear systems
Khalil (1992), Isidori (1989) Vidyasagar (1978).

Digital control and computer-controlled

systems

Franklin and Powell (1990), Houpis and Lamont (1992),


Middleton and Goodwin (1990), Phillips and Nagle (1990),
Hostetter (1988) Ogata (1987) Jac uot (1981), Kuo (1980),
Leigh (1992), Williamson (1991), Istrom and Wittenmark
(1990). Katz (1981).

Adaptive control systems


Narendra and Annaswamy (1989) Astrom and Wittenmark
(1989), Chalam (1987). Goodwin and Sin (1984), Landau
(1979), Butler (1992). Isermann (1992), Sastry (1989).

Multivariable systems
Maciejowski (1989) Dorato (1987). Sinha (1984), Chen
(1984), Callier and Desoer (1982), Owens (1981), Kailath
(1980) Wonham (1979).

Large-scale systems
Gajic and Shen (1993), Jamshidi (1983), Michel and Miller
(1977) Lunze (1992).

IdentiJicationlestimation

theory

Bohlin (1991), Vanden Bosch and Verhaegen (1989). Mendel


(1987), Norton (1986) Stengal (1986), Sorenson (1980).
Kalouptsidis (1993), Watanabe (1991).

Stochastic systems
Kumar and Varaiya (1987), Lewis (1986a), Bagchi (1993).

Control system design/computer-aided

design

Strum and Kirk (1994), Ogata (1994a, b), Saadat (1993),


Shahian and Hassul (1992, 1993), Dickinson (1991), Golten
and Verwer (1991), Hale (1988), DSouza (1988) Vidyasagar
(1985) Boyd (1991), Jamshidi (1992).

Modeling and simulation


Close and Frederick (1993) Shearer and Kulakowski (1990).
Kheir (1988; 2nd ed., 1995), Eyman (1988), Voland (1986),
Matko (1992) Aris (1979).

Optimal control/optimization

techniques

Anderson and Moore (1990), Bagar and Bernhard


and White (1977), Bitmead (1990).

(1991),

Robust control systems


Limebeer and Green (1994) Morari and Zfiriov (1989),
Lunze (1988), Chiang and Safonov (1988).

Robotic systems

Koivo (1989), Wolovich (1987), Fu et al. (1987) Groover et


al. (1986), Vukovratovic and Stokic (1982).

You might also like