0% found this document useful (0 votes)
19 views

EE3002 Part2

This document discusses feedback control systems and provides examples of deriving dynamic models for various systems using physics-based approaches. It introduces concepts like transfer functions and linearization. Examples include models for cruise control, car suspension, and pendulum systems. The goal is to analyze plant models, design dynamic models for specific systems, and simulate them in MATLAB and Simulink.

Uploaded by

Mustafa
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
19 views

EE3002 Part2

This document discusses feedback control systems and provides examples of deriving dynamic models for various systems using physics-based approaches. It introduces concepts like transfer functions and linearization. Examples include models for cruise control, car suspension, and pendulum systems. The goal is to analyze plant models, design dynamic models for specific systems, and simulate them in MATLAB and Simulink.

Uploaded by

Mustafa
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 86

Feedback Control Systems

Fall 2020

September 9, 2020
Today

Disturbances

r e u y
Controller System

ym

Measurements

Objective
I Analyze plant models
I Design dynamic models for particular systems
I Simulate in MATLAB and Simulink
Introduction

I The ultimate goal in feedback control design is to drive the plant


output to follow a desired reference variable
Introduction

I The ultimate goal in feedback control design is to drive the plant


output to follow a desired reference variable
I The first step in this process is to derive the plant dynamic model
Introduction

I The ultimate goal in feedback control design is to drive the plant


output to follow a desired reference variable
I The first step in this process is to derive the plant dynamic model
I Two approaches: Physics-based and system identification
Dynamics of Mechanical Systems

Physics-based derivation
I Newton’s Law of Motion:

F = ma

F : Sum of all forces applied to each body in a system


a: The vector of acceleration of each body w.r.t. a reference
frame
m: Mass of the body
I For instance: Weight of a body: mg, (g : gravitational
acceleration)
Example: Cruise Control Model

Derive the equations of motion. In other words, find the relation


between the input u and the output y
I u: Force enacted by engine
I b: Friction affecting the car’s motion in the opposite direction of
the car’s velocity
I x: Coordinates of the car’s body
I ẋ: Car’s speed
I ẍ: Car’s acceleration
Example: Cruise Control Model

Total force equals the total mass times acceleration:

u − fb ẋ = mẍ

Equivalently:
fb u
ẍ + ẋ =
m m
Our variable of interest is the velocity v. Rewriting yields:
fb u
v̇ + v=
m m
Example: Cruise Control Model

Although frequency response will be covered later, here let’s assume a


solution of the form:
v = V0 est
Time derivative is given as:
v̇ ≡ sV0 est
Rearrange:
 
fb 1
s+ V0 est = U0 est
m m
Thus, we have
1
V0
= m fb
U0 s+ m
Example: Cruise Control Model

We will use the following form to denote the fact that the variables
are transformed to another domain:
1
V (s)
= m fb
U (s) s+ m

I We will name this expression the transfer function of the vehicle.


I The term s denotes the time derivative d/dt.
Example: Cruise Control Model

1
V (s)
= m fb
U (s) s+ m

Time response of the system can be simulated on MATLAB (both on


script file and Simulink):
Example: Car Suspension System

I Also known as quarter car system. Consists of a


mass-spring-damper model
Example: Car Suspension System

I Also known as quarter car system. Consists of a


mass-spring-damper model
I Our aim is to find the relation between the input u and the
output y in the simplified model
Example: Car Suspension System

I Also known as quarter car system. Consists of a


mass-spring-damper model
I Our aim is to find the relation between the input u and the
output y in the simplified model
I x, y: Displacements of the two masses from their equilibrium
positions
I b: Damper coefficient (probably hydraulic)
I ks : Spring constant
Example: Car Suspension System

Focus on the free-body diagram of the system


I Spring and damper act on the two masses with same amount of
force but in the opposite directions

Fs = ks (y − x)
Fb = b(ẏ − ẋ)
Example: Car Suspension System

Focus on the free-body diagram of the system


I Spring and damper act on the two masses with same amount of
force but in the opposite directions

Fs = ks (y − x)
Fb = b(ẏ − ẋ)

I The lower spring denotes the tire compressibility. It softens the


affects of road bumps which changes the road level reference r:

Fw = kw (x − r)
Example: Car Suspension System

Focus on the free-body diagram of the system


I Spring and damper act on the two masses with same amount of
force but in the opposite directions

Fs = ks (y − x)
Fb = b(ẏ − ẋ)

I The lower spring denotes the tire compressibility. It softens the


affects of road bumps which changes the road level reference r:

Fw = kw (x − r)

I Gravity effect can be omitted


Example: Car Suspension System

Overall system equations:

b(ẏ − ẋ) + ks (y − x) − kw (x − r) = m1 ẍ
−ks (y − x) − b(ẏ − ẋ) = m2 ÿ

It is useful to leave the terms with highest order derivatives alone:


b ks kw kw
ẍ + (ẋ − ẏ) + (x − y) + x= r
m1 m1 m1 m1
b ks
ÿ + (ẏ − ẋ) + (y − x) = 0
m2 m2
Example: Car Suspension System

We can obtain the transfer function in the same fashion:


Example: Car Suspension System

We can obtain the transfer function in the same fashion:


I

b ks kw kw
s2 X(s) + s (X(s) − Y (s)) + (X(s) − Y (s)) + X(s) = R(s)
m1 m1 m1 m1
b ks
s2 Y (s) + s (Y (s) − X(s)) + (Y (s) − X(s)) = 0
m2 m2
Example: Car Suspension System

We can obtain the transfer function in the same fashion:


I

b ks kw kw
s2 X(s) + s (X(s) − Y (s)) + (X(s) − Y (s)) + X(s) = R(s)
m1 m1 m1 m1
b ks
s2 Y (s) + s (Y (s) − X(s)) + (Y (s) − X(s)) = 0
m2 m2
I Rearranging yields:

Y (s)
=?
R(s)
Example: Pendulum

Equations of motion
Example: Pendulum

Equations of motion
I Torque: Tc
I Mass concentrated at the tip m
I Gravitational acceleration: g
I Output: The angle θ
Example: Pendulum

Equations of motion
I Torque: Tc
I Mass concentrated at the tip m
I Gravitational acceleration: g
I Output: The angle θ
I Applied torque is equal to the sum of moments at the tip:

Tc − mgl sin(θ) = I θ̈
g Tc
θ̈ + sin(θ) =
l ml2
Example: Pendulum

I Nonlinear due to sin(θ)

g Tc
θ̈ + sin(θ) =
l ml2
Example: Pendulum

I Nonlinear due to sin(θ)


I Can linearize around the equilibrium
θ = 0 such that sin(θ) u θ g Tc
θ̈ + sin(θ) =
l ml2
I Dynamics becomes
g Tc
θ̈ + θ =
l ml2
Example: Pendulum

I Nonlinear due to sin(θ)


I Can linearize around the equilibrium
θ = 0 such that sin(θ) u θ g Tc
θ̈ + sin(θ) =
l ml2
I Dynamics becomes
g Tc
θ̈ + θ =
l ml2
I Apply the transformation s instead
of d/dt:
g 1
s2 Θ(s) + Θ(s) = Tc (s)
l ml2
1
Θ(s) 2
= 2ml g
Tc (s) s + l
Example: Pendulum

I Nonlinear due to sin(θ)


I Can linearize around the equilibrium
θ = 0 such that sin(θ) u θ g Tc
θ̈ + sin(θ) =
l ml2
I Dynamics becomes
g Tc
θ̈ + θ =
l ml2
I Apply the transformation s instead
of d/dt:
g 1
s2 Θ(s) + Θ(s) = Tc (s)
l ml2
1
Θ(s) 2
= 2ml g
Tc (s) s + l

I Shows oscillatory (periodic) behavior


Exercise: Simulation
I Simulate the behavior of the output θ for step input on MATLAB
Exercise: Simulation
I Simulate the behavior of the output θ for step input on MATLAB
I m = 1 kg, l = 1 m, g = 9.81 m/sec2 . System becomes:

θ̈ = −9.81θ + 1

I Simulate for 10 seconds


Exercise: Simulation
I Simulate the behavior of the output θ for step input on MATLAB
I m = 1 kg, l = 1 m, g = 9.81 m/sec2 . System becomes:

θ̈ = −9.81θ + 1

I Simulate for 10 seconds

I
Exercise: Simulation
I Simulate the behavior of the output θ for step input on MATLAB
I m = 1 kg, l = 1 m, g = 9.81 m/sec2 . System becomes:

θ̈ = −9.81θ + 1

I Simulate for 10 seconds

I Linearization yields sufficiently accurate results


Exercise: Simulation
I Repeat the same simulation in Simulink in continuous time

θ̈ = −9.81θ + 1
Exercise: Simulation
I Repeat the same simulation in Simulink in continuous time

θ̈ = −9.81θ + 1

I Block 1/s stands for integrator


Exercise: Simulation
I Repeat the same simulation in Simulink in continuous time

θ̈ = −9.81θ + 1

I Block 1/s stands for integrator


I We can simulate the original nonlinear system as well
Example: Satellite Attitude Model

I Motion executed in three axes


(x, y, z)
Example: Satellite Attitude Model

I Motion executed in three axes


(x, y, z)
I Derive the equations of motion in
each axis
Example: Satellite Attitude Model

I Motion executed in three axes


(x, y, z)
I Derive the equations of motion in
each axis
I Forces and moments:
Fc : Satellite jets produces force with
air-breath (measured in the axis
perpendicular to the paper)
d: Distance between the force point
and the c.o.m.
Md : Disturbance moment caused by
the solar panels
θ: Orientation caused by the moment
Fc d (output)
I: Inertia of the satellite in the
corresponding axis
Example: Satellite Attitude Model

I Dynamics become:

Fc d + Md = I θ̈

Named double integrator plant


Example: Satellite Attitude Model

I Dynamics become:

Fc d + Md = I θ̈

Named double integrator plant


I Transfer function:
Θ(s) 1 1
=
U (s) I s2

where U = Fc d + Md is the input to


the system.
Exercise: Simulation

I Design the model on Simulink

Fc d + Md = I θ̈
Θ(s) 1 1
=
U (s) I s2
Exercise: Simulation

I Design the model on Simulink

Fc d + Md = I θ̈
Θ(s) 1 1
=
U (s) I s2
I Simulate for 10 seconds
Exercise: Simulation

I Design the model on Simulink

Fc d + Md = I θ̈
Θ(s) 1 1
=
U (s) I s2
I Simulate for 10 seconds

I
Exercise: Simulation

I Design the model on Simulink

Fc d + Md = I θ̈
Θ(s) 1 1
=
U (s) I s2
I Simulate for 10 seconds

I Two integrators in series


Inserting Flexibility into the Model: Disk Drive

I Information on the disk is


read by the optical head
Inserting Flexibility into the Model: Disk Drive

I Information on the disk is


read by the optical head
I There is a sort of flexibility
between the read head and
the drive motor
Inserting Flexibility into the Model: Disk Drive

I Information on the disk is


read by the optical head
I There is a sort of flexibility
between the read head and
the drive motor
I Let’s derive the equations of
motion. Recall the quarter car
model (two-mass system). We
have a similar set of forces
and moments here
Inserting Flexibility into the Model: Disk Drive

I Equations of motion on the motor side

I1 θ̈1 + b(θ̇1 − θ̇2 ) + k(θ1 − θ2 ) = Mc + Md


Inserting Flexibility into the Model: Disk Drive

I Equations of motion on the motor side

I1 θ̈1 + b(θ̇1 − θ̇2 ) + k(θ1 − θ2 ) = Mc + Md

I Equations of motion on the head side

I2 θ̈2 + b(θ̇2 − θ̇1 ) + k(θ2 − θ1 ) = 0


Inserting Flexibility into the Model: Disk Drive
Equations of motion:
I1 θ̈1 + b(θ̇1 − θ̇2 ) + k(θ1 − θ2 ) = Mc + Md
I2 θ̈2 + b(θ̇2 − θ̇1 ) + k(θ2 − θ1 ) = 0

I For simplicity, can ignore Md and b. TF from the torque to the


read head:
Θ2 (s) k
=  
Mc (s) 2
I1 I2 s + k
+ k
I1 I2
Inserting Flexibility into the Model: Disk Drive
Equations of motion:
I1 θ̈1 + b(θ̇1 − θ̇2 ) + k(θ1 − θ2 ) = Mc + Md
I2 θ̈2 + b(θ̇2 − θ̇1 ) + k(θ2 − θ1 ) = 0

I For simplicity, can ignore Md and b. TF from the torque to the


read head:
Θ2 (s) k
=  
Mc (s) 2
I1 I2 s + k
+ k
I1 I2

I With the same assumption, TF from the applied torque to the


motor to the motion output θ1 :
Θ1 (s) I s2 + k
= 2 
Mc (s) I1 I2 s2 + Ik1 + k
I2
Inserting Flexibility into the Model: Disk Drive
Equations of motion:
I1 θ̈1 + b(θ̇1 − θ̇2 ) + k(θ1 − θ2 ) = Mc + Md
I2 θ̈2 + b(θ̇2 − θ̇1 ) + k(θ2 − θ1 ) = 0

I For simplicity, can ignore Md and b. TF from the torque to the


read head:
Θ2 (s) k
=  
Mc (s) 2
I1 I2 s + k
+ k
I1 I2

I With the same assumption, TF from the applied torque to the


motor to the motion output θ1 :
Θ1 (s) I s2 + k
= 2 
Mc (s) I1 I2 s2 + Ik1 + k
I2

I Later on, we will see that designing a controller for a flexible


system needs more care than the designs for a non-flexible
system.
Bridged Tee Circuit

I Includes only passive circuit


elements
Bridged Tee Circuit

I Includes only passive circuit


elements
I Derive the equations using the
Kirschoff’s laws and node
analysis
Bridged Tee Circuit

I Includes only passive circuit


elements
I Derive the equations using the
Kirschoff’s laws and node
analysis
I Take node 4 as the ground
node: v4 = 0.
Bridged Tee Circuit

I v4 = 0 → v1 = vi
Bridged Tee Circuit

I v4 = 0 → v1 = vi
I v2 = v1 − R1 i1 , etc.
Bridged Tee Circuit

I v4 = 0 → v1 = vi
I v2 = v1 − R1 i1 , etc.
I Node 2:
v1 − v2 v2 − v3 dv2
− + + C1 =0
R1 R3 dt
Node 3:
v3 − v2 d(v3 − v1 )
+ C2 =0
R2 dt
Bridged Tee Circuit

I v4 = 0 → v1 = vi
I v2 = v1 − R1 i1 , etc.
I Node 2:
v1 − v2 v2 − v3 dv2
− + + C1 =0
R1 R3 dt
Node 3:
v3 − v2 d(v3 − v1 )
+ C2 =0
R2 dt
I HW: Find the relation
between the input vi and
output v3
Opamp Circuit Model
I Assume an ideal opamp
Opamp Circuit Model
I Assume an ideal opamp
I For the opamp circuit shown,
derive the equations of motion
Opamp Circuit Model
I Assume an ideal opamp
I For the opamp circuit shown,
derive the equations of motion
I Positive node connected to
ground: v− = 0
I Thus,
v1
i1 =
R1
v2
i2 =
R2
vo
io =
Rf
I Note that i1 + i2 + i0 = 0. So
Rf Rf
vo = − v1 − v2
R1 R2
Opamp Integrator Model

I Ideal opamp
Opamp Integrator Model

I Ideal opamp
I For the opamp circuit shown,
derive the equations of motion
Opamp Integrator Model

I Ideal opamp
I For the opamp circuit shown,
derive the equations of motion
I iin + iout = 0. Thus
vin dvout
+C =0
Rin dt
Opamp Integrator Model

I Ideal opamp
I For the opamp circuit shown,
derive the equations of motion
I iin + iout = 0. Thus
vin dvout
+C =0
Rin dt
I Again, using the derivative
d
operator (s ≡ ):
dt
Vo (s) 1
=−
Vin (s) Rin Cs
DC Motor Model

I The stator, non-rotating part, has


magnets which causes a magnetic
field
DC Motor Model

I The stator, non-rotating part, has


magnets which causes a magnetic
field
I The input voltage is converted to a
torque
DC Motor Model

I The stator, non-rotating part, has


magnets which causes a magnetic
field
I The input voltage is converted to a
torque
I In a brushed DC motor, the brushes
are in contact with the commutator
part to generate maximum torque
DC Motor Model

I The stator, non-rotating part, has


magnets which causes a magnetic
field
I The input voltage is converted to a
torque
I In a brushed DC motor, the brushes
are in contact with the commutator
part to generate maximum torque
I Derive the equations of motion ia :
The armature current
va : Input voltage
e: Back emf voltage
b: Viscous friction
θm : Rotor rotation angle (output)
θ̇m : Rotor angular velocity
T : Torque
DC Motor Model

I The circuit model:


dia
La + Ra ia = va − Ke θ̇m
dt
DC Motor Model

I The circuit model:


dia
La + Ra ia = va − Ke θ̇m
dt
I On the motor, there are two applied torques: T and bθ̇m :

Jm θ̈m + bθ̇m = Kt ia
DC Motor Model

I The circuit model:


dia
La + Ra ia = va − Ke θ̇m
dt
I On the motor, there are two applied torques: T and bθ̇m :

Jm θ̈m + bθ̇m = Kt ia
I Combining yields
Θm Kt
=
Va (s) s ((Jm s + b)(La s + Ra ) + Kt Ke )
DC Motor Model

I Usually, the inductance effect is ignored (La ≡ 0):


 
Kt Ke Kt
Jm θ̈m + b + θ̇m = va
Ra Ra
DC Motor Model

I Usually, the inductance effect is ignored (La ≡ 0):


 
Kt Ke Kt
Jm θ̈m + b + θ̇m = va
Ra Ra
I Thus, the resulting motor model is found as
Kt
Θm
=  Ra 
Va (s) Jm s2 + b + Kt Ke
s
Ra
Exercise - Gear Model

Generally, motors are used in


combination with gears. Derive
the model for the DC gear motors.
Advanced Models
I Thus far we have seen simple
models for basic systems
Advanced Models
I Thus far we have seen simple
models for basic systems
I Most mechatronic systems
comprise more complicated
motion models
Advanced Models
I Thus far we have seen simple
models for basic systems
I Most mechatronic systems
comprise more complicated
motion models
I Advanced systems are usually
modeled with nonlinear
models and include friction,
backlash, saturation, delay,
etc.
Advanced Models
I Thus far we have seen simple
models for basic systems
I Most mechatronic systems
comprise more complicated
motion models
I Advanced systems are usually
modeled with nonlinear
models and include friction,
backlash, saturation, delay,
etc.
I There is a trade-off: As the
system model gets more
complicated, controller design
becomes more difficult
Advanced Models
I Thus far we have seen simple
models for basic systems
I Most mechatronic systems
comprise more complicated
motion models
I Advanced systems are usually
modeled with nonlinear
models and include friction,
backlash, saturation, delay,
etc.
I There is a trade-off: As the
system model gets more
complicated, controller design
becomes more difficult
I Therefore, simplifications such
as linearization is commonly
applied to ease the controller
design process
Conclusion

I We have seen how to model basic systems with differential


equations in time domain
Conclusion

I We have seen how to model basic systems with differential


equations in time domain
I Converted the resulting models into frequency domain
Conclusion

I We have seen how to model basic systems with differential


equations in time domain
I Converted the resulting models into frequency domain
I Simulated the step response of the models
Conclusion

I We have seen how to model basic systems with differential


equations in time domain
I Converted the resulting models into frequency domain
I Simulated the step response of the models
I Simplifications can be done for nonlinear advanced systems
Conclusion

I We have seen how to model basic systems with differential


equations in time domain
I Converted the resulting models into frequency domain
I Simulated the step response of the models
I Simplifications can be done for nonlinear advanced systems
I Reading assignment: Chp. 2

You might also like