Aa547 Lecture Lec1

EE547: Fall 2018
Lecture 1: Introduction
Lecturer: L.J. Ratliff
Disclaimer: These notes have not been subjected to the usual scrutiny reserved for formal publications.
They may be distributed outside this class only with the permission of the Instructor.
1.1 Today
• Introduction to the broad concepts of modeling and analysis of engineering systems
– modeling
– analysis and control
– verification
– simulation
– validation
• Course overview
1.2 Introduction: What are dynamical systems?
We are interested in studying the dynamical behavior of systems in order to analyze and control them.
Examples of systems exhibiting dynamical behavior, with possible inputs to be controlled and outputs to be
measured, include:
system input output

steam engine amount of fuel engine speed
aircraft rudder and elevator angles altitude
quadrotor thrust & torque position
hard drive disks motor voltage disk speed
rainforest ecology yearly rainfall population of species of interest
cooling system flow-rate of coolant temperature along cooling line
To study such systems we need a model. What is in a model?
• A mathematically precise description of system behavior

• Dynamic models specify how system changes over time
• Balancing act:
– Can’t model movement of every atom of a system
– Therefore, make simplifications
– Need to retain essential properties (salient features) (e.g., a model that predicts negative popula-
tion, or coolant temperature below absolute zero, is probably not useful)
0
1-1
1-2 Lecture 1: Introduction
1.3 Beyond Classical Control?
This course is about linear systems theory and a core aspect of that is controlling linear systems.
Control Theory: focuses on modeling systems and designing inputs to adjust behavior—e.g., stabilize, track
a trajectory, etc.
Classical control (developed largely pre-1960s) largely adopts an input-output approach:
reference input output

System
−
controller
Key theoretical tool : Fourier/Laplace transform (i.e. Frequency domain analysis—root locus, frequency re-
sponse)
Return to differential equations beginning in the ’60s to address:
• numerical simulation
• many inputs/outputs
• ill-defined inputs/outputs
• non-linearities
• optimality
Modern control theory (∼1950’s), i.e. state-space approach: Overcame some limitations of classical
control enabling control of fighter jets, e.g. (related “state space” approach to ODE’s is over 100 years old;
control theorist just adopted it)
• System/model state is defined to capture all relevant info about past

• State often denoted by x ∈ Rn , where n is state-space dimension.
What is the state for the above examples?
• engine speed/velocity
• position and yaw/pitch/roll and velocities (aircraft and quadrotor)
• disk speed/velocity
• species population, food supply, predator population, etc.
• temperature along cooling line
Lecture 1: Introduction 1-3
1.4 Dynamical Systems (non-comprehensive) Taxonomy
(This is not comprehensive as there are other types of systems combining various aspects in the diagram;
however, this picture give a bit of a sense of broad categories of dynamical systems)
dynamical systems
mathematical models
continuous time discrete time discrete state
non-linear non-linear
time- time- time- time-

varying invariant varying invariant
linear linear
time- time- time- time-

varying invariant varying invariant
This course:
given linear differential or difference equations, conduct rigorous analysis and design.
1.5 Finite Dimensional Systems as Models
How to describe a dynamical system? Some options:
• Database/look-up table containing all inputs and resulting outputs. (What if output depends on input
history? What if desired input is not in table?)
• Function/routine in computer code
• Set of mathematical equations
As indicated in our diagram, we will study continuous-time and discrete-time finite-dimensional systems
described by ordinary differential equations or difference equations.
So, in continuous time this might look like. . .

Continuous-Time: (t ∈ R+ := [0, ∞))
ẋ = f (t, x, u), x ∈ Rn , u ∈ Rm
y = g(t, x, u), y ∈ Rp
We have the following nomenclature:
• x is the state,
• u is the (control) input,
• y is the output (observation)
Discrete-Time: (k ∈ N = {0, 1, 2, . . .})
x[k + 1] = f (k, x, u), x ∈ Rn , u ∈ Rm

y[k] = g(k, x, u), y ∈ Rp
What about apparently more exotic systems with higher order derivatives?
For instance, consider
z (n) = f (t, z, z (1) , z (2) , . . . , z (n−1) )
where z (n) indicates the n–th derivative of the function z(t). For simplicity, assume z ∈ R. Define new state
variables
x1 = z, x2 = z (1) , . . . , xn = z (n−1)
Then, we have
ẋ1 = x2
ẋ2 = x3
.. ..
. .
ẋn = f (t, x1 , x2 , . . . , xn )
Thus, without loss of generality (w.l.o.g), we study first-order differential equations.
Example Systems: Vibrating Springs

We consider the motion of an object with mass at the end of a spring that is either vertical (as in Figure 1.1)
or horizontal on a level surface (as in Figure 1.2). Hooke’s Law says that if the spring is stretched (or
compressed) z units from its natural length, then it exerts a force that is proportional to z—that is,
restoring force = −kx
where k is a positive constant called the spring constant. If we ignore any external resisting forces (due to
air resistance or friction) then, by Newton’s Second Law (force equals mass times acceleration), we have
mẍ = −kx or mẍ + kx = 0
k k
Let x1 = z and x2 = ż. Then, ẋ1 = ż = x2 and ẋ2 = z̈ = − m z = −m x1 . Hence,
ẋ1 = x2
k
ẋ2 = − x1
m
Lecture 1: Introduction 1-5
equilibrium
m 0
position
z m
Figure 1.1: Vertical Pull
0 z
Figure 1.2: Horizontal Pull
In this course, we will focus on (finite-dimensional) linear time-varying (LTV) systems. So what do
these look like notationally:
Continuous Time:
ẋ = A(t)x + B(t)u, x ∈ Rn , u ∈ Rm
y = C(t)x + D(t)u, y ∈ Rp
where
• t ∈ R: time
• x(t) ∈ Rn : state (vector)
• u(t) ∈ Rm : input or control
• y(t) ∈ Rp : output
• A(t) ∈ Rn×n : dynamics (matrix)
• B(t) ∈ Rn×m : input matrix
• C(t) ∈ Rp×n : output or sensor matrix
• D(t) ∈ Rp×m : Feedthrough matrix
Equations are often written as
ẋ = Ax + Bu
y = Cx + Du
• A CT LDS is a first order vector differential equation

• also called state equations or m-input, n-state, p-output LDS
Discrete Time:
x[k + 1] = A[k]x[k] + B[k]u[k], x ∈ Rn , u ∈ Rm

y[k] = C[k]x[k] + D[k]u[k], y ∈ Rp
Finally, we will further specialize our results to linear time-invariant (LTI) systems.
Continuous Time LTI:
ẋ = Ax + Bu, x ∈ Rn , u ∈ Rm
y = Cx + Du, y ∈ Rp
where A ∈ Rn×n , B ∈ Rn×m , C ∈ Rp×n , and D ∈ Rp×m are static matrices.
Discrete Time LTI:
x[k + 1] = Ax[k] + Bu[k], x ∈ Rn , u ∈ Rm

y[k] = Cx[k] + Du[k], y ∈ Rp
Some other points:

• most linear systems encountered are time-invariant: A, B, C, D are constant as above—i.e. don’t depend
on t
• when there is no input u (hence, no B or D) system is called autonomous
• very often there is no feedthrough—i.e. D = 0
• when u(t) and y(t) are scalar, the system is called single-input, single-output (SISO); when input & output
signal dimensions are more than one, MIMO
History Lesson:
• parts of LDS theory can be traced to 19th century
• builds on classical circuits & systems (1920s on) (transfer functions . . . ) but with more emphasis on linear
algebra
• first engineering application: aerospace, 1960s
• transitioned from specialized topic to ubiquitous in 1980s (just like digital signal processing, information
theory, . . . )
Many dynamical systems are nonlinear, yet
• most techniques for nonlinear systems are based on linear methods; e.g., linearization to determine stability
or to construct extended Kalman Filters. . .
• methods for linear systems often work unreasonably well, in practice, for nonlinear systems
• if you don’t understand linear dynamical systems you certainly cant understand nonlinear dynamical
systems

Aa547 Lecture Lec1

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Aa547 Lecture Lec1

Uploaded by

Copyright:

Available Formats

EE547: Fall 2018

1.2 Introduction: What are dynamical systems?

system input output

To study such systems we need a model. What is in a model?

• A mathematically precise description of system behavior

1.3 Beyond Classical Control?

Classical control (developed largely pre-1960s) largely adopts an input-output approach:

reference input output

• System/model state is defined to capture all relevant info about past

What is the state for the above examples?

1.4 Dynamical Systems (non-comprehensive) Taxonomy

continuous time discrete time discrete state

time- time- time- time-

time- time- time- time-

1.5 Finite Dimensional Systems as Models

How to describe a dynamical system? Some options:

So, in continuous time this might look like. . .

Continuous-Time: (t ∈ R+ := [0, ∞))

We have the following nomenclature:

Discrete-Time: (k ∈ N = {0, 1, 2, . . .})

x[k + 1] = f (k, x, u), x ∈ Rn , u ∈ Rm

Thus, without loss of generality (w.l.o.g), we study first-order differential equations.

Example Systems: Vibrating Springs

restoring force = −kx

mẍ = −kx or mẍ + kx = 0

Figure 1.1: Vertical Pull

Figure 1.2: Horizontal Pull

• A CT LDS is a first order vector differential equation

x[k + 1] = A[k]x[k] + B[k]u[k], x ∈ Rn , u ∈ Rm

Continuous Time LTI:

where A ∈ Rn×n , B ∈ Rn×m , C ∈ Rp×n , and D ∈ Rp×m are static matrices.

Discrete Time LTI:

x[k + 1] = Ax[k] + Bu[k], x ∈ Rn , u ∈ Rm

Some other points:

You might also like