EED3016 Control Systems
Asst. Prof. Dr. Hatice Doğan 2012‐2013 Spring
Resources
• Textbooks:
• Ogata, Modern Control Eng., Pearson. • Goodwin et.al. Control Syst. Design, Prentice Hall • Kuo Automatic Control Systems, Prentice Hall. • Dorf and Bishop, Modern Control Systems, Wiley.
– Used Slide Presentations:
• EE2010E: Systems and Control, Lecture Slides, Ben M. Chen. • MCEN467:Control Systems, Otto Friedrich.
Lectures
• Attendance is essential!!!
• Ask any question at any time during the lecture.
Engineering
• Engineering is concerned with understanding and controllin g the materials and forces of nature for the benefit of humankind.
• Control system engineers are concerned with understanding and controlling segments of their environment, often called systems, to provide useful economic products for society.
What is system?
• A system is a combination of components that act together and perform a certain objective.
• Examples include financial systems such as stock market, social systems such as government, weather, the human body, electrical systems such as electric circuits, mechanical systems, etc…
Block diagram representation of a system
Linear Systems
• So a system is called linear if the principle of superposition applies.
• The principle of superposition states that the response produced by the simultaneous application of two different forcing functions is the sum of the two individual responses.
Example for nonlinear system
Example for time‐invariant system
Example for time‐variant system
Systems with memory and without memor y
What is a control system?
• A control system is an interconnection of components forming a system configuration that will provide a desired system response.
Control of a fighter aircraft
Historical Perspective
• 250 BC, Hierapolis sawmill: Roman water‐powered stone saw mill combining a crank with a connecting rod
• Similar stone saw mill found also in Ephesus
1136‐1206 El Cezire, Artuks Türkmen Denisty, Diyarbakır‐Cizre Chief Engineer of Pallace
Water
mill
Crankshaft Cshaft.gif
Historical Perspective
• 1600 Drebbel Temperature regulator • 1681 Pressure regulator for steam boilers • 1765 Polzunov water level float regulator=
(For pumping up water out of mine;
• Archimedes screw, Syracuse, 287‐212 BC)
Flyball governor used on a steam engine in a cotton factory near Manchester which was at
the centre of the industrial revolution.
The above flyball governor in the same
cotton factory was used to regulate the
speed of a water wheel driven by the
flow of the river.
Invented! by J. Watt 1788 to reject the
load disturbances also
Steam Engine J. Watt 1736‐1819
●● 1769 1769 James
James Watt’s
Watt’s Steam
Steam Engine and Governor
Engine and Governor
●● 1868
1868
J. J. Clerk
Clerk Maxwell
Maxwell formulates
formulates aa mmathematical
athematical model
model for
for
governor
governor control
control of of aa steam
steam engine
engine
●● 1927
1927
Harold
Harold Black
Black discovers
discovers and and
patents
patents the the feedback
feedback amplifier
amplifier
●● 1927
1927
Hendrik
Hendrik Bode
Bode analyzes
analyzes feedback
feedback amplifiers
amplifiers
●● 1932
1932
Nyquist
Nyquist develops
develops methods
methods for
for analyzing
analyzing feedback
feedback amplifier
amplifier
stability
stability
• 1940s Norbert Wiener leads gun positioning effort; becomes an engineering discipline
• 1950s Increased use of Laplace transform,
s‐plane, root locus (by Evans for stability analysis)
• 1960s Sputnik, highly accurate control systems for space vehicles, missiles and robotics
• 1980s Routine use of digital computers as control elements
• 1990s Feedback control in automobiles, automation, planetary exploration
Basic Terminology
Controlled Variable and Control Signal or Manipulated Variable The controlled variable is the quantity or condition that is measured and controlled. The control signal or manipulated variable is the quantity or condition that is varied by the controller so as to affect the value of the controlled variable. Normally, the controlled variable is the output of the system.
Control means measuring the value of the controlled variable of the system and applying the manipulated variable to the system to correct or limit deviation of the measured value from a desired value.
Plants: A plant may be a piece of equipment, perhaps just a set of machine parts functioning together, the purpose of which is to perform a particular operation. Plant is a physical object to be controlled (such as a mechanical device, a heating furnace, a chemical reactor, or a spacecraft)
Processes: Any operation to be controlled a process. Examples are chemical, economic, and biological processes.
Disturbances: A disturbance is a signal that tends to adversely affect the value of the output of a system. If a disturbance is generated within the system, it is called internal, while an external disturbance is generated outside the system and is an input.
Feedback Control: Feedback control refers to an operation that, in the presence of disturbances, tends to reduce the difference between the output of a system and some reference input and does so on the basis of this difference.
Here only unpredictable disturbances are so specified, since predictable or known disturbances can always be compensated for within the system.
• In speed control system, the plant (controlled system) is the engine.
• The controlled variable is the speed of the engine. •
• The difference between the desired speed and the actual speed is the error signal.
• The control signal (the amount of fuel) to be applied to the plant (engine) is the actuating signal.
• The external input to disturb the controlled variable is the disturbance. An unexpected change in the load is a disturbance.
Open‐Loop Control Systems
Desired
Controller
Plant
Output
Response
Output
•

Those systems in which the output has no effect on the control action are called open‐loop control systems. In other words, in an open‐loop control system the output is neither measured nor fed back for comparison with the input.

•

One practical example is a washing machine. Soaking, washing, and rinsing in the washer operate on a time basis. The machine does not measure the output signal, that is, the cleanliness of the clothes.

• In any open‐loop control system the output is not compared with the reference input. Thus, to each reference input there corresponds a fixed operating condition; as a result, the accuracy of the system depends on calibration. In the presence of disturbances, an open‐loop control system will not perform the desired task. Open‐loop control can be used, in practice, only if the relationship between the input and output is known and if there are neither internal nor external disturbances. Clearly, such systems are not feedback control systems. Note that any control system that operates on a time basis is open loop.
Closed‐Loop (Feedback) Control Systems
• A system that maintains a prescribed relationship between the output and the reference input by comparing them and using the difference as a means of control is called a feedback control system.
• In a closed‐loop control system the actuating error signal, which is the difference between the input signal and the feedback signal, is fed to the controller so as to reduce the error and bring the output of the system to a desired value. The term closed‐loop control always implies the use of feedback control action in order to reduce system error.
Closed‐Loop versus Open‐Loop Control Systems
• An advantage of the closed‐loop control system is the fact
that the use of feedback makes the system response
relatively insensitive to external disturbances and internal
variations in system parameters.
• For systems in which the inputs are known ahead of time
and in which there are no disturbances it is advisable to use
open‐loop control.
• Closed‐loop control systems have advantages when
unpredictable disturbances and/or unpredictable variations
in system components are present.
•
The major advantages of open‐loop systems

1. Simple construction and ease of maintenance.

2. Less expensive than a corresponding closed‐loop system.

3. There is no stability problem.

4. Convenient when output is hard to measure or measuring the output precisely is economically not feasible. (For example, in the washer system, it would be quite expensive to provide a device to measure the quality of the washer's output, cleanliness of the clothes.)
The major disadvantages of open‐loop systems

1. Disturbances and changes in calibration cause errors, and the output may be different from what is desired.

2. To maintain the required quality in the output, recalibration is necessary from time to time.
Open Loop Educational System:
Learning
Objectives‐aims
Disturbances
y = Ka + b
Closed Loop Educational System
Objectives‐aims
Reference
a
Disturbances
b
y
Learner
Assesment/
Evaluation
_
y
y=K(a‐y)+b
y= a.K/(K+1)+b.1/(K+1)
For High Control Gains K’s: y = a
Control System Design
• The goal of control enginnering design is to obtain the
configuration, specifications and identification of the key
parameters of a proposed system to meet an actual need.
The design process:

1. Establishment of goals and variables to be controlled and definition of specifications (metrics) againist which to measure performance

2. System definition and modelling

3. Control system design and integrated system simulation and analysis
Cruise Controller Design
• Objective: Design a cruise controller to regulate the speed
of a car to a desired value (say v _{0} )
• Design Procedure
– Step 1: Identif y the relevant components/systems
– Step 2: Specify their interconnection through block
diagrams
(accelerator pedal (throttle) position ,, brake pedal position)
(accelerator pedal (throttle) position brake pedal position)
Disturbances
Disturbances
(air drag, grade,
(air drag, grade,
Actuator Noise
Actuator Noise
Control I/P
Control
I/P
friction etc)
friction etc)
Physical Process
Physical Process
Actuators
Actuators
Actual output
Actual output
Desired
Desired
(vehicle speed)
(vehicle speed)
Control I/P
Control I/P
Sensed O/P
Sensed O/P
Cruise Cont.
Cruise Cont.
Sensors
Sensors
Reference
Reference Input
Input
Sensor Noise
Sensor Noise
(desired vehicle speed)
(desired vehicle speed)
(wheel speed sensor)
(wheel speed sensor)
Cruise Controller Design
• Step 3: Establish a map from control input to the output(s) of interest i.e., establish a mathematical model for the physical process. This map will be used to determine the control algorithm.
– Required attributes of the model:
•
The model should adequately represent reality
• For ease of control design, the model should be simple – “Usual” procedure to construct control‐oriented models • Invoke relevant physics to describe the process under investigation • Make approximations/assumptions to simplify model structure
Disturbances
Disturbances
Physical Process
Physical Process
Actual Output
Actual
Output
Controller Design
• Step 4: Controller Design
– The controller is a map from actual output (v) and
desired output (v _{0} ) to the desired control input
– The controller should account for the dynamics of the
process under control
Example Control Systems
• Mechanical and Electro‐mechanical Control Systems • Thermal (e.g. Temperature) Control System • Pneumatic Control System • Fluid (Hydraulic) Control Systems • Complex Control Systems
• Industrial Controllers
–
–
–
–
On‐off Controllers Proportional Controllers Integral Controllers Proportional‐plus‐Integral Controllers Proportional‐plus‐Derivative Controllers
– – Proportional‐plus‐Integral‐plus‐Derivative Controllers
General Analysis Approach for Control Systems
Define the physical problem properly
Define the system and its components
Formulate the mathematical model
─ Newton’s laws: translational and rotational forces
─ Kirchoff’s laws: mesh and node equations
List the necessary assumptions
Write the differential equations for the model
Put the equations in standard form
Solve the equations for the desired variables
Examine the solution for reasonableness
Mechanical‐Chemical‐Electrical… Control Systems
How to model the system‐plant?
How to Analyze the model?
How to Design a Controller?
How to redesign‐tune?
Modeling (Art&Science)
• Select a model (model identification)
• Realistic (enough complexity)
• Simple (Occam’s razor & generalization)
• Find model parameters (parameter identification)
• Describe model errors
• Linearize nonlinear models
The first
•• The
first way way:: either
either by by trial
trial‐‐error
error or
or by by anan algorithm,
algorithm, model
model parameters
parameters until until
ddyynamic namic behavior behavior of of model model and and pplant lant match sufficientlyy well
match sufficientl well
•• An An alternative
alternative approach
approach for
for dealing
dealing with with modeling
modeling problem
problem isis toto use use physical
physical
laws (such as conservation
laws (such as
conservation ooff mass, energy and
mass, energy and momentum)
momentum) toto construct
construct the
the
model.
model.
•• Control
Control relevant
relevant models
models are
are often
often quite
quite simple
simple compared
compared toto the
the true
true process
process
and and usually
usually ccombine
ombine physical
physical reasoning with experimental
reasoning with
experimental data.
data.
Mathematical Model
• To understand and control complex systems, one must obtain quantitative mathematical models of these systems. It is necessary therefore to analyze the relationships between the system variables and to obtain a mathematical model.
• The systems are dynamic in nature so the descriptive equations are usually differential equations.
Ideal Control System Elements
Element Type

Physical Element

Describing Equation

Energy (E) or power (Þ)

Inductive storage

Electrical Inductance

v

= L di/dt


E = (1/2) L i ^{2}



Translational spring

dx/dt = (1/k) dF/dt

E = (1/2) F ^{2} / k



Rotational spring

ω = (1/k) dT/dt

E = (1/2) T ^{2} / k



Fluid inertia


P = I dQ/dt


E = (1/2) I Q ^{2}


Capacitive storage

Electrical capacitance



E = (1/2) C v ^{2}



Translational mass

F = M d ^{2} x/dt ^{2}

E = (1/2) M ( dx/dt ) ^{2}


R
t
ti
l
o a ona mass

T

J d
=

ω

/dt

E

1/2 J
)
= (

ω



Fluid capacitance

Q = C _{f} d P/dt

E = (1/2) C _{f} P ^{2}



Thermal capacitance

q = C dŦ/dt


E = C Ŧ ^{2}


Energy dissipators

Electrical resistance



Þ = i ^{2} R = v ^{2} / R


Translational damper

F = b dx/dt


Þ = b ( dx/dt ) ^{2}



Rotational damper

T = bω


Þ = bω ^{2}



Fluid resistance

Q = ( 1/ R _{f} ) P

Þ = ( 1/ R _{f} ) P ^{2}



Thermal resistance

q = ( 1/ R _{t} ) Ŧ

Þ = ( 1/ R _{f} ) Ŧ


Differential Equations of Physical Systems:
Example 1: Mechanical System
2
d
y ( t )
dy t
( )
M
+
b
+
ky t
( )
=
r t
( )
2
dt
dt
Example2: Electrical System
v ( t )
R
+
C
dv t
( )
d
t
+
1
L
t
∫
( )
v t
dt
0
=
( )
r t
Example: Cruise‐control System
Linear Differential Equations
I/O REPRESENTATIONS for LINEAR SYSTEMS
x(t)
Linear System
• A system is linear iff
If x 1 (t)
y 1 (t)
and
x 2 (t)
y 2 (t)
y(t)
then k _{1} x _{1} (t) + k _{2} x _{2} (t)
Example:: Affine
Example
Affine System
System
k _{1} y _{1} (t) + k _{2} y _{2} (t) for all constants k _{1} _{,} k _{2}
If x (t)
1
y (t) = kx (t) + b and x (t)
1
1
2
y (t) = kx (t) + b
2
2
Let x _{3} (t) =x _{1} (t) + x _{2} (t) ≠ y _{1} (t) + y _{2} (t)
y _{3} (t) = k x _{3} (t) + b = k[x _{1} (t) + x _{2} (t)] + b
System does not satisfy the superposition property, so it is not linear
Linearization
• Real models usually exhibit nonideal and nonlinear
characteristics
• Analysis of a system represented by nonlinear partial
differential equations, with time varying coefficients is
extremely difficult and requires heavy computations
• There is no general analytic method available for solving
nonlinear systems
Dealing with Nonlinear Systems
In general, we can take one of the following three options:

a) Replace nonlinear elements with “roughly equivalent linear elements”, which often leads to invalid models;

b) Develop and solve a nonlinear model, which results in most accurate results, BUT the analysis is too expensive
since there is no general analytic methods available for
solving.
well, or at least fairly well, a nonlinear one) in order to
make possible more efficient analysis and control design
based on linear models.
Why are linear good?
To summarize: why are linear models useful?
• They are easy to compute, understand and visualize.
• They give predictable outputs, in time and over iterations
• The analysis of linear theory is complete, developed and efficient
• Linear differential equations are easy to solve!
What is a Linear Function?
• Linear functions are the first type of functions one learns in mathematics, yet there is not one single definition of linearity…
• Different answers apply to different contexts, discipline or purposes.
Possible Answers…
A linear function is:…

1. … a function of the general form y = bx + c

2. … a function whose derivative is a constant

3. … a function in which the output is proportional to the input

4. … a straight line? (careful, this only works in 2D representations)
Intuitively, linearity means proportionality of the output with
respect to a variable. One variable function are most
familiar but functions can be linear in many variables,
^{e}^{.}^{g}^{:}
y
n
= b x
1
1
+ b x
2
2
+ K + b x
n
n
Linear and Nonlinear Functions in 3D
The Real World
• Learning about linear behavior is good, but how useful is it? Is the real word linear at all? The answer is no most of the time.
• Unfortunately, nonlinear dynamics are not fully understood and the best we can do is simulate the real world with linear or low‐ order approximations.
• To be more precise, linear behavior is simulated locally, at a point or along a small interval in space‐time, and then the results are extrapolated about the general domain.
• That means that some degree of prediction is possible, but yet, we do not know everything about nonlinearity.
Small‐signal Linearization
Approximating the function while considering small disturbances around stable equilibrium points
Applications & Method
• Small signal linearization method is the most widely used
• It is in general done with the help of Taylor series.
• The Taylor expansion of a function f(x) around a point x
is given by:
f ( x )
=
f ( x )
+
)
2
2
df
(x −
x
d
f
(
x
−
x
)
( x )
+
( x )
2
dx
2 !
dx
+
...
(x x )
n
n
−
d
f
( x )
n
n!
dx
Linearization around a point
• If we define ˆx _{=} _{x} _{−} _{x}
the Taylor expansion becomes:
f ( x )
=
f ( x )
+ ˆx
+
2
n
x
ˆ 2
d
f
ˆ x
+ ...
2
2
!
dx
n!
x
• Note that the functions and all the derivatives are
evaluated at the linearization point
• If ˆx is small (i.e. x is close to x ), then we may drop the
second and higher‐order terms as follows:
f ( x ) ≈ f ( x ) + ˆx
Functions of several independent variables
• For a function f of a single variable, x:
f ( x ) − f ( x ) ≈ ˆx
• For a function f of two independent variables, x and y:
f ( x
,
y )
−
f ( x
,
y )
≈ ˆx
∂ f
∂ f
+ ˆy
∂ x
∂ y
x
, y
Around what point is it proper to linearize?
• Intuitively, one would say around zero
• But the general answer is around stable equilibrium points
y=sin x and y=x nearby zero are very close
Laplace Transform
The Laplace transform method substitutes relatively easily
solved algebraic equations for the more difficult
differential equations. The time‐response solution is
obtained by the following operations:

1. Obtain the linearized differential equations.

2. Obtain the Laplace transformation of the differential equations.

3. Solve the resulting algebraic equation for the transform of the variable of interest.
Laplace transform properties
Some commonly used Laplace transform pairs
Inverse Laplace transform
Summary of partial‐fraction technique
Inverse Laplace transform
Nth order real pole
X(s) = N(s)/(s+a) ^{N} = Σ A _{i} /(s+a) ^{i} where A _{i} = [1/(Ni)!][d ^{N}^{}^{i} /ds ^{N}^{}^{i} {(s+a) ^{N} X(s)}]
s = a
example: X(s) = 1/(s+2)(s+1) ^{3} = B/(s+2)+A _{1} /(s+1)+A _{2} /(s+1) ^{2} +A _{3} /(s+1) ^{3}
B = (s+2) X(s)
= 1
s = 2
A _{3} = (s+1) ^{3} X(s)
= 1
s = 1
A _{2} = d/ds {(s+1) ^{3} X(s)} = d/ds {1/(s+2)} = 1/(s+2) ^{2} = 1 A _{1} = [1/2!] d /ds {(s+1) X(s)} = 1
2
2
s = 1
2
s = 1
s = 1
s = 1
X(s) = 1/(s+2)+1 /(s+1) 1/(s+1) ^{2} +1/(s+1) ^{3} x(t) = [e ^{–}^{2}^{t} + e ^{–}^{t}  te ^{–}^{t} + (t ^{2} /2) e ^{–}^{t} ] u(t)
Solution of a differential equation
d
dt
dy ( t )
dt
+
3
( )
y t
=
2 ( )
r t
•
Where the initial conditions y(0)=1, dy(0)/dt=0, r(t)=1 t>=0.The Laplace
⎡
⎢
⎣
s
transform:
2
Y
(
s
)
−
sy
(
0 ) −
y
⎤
•
(
0 + 4
)
⎥
⎦
[
s
Y
(
s
)
−
y
(
0
)
]
+
(
s
)
=
2
R
( ),
s
R
(
s
)
=
1 /
s
Y ( s ) =
Y
(
s )
=
s + 4
2
+
s
2
+
4
⎡ 3
⎢
⎢
⎣
s
+ 1
(
s s
2
+ 4 s +
3
)
⎤
⎥
⎥
⎦
+
⎡
⎢
⎢
⎣
− 1
s
+ 1
+
⎤
⎥
s
+ 3 ⎥
⎦
+
s
y
( t ) =
⎡ 3
⎢
⎣
2
e
− t
−
e
3
t
⎤
⎥
⎡
⎢
+ −
⎣
1
e
− t
+
1
3
e
−
3
t
⎤
⎥
⎦
+
2
3
The steady‐state response is
lim ( ) =
y
t
t → ∞
2
3
.
The Transfer Function
• The transfer function of a linear, time‐invariant, differential equation system is defined as the ratio of the Laplace transform of the output (response function) to the Laplace transform of the input (driving function) under the assumption that all initial conditions are zero.
• TF description does not include any information concerning the internal structure of the system and its behavior.
Spring‐mass‐damper system:
M
d
y ( t )
dt
+
b
dy t
( )
dt
+
ky t
( )
=
( )
r t
Ms Y ( s ) + bY ( s ) + kY ( s ) = R ( s )
Transfer function:
Output
Input
= G ( s
)
Initial conditions are zero!!!!
Y
(
s
)
1
=
=
2
R s
(
)
Ms
+
bs
+
k
Consider the system represented by the following differential equation:
d
y
dt
+ q
d
y
dt
+
L
+
q
y
=
p
d
x
dt
+
p
d
x
dt
+
L
+
p
x
h
h
h
f
f
f h
l
W ere y(t) is t e response and r(t) is t e input or orcing unctions. I t e initia
conditions are all zero, then the TF is the coefficient of the R(s).
Y
(
s
)
=
G s R s
(
)
(
)
=
(
p s
)
(
q s
)
R s
(
)
=
p
s
+
p
s
+
L
+
p
s
^{n} +
q
s
+
L
+
q
R s
( ).
• Consider the linear time‐invariant system defined by the following differential equation:
where y is the output of the system and x is the input. The transfer function of this system is the ratio of the Laplace transformed output to the Laplace transformed input when all initial conditions are zero, or
By using the concept of transfer function, it is possible to represent system
dynamics by algebraic equations in s. If the highest power of s in the denominator of the transfer function is equal to n, the system is called an nth‐ order system.
Convolution Integral
For a linear, time‐invariant system the transfer function H(s) is
•
Y(s)=H(s)X(s)
where X(s) is the Laplace transform of the input and Y (s) is the Laplace transform of the output, where we assume that all initial conditions involved are zero. Multiplication in the complex domain is equivalent to convolution in the time domain so the inverse Laplace transform of Y(s) is given by the following convolution integral:
t
t
y(t) = ∫ x(τ) h(t  τ) d τ = ∫ h(τ) x(t  τ) d τ
0
0
y(t) = x(t)
*
h(t) = h(t)
*
x(t)
• Where both h(t) and g(t) are 0 for t<0.
Impulse Response of LTI System
δ(t)
h(t)
h(t) completely characterizes the LTI system in the time domain
∞
∞
y(t) = ∫ x(τ) h(t  τ) d τ = ∫ h(τ) x(t  τ) d τ
y(t) = x(t) * h(t) = h(t) * x(t)
∞
∞
Let x(t) = u(t) and h(t) = e ^{}^{a}^{t} u(t), a>0, Find the response (Step response!) y(t) = ∫ h(τ) x(t  τ) d τ = ∫ e ^{}^{a} ^{τ} u(τ) u(t τ)d τ
y(t) = ∫ e ^{}^{a} ^{τ} d τ = 1/a e ^{}^{a} ^{τ}
= 1/a (1  e ^{}^{a}^{t} ) u(t)
Modeling and Laplace Transforms