the behavioral approach - ku leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. a brief history...
TRANSCRIPT
![Page 1: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/1.jpg)
THE BEHAVIORAL APPROACHto
SYSTEMS THEORY
Jan C. WillemsK.U. Leuven, Belgium
SCTW 2005 Gebze Yuksek Teknoloji Enstitusu, September 7, 2005– p.1/76
![Page 2: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/2.jpg)
Open and Connected
The central tenets of our field:
Systems are open and consist of
interconnected subsystems.
Synthesis of systems consists of
interconnecting subsystems
– p.2/76
![Page 3: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/3.jpg)
Open
SYSTEM
ENVIRONMENT
Boundary
In this lecture, we think of this interaction boundary as ‘terminals’
SYSTEMENVIRONMENT
electrical components with ‘wires’
mechanical components with ‘pins’
fluidic components with ‘ducts’
signal processors with inputs and outputs
motors with terminals & pins
computer terminal, etc., etc., etc.
– p.3/76
![Page 4: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/4.jpg)
Open
In this lecture, we think of this interaction boundary as ‘terminals’
SYSTEMENVIRONMENT
electrical components with ‘wires’
mechanical components with ‘pins’
fluidic components with ‘ducts’
signal processors with inputs and outputs
motors with terminals & pins
computer terminal, etc., etc., etc.
– p.3/76
![Page 5: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/5.jpg)
Connected
An interconnection architecture with subsystems
– p.4/76
![Page 6: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/6.jpg)
Connected
Think of:
electrical circuits
mechanical constructions
fluidic systems
networks of signal processors
computers
essentially all engineering systems– p.4/76
![Page 7: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/7.jpg)
Connected
Observe the hierarchical nature
– p.4/76
![Page 8: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/8.jpg)
Interconnect
Reverse process: ‘tearing’ & ’zooming’ & ’linking’:
very useful in modeling.
– p.5/76
![Page 9: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/9.jpg)
Interconnect
Reverse process: ‘tearing’ & ’zooming’ & ’linking’:
very useful in modeling.
– p.5/76
![Page 10: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/10.jpg)
Interconnect
Reverse process: ‘tearing’ & ’zooming’ & ’linking’:
very useful in modeling.– p.5/76
![Page 11: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/11.jpg)
Mathematization
What are the appropriate concepts / mathematization?
What is an open dynamical system?
How do we deal with interconnections?
How does control fit in?
1. Get the physics right
2. The rest is mathematics
R.E. Kalman, Opening lectureIFAC World Congress, Prague, July 4, 2005
– p.6/76
![Page 12: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/12.jpg)
Mathematization
1. Get the physics right
2. The rest is mathematics
R.E. Kalman, Opening lectureIFAC World Congress, Prague, July 4, 2005
– p.6/76
![Page 13: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/13.jpg)
THEMES
First part:
1. Open and connected
2. A brief history of systems theory
3. Why are better framework is needed
4. Models and behaviors
Second part:
5 Linear time-invariant differential systems
6 Controllability and stabilizability
7 Representations of linear differential systems
8 PDE’s
– p.7/76
![Page 14: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/14.jpg)
How it all began ...
– p.8/76
![Page 15: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/15.jpg)
Planet ???
How does it move?
– p.9/76
![Page 16: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/16.jpg)
Kepler’s laws
Johannes Kepler (1571-1630)PLANET
SUN
Kepler’s laws:
Ellipse, sun in focus; = areas in = times; (period)
� ��� (diameter)
�
– p.10/76
![Page 17: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/17.jpg)
The equation of the planet
Consequence:
acceleration = function of position and velocity� ��� � � � � � � � � ��� � �
� via calculus and calculation
� ��� � ��� � ��� �� �� � � � � �
Hypotheses non
fingo
– p.11/76
![Page 18: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/18.jpg)
The equation of the planet
Consequence:
acceleration = function of position and velocity� ��� � � � � � � � � ��� � �
� via calculus and calculation
� ��� � ��� � ��� �� �� � � � � � Hypotheses non
fingo
– p.11/76
![Page 19: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/19.jpg)
Newton’s laws
2-nd law
� � � � � � �! � �
gravity
� � � � � � "$# %& '(*) + ! , ( 3-rd law
� � � .- � � � � � /0
1 213 2
4 � � 576 8 3 9 5 2
– p.12/76
![Page 20: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/20.jpg)
The paradigm of closed systems
– p.13/76
![Page 21: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/21.jpg)
‘Axiomatization’
K.1, K.2, & K.3
� ��� � ��� � �:� �� �� ;;� � � � � �
��� < � � < �‘dynamical systems’, flows
closed systems as paradigm of dynamics
– p.14/76
![Page 22: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/22.jpg)
‘Axiomatization’
Henri Poincare (1854-1912)
George Birkhoff (1884-1944)
Stephen Smale (1930- )
– p.15/76
![Page 23: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/23.jpg)
‘Axiomatization’
XA dynamical system is defined by
a state space anda state transition function=?> @ @ @ such that @ @ @
= � � A = state at time
�
starting from state A
How could they forget about Newton’s second law,about Maxwell’s eq’ns,
about thermodynamics,about tearing & zooming & linking, ...?
SYSTEM
ENVIRONMENT
Boundary
a property of but just as much of
– p.16/76
![Page 24: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/24.jpg)
‘Axiomatization’
Reply: assume ‘fixed boundary conditions’
SYSTEM
ENVIRONMENT
Boundary
� an absurd situation: to model a system,we have to model also the environment!
a property of but just as much of
– p.16/76
![Page 25: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/25.jpg)
‘Axiomatization’
SYSTEM
ENVIRONMENT
Boundary
Chaos theory, cellular automata, sync, etc.,‘function’ in this framework ...
Chaos: not a property of the physical laws,but just as much of what the system is
interconnected to.
– p.16/76
![Page 26: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/26.jpg)
‘Axiomatization’
SYSTEM
ENVIRONMENT
Boundary
Turbulence may not be a property ofNavier-Stokes, but just as much of
the boundary conditions.
– p.16/76
![Page 27: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/27.jpg)
Meanwhile, in engineering, ...
– p.17/76
![Page 28: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/28.jpg)
Input/output systems
SYSTEMstimulus response
causeinput
effectoutput
u1u2
u
1y
um
y2
p
input SYSTEM output
– p.18/76
![Page 29: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/29.jpg)
The originators
Lord Rayleigh (1842-1919)
Oliver Heaviside (1850-1925)
Norbert Wiener (1894-1964)
and the many electrical circuit theorists ...
– p.19/76
![Page 30: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/30.jpg)
Mathematical description
u1u2
u
1y
um
y2
p
input SYSTEM output
B CD EGF H IJ
or K L M CD N D O EP CD O EQ D O
B CD EF M J CD ESR IK L MUT CD N D O EP CD O E Q D OR
IK L
I VK L M � CD N D OW D O N D O O EP CD O EP CD O O E Q D OQ D O OR X X X
These models fail to deal with ‘initial conditions’.A physical system is SELDOM an i/o map
– p.20/76
![Page 31: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/31.jpg)
Input/state/output systems
� ��! Y � Z Y� [ � \ � ] Y� [
Rudolf Kalman (1930- )
– p.21/76
![Page 32: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/32.jpg)
‘Axiomatization’
State transition function:= � � A� [ > state reached at time
�
from A using input [.
X
Read-out function:] A� ^ > output value with state A and input value ^.– p.22/76
![Page 33: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/33.jpg)
The input/state/output view turned out to bea very effective and fruitful paradigm
for control (stabilization, robustness, ...)
to−be−controlled outputs
Actuators PLANT
FEEDBACKCONTROLLER
controlinputs
measuredoutputs
CLOSED−LOOPSYSTEM
exogenous inputs
Sensors
prediction of one signal from another, filtering
understanding system representations(transfer f’n, input/state/output, etc.)
model simplification, reduction
system ID: models from data
etc., etc., etc.
– p.23/76
![Page 34: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/34.jpg)
The input/state/output view turned out to bea very effective and fruitful paradigm
for control (stabilization, robustness, ...)
prediction of one signal from another, filtering
understanding system representations(transfer f’n, input/state/output, etc.)
model simplification, reduction
system ID: models from data
etc., etc., etc.
– p.23/76
![Page 35: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/35.jpg)
Let’s take a closer look at the i/o framework ...
in control
– p.24/76
![Page 36: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/36.jpg)
Difficulties with i/o
active control
to−be−controlled outputs
Actuators PLANT
FEEDBACKCONTROLLER
controlinputs
measuredoutputs
CLOSED−LOOPSYSTEM
exogenous inputs
Sensors
versus passive controlDampers, heat fins, pressure valves, ...
Controllers without sensors and actuators– p.25/76
![Page 37: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/37.jpg)
Difficulties with i/o
active control versus passive control
Controlling turbulence
for airplanes, sharks, dolphins, golf balls, bicycling helmets, etc.
Nagano 1998
These are beautiful controllers! The appropriate figure is
SYSTEM
to−be−controlled
CONTROLLED
CONTROLLERvariables PLANTcontrolvariables
With the ‘classical’ interconnection figure
output
PLANT +++
disturbanced1 d2 disturbance
+
CONTROLLER
ucontrolinput
yobserved
such controllers do not stabilize, because
dynamic order controlled system dynamic order plant dynamic order controller
– p.26/76
![Page 38: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/38.jpg)
Difficulties with i/o
active control versus passive control
Controlling turbulence
Nagano 1998
These are beautiful controllers! The appropriate figure is
SYSTEM
to−be−controlled
CONTROLLED
CONTROLLERvariables PLANTcontrolvariables
With the ‘classical’ interconnection figure
output
PLANT +++
disturbanced1 d2 disturbance
+
CONTROLLER
ucontrolinput
yobserved
such controllers do not stabilize, because
dynamic order controlled system dynamic order plant dynamic order controller
– p.26/76
![Page 39: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/39.jpg)
Difficulties with i/o
active control versus passive control
Controlling turbulence
Nagano 1998
These are beautiful controllers! The appropriate figure is
SYSTEM
to−be−controlled
CONTROLLED
CONTROLLERvariables PLANTcontrolvariables
With the ‘classical’ interconnection figure
output
PLANT +++
disturbanced1 d2 disturbance
+
CONTROLLER
ucontrolinput
yobserved
such controllers do not stabilize, because
dynamic order controlled system dynamic order plant dynamic order controller
– p.26/76
![Page 40: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/40.jpg)
Difficulties with i/o
active control versus passive control
Controlling turbulence
Nagano 1998
These are beautiful controllers! The appropriate figure is
SYSTEM
to−be−controlled
CONTROLLED
CONTROLLERvariables PLANTcontrolvariables
With the ‘classical’ interconnection figure
output
PLANT +++
disturbanced1 d2 disturbance
+
CONTROLLER
ucontrolinput
yobserved
such controllers do not stabilize, because
dynamic order controlled system dynamic order plant dynamic order controller
– p.26/76
![Page 41: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/41.jpg)
Difficulties with i/o
active control versus passive control
Controlling turbulence
Nagano 1998
These are beautiful controllers!
The appropriate figure is
SYSTEM
to−be−controlled
CONTROLLED
CONTROLLERvariables PLANTcontrolvariables
With the ‘classical’ interconnection figure
output
PLANT +++
disturbanced1 d2 disturbance
+
CONTROLLER
ucontrolinput
yobserved
such controllers do not stabilize, because
dynamic order controlled system dynamic order plant dynamic order controller
– p.26/76
![Page 42: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/42.jpg)
Difficulties with i/o
active control versus passive control
Controlling turbulence
Nagano 1998
These are beautiful controllers! But, the only people not callingthis ”control”, are the control engineers ...
The appropriatefigure is
SYSTEM
to−be−controlled
CONTROLLED
CONTROLLERvariables PLANTcontrolvariables
With the ‘classical’ interconnection figure
output
PLANT +++
disturbanced1 d2 disturbance
+
CONTROLLER
ucontrolinput
yobserved
such controllers do not stabilize, because
dynamic order controlled system dynamic order plant dynamic order controller
– p.26/76
![Page 43: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/43.jpg)
Difficulties with i/o
active control versus passive controlAnother example: the stabilizer of a ship
These are beautiful controllers! But, the only people not calling this”stabilization”, are the control engineers ...
Btw, this interconnection is, but shouldn’t be, called ‘singular’
The appropriate figure is
SYSTEM
to−be−controlled
CONTROLLED
CONTROLLERvariables PLANTcontrolvariables
With the ‘classical’ interconnection figure
output
PLANT +++
disturbanced1 d2 disturbance
+
CONTROLLER
ucontrolinput
yobserved
such controllers do not stabilize, because
dynamic order controlled system dynamic order plant dynamic order controller
– p.26/76
![Page 44: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/44.jpg)
Difficulties with i/o
active control versus passive control
The appropriate figure is_ _ _ _ ` ` ` `a a a a b b b bc c c c d d d de e e e
SYSTEM
to−be−controlled
CONTROLLED
CONTROLLERvariables PLANTcontrolvariables
With the ‘classical’ interconnection figure
output
PLANT +++
disturbanced1 d2 disturbance
+
CONTROLLER
ucontrolinput
yobserved
such controllers do not stabilize, because
dynamic order controlled system
f
dynamic order plant
R
dynamic order controller– p.26/76
![Page 45: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/45.jpg)
Let’s take a closer look at the i/o framework ...
for interconnection
– p.27/76
![Page 46: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/46.jpg)
i/o and interconnection
f11 f12
p12
h 1p
21f21
p22
f22
p 2h11
gg I hji k li m hjion pi ion pi qn ri i k si i m hion pi i qn ri k si m hion pi qgg I h k li m h n p ion p qn r i k s i m h n p i qn r k s m h n p q
inputs: the pressures t T T � t T �� t � T � t � �
outputs: the flows
ZT T � ZT �� Z � T � Z � �
f11
h 1p
22f22
h 211p
Interconnection:
This identifies 2 inputs AND (NOT WITH) 2 outputs,
the sort of thing SIMULINK c forbids.
This situation is the rule, not the exception (in fluidics, mechanics,...)
Interconnection is not input-to-output assignment!
– p.28/76
![Page 47: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/47.jpg)
i/o and interconnection
f11
h 1p
22f22
h 211p
gg I hji k li m hjion pi ion pi qn ri i k si i m hion pi i qn ri k si m hion pi qgg I h k li m h n p ion p qn r i k s i m h n p i qn r k s m h n p q
Interconnection: t T � � t � T � ZT �- Z � T � /
This identifies 2 inputs AND (NOT WITH) 2 outputs,
the sort of thing SIMULINK c
u
forbids.
This situation is the rule, not the exception (in fluidics, mechanics,...)
Interconnection is not input-to-output assignment!– p.28/76
![Page 48: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/48.jpg)
Sharing variables, not input-to-output assignment, is the basic
mechanism by which systems interact.
block 1Building block 2
Building block 1
Building block 2
Building
Before interconnection:
the variables on the interconnected terminals are independent.
After interconnection: they are set equal.
– p.29/76
![Page 49: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/49.jpg)
Let’s take a closer look at the i/o framework ...
for modeling
– p.30/76
![Page 50: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/50.jpg)
i/o in modeling
Physical systems often interact with their environment through
physical terminals
SYSTEMENVIRONMENT
On each of these terminals many variables ‘live’:
voltage & current
position & force
pressure & flow
price & demand
angle & momentum
etc. & etc.
The selection of what is an input and what is an output
does not need to be madeif it made, it should be made after the modeling is done
sometimes it cannot be made
– p.31/76
![Page 51: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/51.jpg)
i/o in modeling
Physical systems often interact with their environment through
physical terminals
SYSTEMENVIRONMENT
Situation is NOT:
on one terminal there is an input, on another there is an output.
u1u2
u
1y
um
y2
p
input SYSTEM output
This picture is misleading, if superficially interpreted.
The selection of what is an input and what is an output
does not need to be made
if it made, it should be made after the modeling is done
sometimes it cannot be made
– p.31/76
![Page 52: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/52.jpg)
i/o in modeling
Physical systems often interact with their environment through
physical terminals
The selection of what is an input and what is an output
most often does not need to be made
if it made, it should be made after the modeling is done
sometimes it cannot be made
The selection of what is an input and what is an output
does not need to be made
if it made, it should be made after the modeling is done
sometimes it cannot be made
– p.31/76
![Page 53: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/53.jpg)
i/o in modeling
Physical systems often interact with their environment through
physical terminals
The selection of what is an input and what is an output
does not need to be made
if it made, it should be made after the modeling is done
V"
I’ I"+
−V’
+
−
voltage controlled?
sometimes it cannot be made
– p.31/76
![Page 54: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/54.jpg)
i/o in modeling
Physical systems often interact with their environment through
physical terminals
The selection of what is an input and what is an output
does not need to be made
if it made, it should be made after the modeling is done
sometimes it cannot be madeVoltage
Currentinto diode
across diode
variables:
Y� v gg I Y � v
tangent bundle of the sphere is not ‘trivial’– p.31/76
![Page 55: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/55.jpg)
Conclusion
The inability of the i/o framework to properly deal with
(i) interconnections
and
(ii) passive control
is lethal.
Just as the state, the input/output partition needs to be constructed
from first principles models. Contrary to the state, such a partition
may not be useful, or even possible
We need a better, more flexible, universal, simpler framework that
properly deals with
open & connected.– p.32/76
![Page 56: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/56.jpg)
General formalism
– p.33/76
![Page 57: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/57.jpg)
Generalities
What is a model? As a mathematical concept.What is a dynamical system? What is the role of differentialequations in thinking about dynamical models?
So, before Boyle, Charles, and Avogadro got into the act,and may have seemed unrelated, yielding
The ideal gas law restricts the possibilities to
– p.34/76
![Page 58: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/58.jpg)
Generalities
IntuitionWe have a ‘phenomenon’ that produces ‘outcomes’ (‘events’).We wish to model the outcomes that can occur.
Before we model the phenomenon:the outcomes are in a set, which we call the universum.
After we model the phenomenon:the outcomes are declared (thought, believed)to belong to the behavior of the model,a subset of this universum.
This subset is what we consider the mathematical model.
So, before Boyle, Charles, and Avogadro got into the act,and may have seemed unrelated, yielding
The ideal gas law restricts the possibilities to
– p.34/76
![Page 59: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/59.jpg)
Generalities
This way we arrive at the
Definition
A math. model is a subset of a universum of outcomes
wis called the behavior of the model.
For example, the ideal gas law states that the temperature
x
,pressure
y
, volume
z, and quantity (number of moles) of an
ideal gas satisfy y zx � {
with
{a universal constant.
So, before Boyle, Charles, and Avogadro got into the act,and may have seemed unrelated, yielding
The ideal gas law restricts the possibilities to
– p.34/76
![Page 60: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/60.jpg)
Generalities
So, before Boyle, Charles, and Avogadro got into the act,x� y� z
and may have seemed unrelated, yielding
� | }~ wThe ideal gas law restricts the possibilities to
� � x� y� z� �� | }~ � y z � x � { �
– p.34/76
![Page 61: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/61.jpg)
Features
Generality, applicability
shows the role of model equations� notion of equivalent models� notion of more powerful model
Structure, symmetries
...
– p.35/76
![Page 62: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/62.jpg)
We will only consider deterministic models.
Stochastic models: there is a map
y
(the ’probability’)
y > � /� � �with a ‘ �-algebra’ of subsets of .
y � ‘the degree of certainty (belief, plausibility, propensity,relative frequency) that outcomes are in ;� � the degree of validity of as a model.
Fuzzy models: there is a map (the ‘membership function’)
– p.36/76
![Page 63: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/63.jpg)
We will only consider deterministic models.
Stochastic models: there is a map
y
(the ’probability’)
y > � /� � �with a ‘ �-algebra’ of subsets of .
Fuzzy models: there is a map � (the ‘membership function’)
�> � /� � �
� Y � ‘the extent to which Y� belongs to the model’.
– p.36/76
![Page 64: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/64.jpg)
We will only consider deterministic models.
Stochastic models: there is a map
y
(the ’probability’)
y > � /� � �with a ‘ �-algebra’ of subsets of .
Determinism: � ��� � � ��� � ���� �� �� �� y � �
.
Fuzzy models: there is a map � (the ‘membership function’)
�> � /� � �
Determinism: � is ‘crisp’:�?� ��� � � � /� � �� � � � " � � � > � � Y� � � Y � � �
– p.36/76
![Page 65: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/65.jpg)
Dynamical systems
In dynamics, the outcomes are functions of time �EVENTS
SYSTEMtime
time
time
Which event trajectories are possible?
Definition
A dynamical system =
with , the time-axis (= the relevant time instances),, the signal space
(= where the variables take on their values),
the behavior (= the admissible trajectories).
– p.37/76
![Page 66: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/66.jpg)
Dynamical systems
Definition
A dynamical system =
� > � � � � with
� |
, the time-axis (= the relevant time instances),, the signal space
(= where the variables take on their values),�
the behavior (= the admissible trajectories).
– p.37/76
![Page 67: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/67.jpg)
Dynamical systems
Definition
A dynamical system =
� > � � � � with
� |
, the time-axis (= the relevant time instances),, the signal space
(= where the variables take on their values),�
the behavior (= the admissible trajectories).signal space
time
Totality of ‘legal’ trajectories =: the behavior
– p.37/76
![Page 68: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/68.jpg)
End of Part I
– p.38/76
![Page 69: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/69.jpg)
Part II: Linear Differential Systems
– p.39/76
![Page 70: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/70.jpg)
Dynamical systems
Definition
A dynamical system =
� > � � � � with
� |
, the time-axis (= the relevant time instances),, the signal space
(= where the variables take on their values),�
the behavior (= the admissible trajectories).
– p.40/76
![Page 71: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/71.jpg)
Dynamical systems
Definition
A dynamical system =
� > � � � � with
� |
, the time-axis (= the relevant time instances),, the signal space
(= where the variables take on their values),�
the behavior (= the admissible trajectories).signal space
time
Totality of ‘legal’ trajectories =: the behavior
– p.40/76
![Page 72: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/72.jpg)
Dynamical systems
Definition
A dynamical system =
� > � � � � with
� |
, the time-axis (= the relevant time instances),, the signal space
(= where the variables take on their values),�
the behavior (= the admissible trajectories).
For a trajectory (‘an event’) � > � � we thus have:
�� : the model allows the trajectory ��� ��
: the model forbids the trajectory � w
– p.40/76
![Page 73: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/73.jpg)
Dynamical systems
Definition
A dynamical system =
� > � � � � with
� |
, the time-axis (= the relevant time instances),, the signal space
(= where the variables take on their values),�
the behavior (= the admissible trajectories).
Usually,� � |
, or
� /� � , etc. (in continuous-time systems),or
�� or
�
, etc. (in discrete-time systems).
– p.40/76
![Page 74: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/74.jpg)
Dynamical systems
Definition
A dynamical system =
� > � � � � with
� |
, the time-axis (= the relevant time instances),, the signal space
(= where the variables take on their values),�
the behavior (= the admissible trajectories).
Usually, |�
(in lumped systems),a function space
(in distributed systems, time a distinguished variable),a finite set (in DES)’ etc.
– p.40/76
![Page 75: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/75.jpg)
Dynamical systems
Definition
A dynamical system =
� > � � � � with
� |
, the time-axis (= the relevant time instances),, the signal space
(= where the variables take on their values),�
the behavior (= the admissible trajectories).
Emphasis:� � |�� |� �� solution set of system of (linear constant coefficient)ODE’s, or difference eqn’s, or PDE’s. � ‘differential systems’.
– p.40/76
![Page 76: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/76.jpg)
A series of examples
– p.41/76
![Page 77: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/77.jpg)
Examples
Let’s put Kepler and Newton in this setting.
K1+K2+K3 obviously define a dynamical system� � � � �
� � |� � | ��� all � > | | �
that satisfy Kepler’s 3 laws.
Nice example of a dynamical model ‘without equations’.
Is it a differential system?
This question turned out to be of revolutionary importance...
Input / output systems
(time),
(input output signal spaces),
all input / output pairs.
Thermodynamics: a theory of open systems
terminal(Q ,T )hhWorking
W
(Q ,T )cc
terminalCooling
Heatingterminal
ThermodynamicEngine
First and second law:
These laws deal with ‘open’ systems.
But not with input/output systems!
– p.42/76
![Page 78: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/78.jpg)
Examples
Let’s put Kepler and Newton in this setting.
K1+K2+K3 obviously define a dynamical system� � � � �
� � |� � | ��� all � > | | �
that satisfy Kepler’s 3 laws.
Nice example of a dynamical model ‘without equations’.
Is it a differential system?
This question turned out to be of revolutionary importance...
Input / output systems
(time),
(input output signal spaces),
all input / output pairs.
Thermodynamics: a theory of open systems
terminal(Q ,T )hhWorking
W
(Q ,T )cc
terminalCooling
Heatingterminal
ThermodynamicEngine
First and second law:
These laws deal with ‘open’ systems.
But not with input/output systems!
– p.42/76
![Page 79: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/79.jpg)
Examples
Flows:
��� Y � � Z Y � �
� all state trajectories.
Observed flows:
��� Y � � Z Y � ¡ \ � � ¢ Y � �
� all possible output trajectories.
Note:
1. It may be impossible to express as the solutions of a
differential equation involving only \.2. The auxiliary (latent variable) nature of Y.
Input / output systems
(time),
(input output signal spaces),
all input / output pairs.
Thermodynamics: a theory of open systems
terminal(Q ,T )hhWorking
W
(Q ,T )cc
terminalCooling
Heatingterminal
ThermodynamicEngine
First and second law:
These laws deal with ‘open’ systems.
But not with input/output systems!
– p.42/76
![Page 80: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/80.jpg)
Examples
Input / output systems
ZT \ � � ��� \ � � � ��� � \ � � w w w � �
� Z � [ � � �� � [ � � � ��� � [ � � w w w � �
� � |
(time),� £�¤ ¥
(input
¤
output signal spaces),� all input / output pairs.
Thermodynamics: a theory of open systems
terminal(Q ,T )hhWorking
W
(Q ,T )cc
terminalCooling
Heatingterminal
ThermodynamicEngine
First and second law:
These laws deal with ‘open’ systems.
But not with input/output systems!
– p.42/76
![Page 81: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/81.jpg)
Examples
Input / state / output systems
��� Y � � Z Y � � [ � � � � \ � � ¢ Y � � [ � � �
What do we want to call the behavior?
the
[� \� Y ’s, or the [� \ ’s?
Is the
[� \ behavior described by a differential eq’n?
Thermodynamics: a theory of open systems
terminal(Q ,T )hhWorking
W
(Q ,T )cc
terminalCooling
Heatingterminal
ThermodynamicEngine
First and second law:
These laws deal with ‘open’ systems.
But not with input/output systems!
– p.42/76
![Page 82: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/82.jpg)
Examples
Codes¦ § ¨ � the code; yields the system
� � © � §� ¦ wRedundancy structure, error correction possibilities, etc., are visible
in the code behavior
¦
. It is the central object of study.
Formal languages§ � a (finite) alphabet,ª §« � the language = all ‘legal’ ‘words’ ¬ T ¬ � @ @ @ ¬® @ @ @§« � all finite strings with symbols from
§
.
yields the system
� � �� §� ª w
Examples: All words appearing in the Webster dictionary
All LATEX documents.
Thermodynamics: a theory of open systems
terminal(Q ,T )hhWorking
W
(Q ,T )cc
terminalCooling
Heatingterminal
ThermodynamicEngine
First and second law:
These laws deal with ‘open’ systems.
But not with input/output systems!
– p.42/76
![Page 83: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/83.jpg)
Examples
Thermodynamics: a theory of open systems
Thermodynamics is the only theory of a general nature of which I
am convinced that it will never be overthrown.
Albert Einstein
The law that entropy always increases – the second law of
thermodynamics – holds, I think, the supreme position among the
laws of nature.
Arthur Eddington
terminal(Q ,T )hhWorking
W
(Q ,T )cc
terminalCooling
Heatingterminal
ThermodynamicEngine
First and second law:
These laws deal with ‘open’ systems.
But not with input/output systems!
– p.42/76
![Page 84: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/84.jpg)
Examples
Thermodynamics: a theory of open systems
terminal(Q ,T )hhWorking
W
(Q ,T )cc
terminalCooling
Heatingterminal
ThermodynamicEngine
time-axis:
|
Q: Variables of interest? A: h� x h� ¯� x ¯�� signal space: � |±° ¤ |° ¤ |° ¤ |±° ¤ |
Behavior : a suitable family of trajectories.
But, there are some universal laws that restrict the ’s that are
‘thermodynamic’.
First and second law:
These laws deal with ‘open’ systems.
But not with input/output systems!
– p.42/76
![Page 85: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/85.jpg)
Examples
Thermodynamics: a theory of open systems
terminal(Q ,T )hhWorking
W
(Q ,T )cc
terminalCooling
Heatingterminal
ThermodynamicEngine
First and second law:
h² ¯² �� � / hx h ² ¯x ¯ �� / w
These laws deal with ‘open’ systems.
But not with input/output systems!
– p.42/76
![Page 86: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/86.jpg)
³
: Linear time-invariant differential systems
– p.43/76
![Page 87: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/87.jpg)
More structure
� � � � � is said to be linear
if is a vector space, and a linear subspace of
�
.
– p.44/76
![Page 88: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/88.jpg)
More structure
� � � � � is said to be time-invariant
if
� � |� | ~ � �� or
� ~ and if satisfies
� ! for all
� � � w
� ! denotes the shift, � ! Z � � > � Z � �- �
.
– p.44/76
![Page 89: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/89.jpg)
More structure
� � � � � is said to be differential
if
� � |� or
| ~ , etc., and if is the solution set of a (system of)ODE’s.
a difference system if, etc.
– p.44/76
![Page 90: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/90.jpg)
More structure
� � � � � is said to be symmetric
w.r.t. the transformation group
� xµ´ � ]� ¶ �on
�
if
x´ � for all ]� ¶
.
Examples:
1. time-invariance, time-reversibility
2. permutation symmetry, rotation symmetry, translationsymmetry, Euclidean symmetry,
3. etc., etc.
– p.44/76
![Page 91: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/91.jpg)
·¸
{� |¹ º ¸ �» � { gg I � � /
defines the
linear, time-invariant, differential system:
� � |� |¸ � with
� � �� ¦ L |� |¸ � { gg I � � / � w
– p.45/76
![Page 92: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/92.jpg)
·¸
{� |¹ º ¸ �» � { gg I � � /
defines the
linear, time-invariant, differential system:
� � |� |¸ � with
� � �� ¦ L |� |¸ � { gg I � � / � w
NOTATIONª ¹ > all such systems (with any - finite - number of variables)ª¸ > with ¼ variables
� ª¸(no ambiguity regarding
� � )
– p.45/76
![Page 93: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/93.jpg)
·¸
{� |¹ º ¸ �» � { gg I � � /
defines the
linear, time-invariant, differential system:
� � |� |¸ � with
� � �� ¦ L |� |¸ � { gg I � � / � w
NOMENCLATURE
Elements of
ª ¹ > linear differential systems{ gg I � � / > a kernel representation of the
corresponding
�� ª ¹
or
� ª ¹
– p.45/76
![Page 94: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/94.jpg)
Overview
Starting from this vantage point, a rich theory has been developed
1. Modeling by tearing, zooming, and linking
2. Controllability and stabilizability
3. Control by interconnection:
from stabilization to LQ and L -control
4. Observability, observers and the like
5. SYSID, the MPUM, subspace ID
6. System representations
7. PDE’s
8. etc., etc., ...
– p.46/76
![Page 95: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/95.jpg)
Controllability
– p.47/76
![Page 96: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/96.jpg)
Controllability
Take any two trajectories � "� � �� .
2
0
1w
w
W
time
Controllability:
2
0 T
1w
w
σ wT
W
time
W
– p.48/76
![Page 97: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/97.jpg)
Controllability
Take any two trajectories � "� � �� .
2
0
1w
w
W
time
Controllability:
2
0 T
1w
w
σ wT
W
time
W
– p.48/76
![Page 98: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/98.jpg)
Controllability
The time-invariant system
� � � � �
is said to be
controllable
if for all � "� � �� there exists �� and
x /such that
� � � � " � � ½ /
� � � ² x � x
Controllability >legal trajectories must be ‘patch-able’, ‘concatenable’.
– p.49/76
![Page 99: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/99.jpg)
State Controllability
Special case: classical Kalman definitions for��! Y � Z Y� [ w
controllability: variables = state or (input, state)This is a special case of our controllability:
?
X X
x1
x2
X
x 1
x 2
time
If a system is not (state) controllable, why is it?Insufficient influence of the control?Or bad choice of the state?Or not properly editing the equations?
Kalman’s definition addresses a rather special situation.
– p.50/76
![Page 100: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/100.jpg)
State Controllability
Special case: classical Kalman definitions for��! Y � Z Y� [ w
controllability: variables = state or (input, state)
If a system is not (state) controllable, why is it?Insufficient influence of the control?Or bad choice of the state?Or not properly editing the equations?
Kalman’s definition addresses a rather special situation.
– p.50/76
![Page 101: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/101.jpg)
Tests
Given a system representation, derive algorithms in terms of theparameters for controllability.
Consider the system
� ª ¾
defined by
{ ��� � � / wUnder what conditions on
{� | ¾ ¿ � �» �does it define a
controllable system?
Theorem:
{ ��! � � /defines a controllable system
À �Á  { à � constant over
à � Ä w
– p.51/76
![Page 102: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/102.jpg)
Tests
Notes:
If
{ ��! � � /
has
{
of full row rank, then
controllability
{ Ã
is of full row rank
Å Ã � Ä
.
Equivalently,
{
is right-invertible as a polynomial matrix( ‘left prime’).
– p.52/76
![Page 103: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/103.jpg)
Tests
Notes:
��! Y � � Y- Æ [� � � Y or
Y� [ is controllable iff
À �Á  � �² ÃÇ Æ � � È �?� Y Å Ã � Ä w
Popov-Belevich-Hautus test for controllability.
Of course,
À �Á Â É Æ � Æ @ @ @ � Ê Ë� +*Ì , � " Æ Í � È �?� Y w
– p.52/76
![Page 104: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/104.jpg)
Tests
Notes:
When is t ��� � " � Î ��� � �controllable? t� Î� | �» �
, not both zero.
Controllable rank � t à ² Î Ã � � � Åà � Ä w
Iff t and Î are co-prime. No common factors!
Testable via Sylvester matrix, etc.
Generalizable.
– p.52/76
![Page 105: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/105.jpg)
Stabilizability
The system
� � � � |� �
is said to be stabilizable if, for all�� , there exists � �� such that
� � � � � �
for
� ½ /
and � � � ² !Ï Ð /.
Stabilizability >
legal trajectories can be steered to a desired point.
w’0
1w W
time
– p.53/76
![Page 106: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/106.jpg)
Stabilizability
Consider the system defined by
{ ÑÑÒ ÓÔ / wUnder which conditions on
{� | ¾ ¿ � �» �does it define a
stabilizable system?
Theorem:
{ ��! Ó Ô /defines a stabilizable system
À �Á Â Õ { Õ Ã Ö Ö Ô constant over
� à � Ä ×Ø � � Ù Õ Ã Ö Ú � w
– p.54/76
![Page 107: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/107.jpg)
Image representations
Representations of
ª ¾
:
{ ÛÛÜ Ó Ô Ú
called a ‘kernel’ representation. Sol’n set
� ª ¾, by definition.
{ ÛÛÜ ÓÔ ÛÛÜ Ýcalled a ‘latent variable’ representation of the behavior of theÓ-variables.
‘Elimination th’m’
� ª ¾
.
Missing link:
called an ‘image’ representation of
Elimination theorem every image is also a kernel.
¿¿ Which kernels are also images ?? Controllability!
– p.55/76
![Page 108: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/108.jpg)
Image representations
Representations of
ª ¾
:
{ ÛÛÜ Ó Ô Ú
called a ‘kernel’ representation. Sol’n set
� ª ¾, by definition.
{ ÛÛÜ ÓÔ ÛÛÜ Ýcalled a ‘latent variable’ representation of the behavior of theÓ-variables.
‘Elimination th’m’
� ª ¾
.
Missing link: ÓÔ ÛÛÜ Ý
called an ‘image’ representation of Ô �?� ÛÛÜ w
Elimination theorem every image is also a kernel.
¿¿ Which kernels are also images ?? Controllability! – p.55/76
![Page 109: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/109.jpg)
Image representations
Theorem: (Controllability and image representations):
The following are equivalent for
� ª ¾ßÞ1. is controllable
2. admits an image representation
ÓÔ ÛÛÜ Ý
3. etc., etc.
– p.56/76
![Page 110: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/110.jpg)
Numerical test
Image representation leads to an effective numerical test.à
similar results & algorithms for time-varying systems.à
partial results for nonlinear systems.
– p.57/76
![Page 111: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/111.jpg)
Controllable part
The ‘controllable part’ of
� ª ¹
can be defined in many equivalent
ways. Most expedient:
áâã äæå â ç çéè ê çéë Þ Ô largest controllable
ì� ª¸îí ì
Two systems
y±ï Õ ÑÑÒ Ö Ó ï Ô ï Õ ÑÑÒ Ö Óéð yð Õ ÑÑÒ Ö Ó ï Ô ð Õ ÑÑÒ Ö Óð
have the same controllable part iff they have the same transfer
function yñ ïï ï Ô Þ ò ï Ô òð Þ Ô yñ ïð ð
Transfer function: determines the controllable part only.
Limited description. Limitation of tf. f’n manipulations.– p.58/76
![Page 112: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/112.jpg)
Polynomial representations
Representations with
| �» �
-matrices of
� ª ¹1.
{ ó ggô õ ÓÔ Ú
by definition
2. WLOG:
{
full row rank
3.
{
left prime over
| �» �
(
à öÞ { öÔ Ç) controllable
4. ÓÔ ó ggô õ Ý
controllable
5. if controllable,
WLOG: right prime over
| �» �(
à Þ Ô Ç
)
‘observable image representation’:
à Þ ÝÔ Õ ggô Ö Ó .
– p.59/76
![Page 113: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/113.jpg)
Representations with rational functions
Let
ò� |¹ º ¸ �» �
. What does
ò Õ ggô Ö ÓÔ Úmean?
The behavior defined by is defined as that of
a left co-prime factorization over of
Representations with -matrices of .
1. WLOG, with (strictly) proper, etc.
2. left prime over ring of stable rational f’ns stabilizable
3. controllable
4. if controllable, WLOG: right prime over stable rational f’ns
‘observable im. repr’on’: stable rational .
– p.60/76
![Page 114: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/114.jpg)
Representations with rational functions
Let
ò� |¹ º ¸ �» �
. What does
ò Õ ggô Ö ÓÔ Úmean?
Joint work with
Yutaka Yamamoto
The behavior defined by is defined as that of
a left co-prime factorization over of
Representations with -matrices of .
1. WLOG, with (strictly) proper, etc.
2. left prime over ring of stable rational f’ns stabilizable
3. controllable
4. if controllable, WLOG: right prime over stable rational f’ns
‘observable im. repr’on’: stable rational .
– p.60/76
![Page 115: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/115.jpg)
Representations with rational functions
The behavior defined by
ò Õ ggô Ö Ó Ô Ú
is defined as that of
Õ ggô Ö ÓÔ Ú
òÔ yñ ï
a left co-prime factorization over
| �» �
of
ò
Representations with
| Õ » Ö
-matrices of� ª ¹
.
1. WLOG, with
ò
(strictly) proper, etc.
2.
ò
left prime over ring of stable rational f’ns stabilizable
3. ÓÔ ò Õ ggô Ö Ýcontrollable
4. if controllable, WLOG:
òright prime over stable rational f’ns
‘observable im. repr’on’:
à÷
stable rational Þ ÝÔ ÷ Õ ggô Ö Ó .– p.60/76
![Page 116: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/116.jpg)
PDE’s
– p.61/76
![Page 117: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/117.jpg)
PDE’s
Much of the theory also holds for PDE’s.� Ô | øí the set of independent variables, often ùÔ ú,Ô |� í the set of dependent variables,Ô sol’ns of a linear constant coefficient system of PDE’s.
Let and consider
Define the associated behavior
holds
Notation for n-D linear differential systems:or
– p.62/76
![Page 118: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/118.jpg)
PDE’s
Much of the theory also holds for PDE’s.� Ô | øí the set of independent variables, often ùÔ ú,Ô |� í the set of dependent variables,Ô sol’ns of a linear constant coefficient system of PDE’s.
Let
{� | ¾ ¿ � �»üû í ý ý ý í » ø �í and consider
{ þþÌÿ í ý ý ý í þþÌ�� Ó Ô Ú w Õ � Ö
Define the associated behavior
Ô � Ó� ¦ Ð Õ | øí |� Ö × Õ � Ö holds
� w
Notation for n-D linear differential systems:Õ | øí |� í Ö � ª � øí or
� ª � ø w – p.62/76
![Page 119: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/119.jpg)
Example
Maxwell’s eq’ns, diffusion eq’n, wave eq’n, w w w
��� ��� � ï � �
� º ��� � ñ �� ô ������� ��� � � �¯ � � º � � � ï ���� �� ô �����
(time and space) ,
(electric field, magnetic field, current density, charge density),,
set of solutions to these PDE’s.
Note: 10 variables, 8 equations! free variables. ‘open’ system.
– p.63/76
![Page 120: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/120.jpg)
Example
Maxwell’s eq’ns, diffusion eq’n, wave eq’n, � � �
��� ��� � ï � �
� � ��� � ñ �� ô � ���� ��� � � �� � � � � � � ï ���� �� ô �����
� Ô � � � �
(time and space) ùÔ ú,Ó Ô �! í �#"í �%$í &
(electric field, magnetic field, current density, charge density),Ô � � � � � � � � � �í 'Ô ( Ú
,Ô set of solutions to these PDE’s.
Note: 10 variables, 8 equations!
à
free variables. ‘open’ system.– p.63/76
![Page 121: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/121.jpg)
Submodule theorem
)+* � , - , ./üû í ý ý ý í / ø 0 defines 1 2�34 ) þþ�5ÿ í ý ý ý í þþ�5� ,
but not vice-versa.
¿¿
à
‘intrinsic’ characterization of* 6 7 ø 8 8
Is there a mathematical ‘object’ that characterizes a
* 6 7 ø 8
Define the annihilators of
* 6 7 ø by
9 Þ 1 :<; * � 7 ./üû í ý ý ý í / ø 0= ; > þþ�5ÿ í ý ý ý í þþ�5� 1 Ú ? �
Proposition: 9 is a� ./7û í ý ý ý í / ø 0 sub-module of
� 7 ./7û í ý ý ý í / ø 0 �
Theorem:bijective
submodules of
– p.64/76
![Page 122: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/122.jpg)
Submodule theorem
)+* � , - , ./üû í ý ý ý í / ø 0 defines 1 2�34 ) þþ�5ÿ í ý ý ý í þþ�5� ,
but not vice-versa.
¿¿
à
‘intrinsic’ characterization of* 6 7 ø 8 8
Is there a mathematical ‘object’ that characterizes a
* 6 7 ø 8
Define the annihilators of
* 6 7 ø by
9 Þ 1 :<; * � 7 ./üû í ý ý ý í / ø 0= ; > þþ�5ÿ í ý ý ý í þþ�5� 1 Ú ? �
Proposition: 9 is a� ./7û í ý ý ý í / ø 0 sub-module of
� 7 ./7û í ý ý ý í / ø 0 �
Theorem:6 7 ø bijective
submodules of
� 7 ./7û í ý ý ý í / ø 0
– p.64/76
![Page 123: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/123.jpg)
Elimination theorem
Motivation: In many problems, we want to eliminate variables. Forexample, first principle modeling
@ model containing both variables the model aims at (‘manifest’variables), and auxiliary variables introduced in the modelingprocess (‘latent’ variables).
¿ Can these latent variables be eliminated from the equations ?– p.65/76
![Page 124: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/124.jpg)
Elimination theorem
This leads to the following important question, first in polynomialmatrix language. Consider
)û A BBDC û í ý ý ý í BBC ø E F û 1 )�G A BBDC û í ý ý ý í BBC ø EF G �
Obviously, the behavior of the
AF û í F G E’s is described by a system
of PDE’s. ¿ Is the behavior of the F û ’s alone also ?
B
W
W1
W x 21
B1
– p.66/76
![Page 125: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/125.jpg)
Elimination theorem
In the language of behaviors:
Let
* 6 7ÿ H 7 �ø . Define
û 1 :F û * I J A � øí � 7ÿ E= à F G such thatAF û í F G E * ? �
Does this ‘projection’ û belong to
6 7ÿ ø ?
B
W
W1
W x 21
B1
Theorem: It does!
6 ,
is closed under projection !!– p.66/76
![Page 126: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/126.jpg)
The Fundamental Principle
Proof: ‘Fundamental principle’. Consider
÷ A C E 1 K
Given:
÷ Þ L Mí K * M
; Unknown: C * L
.
¿ Does there exists a sol’n C ?
Examples:
1.
2.
3.
– p.67/76
![Page 127: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/127.jpg)
The Fundamental Principle
Proof: ‘Fundamental principle’. Consider
÷ A C E 1 K
Given:
÷ Þ L Mí K * M
; Unknown: C * L
.
¿ Does there exists a sol’n C ?
Examples:
1.
÷ * � øÿ - ø �í K * � ø �í C * � øÿ2.
3.
– p.67/76
![Page 128: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/128.jpg)
The Fundamental Principle
Proof: ‘Fundamental principle’. Consider
÷ A C E 1 K
Given:
÷ Þ L Mí K * M
; Unknown: C * L
.
¿ Does there exists a sol’n C ?
Examples:
1.
2. ODE’s: ÷ A NNO E C 1 K
with
÷ * � øÿ - ø � ./ 0í K * I J A �í � ø � Eí C * I J A �í � øÿ E
.
Or over distributions, K * P A �í � ø � Eí C * P A �í � øÿ E
.
3.– p.67/76
![Page 129: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/129.jpg)
The Fundamental Principle
Proof: ‘Fundamental principle’. Consider
÷ A C E 1 K
Given:
÷ Þ L Mí K * M
; Unknown: C * L
.
¿ Does there exists a sol’n C ?
Examples:
1.
2.
3. PDE’s: ÷ A BBC û í ý ý ý í BBDC ø E C 1 K
÷ * � øÿ - ø � ./üû í ý ý ý í / ø 0í K * I J A � øí � ø � EíC * I J A � øí � øÿ E
, or over distributions. – p.67/76
![Page 130: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/130.jpg)
The Fundamental Principle for PDE’s
÷ A þþ�5ÿ í ý ý ý í þþ�5� E C 1 KGiven:
÷ * � øÿ - ø � ./7û í ý ý ý í / ø 0í K * I J A � øí � ø � EíUnknown: C * I J A � øí � øÿ E
.
¿ Does there exists a sol’n C ?
Obvious necessary condition:
A; * � Qÿ ./üû í ý ý ý í / ø 0 ESR A ù > A /üû í ý ý ý í / ø E ÷ A /üû í ý ý ý í / ø E 1 Ú E
; > A BBDC û í ý ý ý í BBDC ø E K 1 Ú �
Theorem (Fundamental principle): This is a n.a.s.c.
– p.68/76
![Page 131: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/131.jpg)
The Fundamental Principle for PDE’s
÷ A þþ�5ÿ í ý ý ý í þþ�5� E C 1 KGiven:
÷ * � øÿ - ø � ./7û í ý ý ý í / ø 0í K * I J A � øí � ø � EíUnknown: C * I J A � øí � øÿ E
.
¿ Does there exists a sol’n C ?Theorem (Fundamental principle): This is a n.a.s.c.
Since the ; ’s form a (finitely generated)
� ./ û í ý ý ý í / ø 0 -module,this is a finite condition!
Example:
Take
Ú T 1 ÷ * � ./üû í ý ý ý í / ø 0 . PDE
÷ A þþ�5ÿ í ý ý ý í þþ�5� E C 1 K.Always solvable!
– p.68/76
![Page 132: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/132.jpg)
The elimination theorem
There exist effective algorithms for
A )û í )�G EVU ).@ Computer algebra, Grobner bases.
It follows from all this that
6 ,ø has very nice properties. Inparticular, it is closed under:
Intersection:
A û í G * 6 7 ø E A û W G * 6 7 E
Addition:
A û í G * 6 7 ø E A û X G * 6 7 ø E
Projection:
YZ [ \]ÿ � ] �^ _a` Yb cÿ Z [ \]ÿed _ b cÿ f projection
Action of a linear differential operator:YZ [ \]ÿ d g h [ i] � � ]ÿ jk�l g m m m g k ^ n _` Y h Y oop _ Z [ \] �^ _
.
Inverse image of a linear differential operator:YZ [ \] �^ g h [ i] � � ]ÿ jkql g m m m g k ^ n _` Y h Y oop _ _sr l Z [ \]ÿ ^ .– p.69/76
![Page 133: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/133.jpg)
Elimination theorem
Which PDE’s describe ( &ut �! t �%$
) in Maxwell’s equations ?
Eliminate
�#"
from Maxwell’s equations @ý �! 1 (
vxw & t
vxw BB O ý �! X ý �$ 1 y t
vzw B GB O G �! X vzw { G � � �! X BB O �$ 1 y �
– p.70/76
![Page 134: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/134.jpg)
) ||�5} t ~ ~ ~ t ||�5�� F 1 yis called a kernel representation of the associated
* 6 7�.Another representation: image representation
F 1 ||�5} t ~ ~ ~ t ||�5�� � �
‘Elimination’ thm��� ||�5} t ~ ~ ~ t ||�5�� * 6 7� �
Which linear diff. systems admit an image representation???
admits an image representation iff it is ‘controllable’.
– p.71/76
![Page 135: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/135.jpg)
) ||�5} t ~ ~ ~ t ||�5�� F 1 yis called a kernel representation of the associated
* 6 7�.Another representation: image representation
F 1 ||�5} t ~ ~ ~ t ||�5�� � �
‘Elimination’ thm��� ||�5} t ~ ~ ~ t ||�5�� * 6 7� �
Which linear diff. systems admit an image representation???
* 6 7� admits an image representation iff it is ‘controllable’.– p.71/76
![Page 136: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/136.jpg)
Controllability for PDE’s
Controllability def’n in pictures:
� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �
� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �
� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �
�
��
���
��� �
��
F�� t F G *
.
– p.72/76
![Page 137: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/137.jpg)
Controllability for PDE’s
� F *
‘patches’ F � t F G *
.
� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �
� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �� � � � � � � � � � � � � � � � � � � � � � � �
� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �
� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �
� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � �� � � � � � � ��
�
��
Controllability : ‘patch-ability’.
– p.73/76
![Page 138: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/138.jpg)
Are Maxwell’s equations controllable ?
The following equations in the scalar potential
��� � � � �
and the vector potential
¡£¢� � � � �
, generate exactly thesolutions to Maxwell’s equations:
¤�¥ ¦ r §§ p ¤©¨ r ª «¬¤�® ¦ ª � ¤a¨¬¤©¯ ¦ °± § ²§ p ² ¤ ¨ r °± � ² ª ² ¤³¨´ °± � ² ª µ ª�¶ ¤³¨ · ´ °± §§ p ª «¬
¸ ¦ r °± §§ p ª¶ ¤³¨ r °± ª ² «¹Proves controllability. Illustrates the interesting connection
controllability
�
potential!
– p.74/76
![Page 139: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/139.jpg)
Conclusion
The flexibility and generality of the behavioral approach in modeling,
for system representations, for passive control, dealing with PDE’s,
etc. is evident.
Exemplified by the notion of controllability.
Nature and Nature’s laws lay hid in night
God said, ‘Let Newton be’ and all was light
Mathematical Systems Theory lay bound by might
Ratio said, ‘Let Behaviors be’ and all was right
– p.75/76
![Page 140: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/140.jpg)
Conclusion
The flexibility and generality of the behavioral approach in modeling,
for system representations, for passive control, dealing with PDE’s,
etc. is evident.
Exemplified by the notion of controllability.
Nature and Nature’s laws lay hid in night
God said, ‘Let Newton be’ and all was light
Mathematical Systems Theory lay bound by might
Ratio said, ‘Let Behaviors be’ and all was right
– p.75/76
![Page 141: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/141.jpg)
Details & copies of the lecture frames are available from/at
http://www.esat.kuleuven.be/ ºjwillems
Thank youThank you
Thank youThank you
Thank you
Thank you
Thank you
Thank you
– p.76/76
![Page 142: THE BEHAVIORAL APPROACH - KU Leuvensistawww/smc/jwillems/... · 2005-09-07 · 2. A brief history of systems theory 3. Why are better framework is needed 4. Models and behaviors Second](https://reader034.vdocuments.net/reader034/viewer/2022042401/5f1056217e708231d4489b59/html5/thumbnails/142.jpg)
Details & copies of the lecture frames are available from/at
http://www.esat.kuleuven.be/ ºjwillemsThank you
Thank youThank you
Thank you
Thank you
Thank you
Thank you
Thank you
– p.76/76