Dr Ya-Jun Pan Pan alsent today her PHD student Lucas led this class

.
was
,
Mech 3900
Wednesday November 6th ,

Handout – Ya-Jun Pan Chapter 8 - Page 1

Poles : values of S that

cause 5-00 MECH 3900 Systems I

Dept. of Mechanical Engineering

Dalhousie University

Chapter 8: Transfer Function

8.1 Transfer Function (Sections 2.5 and 5.1 in the textbook by W. Palm)

A transfer function is to describe the relationship of the system from the input to the output.

U(s) Y(s)
G(s)

Y (s) Numerator polynomialins h(s -

2,
) ...
(5 -

2 m)
G(s) = = =

U(s) Denominator polynomialins (S -

P, ) ... (S Pn) -

The roots of the numerator are called zeros; the roots of the denominator are called poles.
The denominator is called the characteristic equation of the system. 23wns Wa
152 ,
+

The roots of the characteristic equation are poles of the Transfer Function. O =
(S Si)(5 Sa)
-
-

8.2 Relationship to State-Space Representation (Section 12.7 in the textbook textbook by R&W)
Take Laplace Transform (assume zero initial conditions (IC’s)) of the following SS form:
! x = Ax + Bu
"
# y = Cx + Du
We can get the transfer function represented as:

v
Y (s) −1 is
G(s) =
U(s)
= C ( SI − A) B + D our
goal
To demonstrate the derivation in the class:

1
note'
not matrix is it must be
I =
[ ] ,
S is a ,
will
multiplied by I the I matrix always be
equal
to one
 transform I
take laplace of and
y

①
Bu()
2(f) 2(Ax Bu)
+ = SX(s) = AX(s) +

Sy (s) -

Ax(s) = Bu(s)

(SI
-

A)"(SI A) -
= (s) =
Bu(s)(SI A) -

multiply by inverse

[i] (SI A) bu(s)

x(s)
-
=

Y =
(x + umY(s) = (x(s) + Du(s)8
take L

[(SFA) Bucs)] DuCs)

+
C
: Y() =
+

=
[c(Sy A)
-
B +
0]u(s)

G = (SFA
:
Handout – Ya-Jun Pan Chapter 8 - Page 2

Example: For the mass-spring-damper system, we have derived the state space equation as:

! $ ! B 1 $! $ ! 1 $
d # vm &=# − m − &# vm &+# m &F
m
dt #" Fk &% # &# F &% # &
#" k 0 &%" k #" 0 &%

Find the single ODE from mass displacement, i.e. y = xm . Use the transfer function expression as

Y (s)
= C(sI − A)−1 B + D .
U(s)
Solution:

Three representations of the system dynamics:

Example on the Mass-damper-spring system:
1. Three representations: System Representation in Time Domain- ODE:
m
x + bx + kx = F
 =0:
Transfer Function with zero initial condition: x(0) = x(0)

ms 2 X(s) + bsX(s) + kX(s) = F(s)

1
X(s) 1 m
G(s) = = 2 =
F(s) ms + bs + k s 2 + s + k
b
m m
State-space Form:

x1 = x2 " ' ' 1 *' x * ' 0 *

$ x1 * ) 0 , 1 , ) ,
b k 1 #⇒) ,=
) k b ,) + ) 1 , F(t)
x2 = − x2 − x1 + F $ )( x2 ,+ − − ) x ,
m m m % )( m m ,+( 2 + )( m ,+
  
A B

! x #
For the output: ! #
y = x1 = " 1 0 $%
1
& + [ 0 ] F(t)
  " 2 &$ 
% x D
C

2
Handout – Ya-Jun Pan Chapter 8 - Page 3

2. Simulation with m = 5; b = 1; k = 2 => 1/m=0.2, b/m=0.2, k/m = 0.4.

b k 1
ODE: 
x=− x − x + F à Block diagram as we have done in the labs.
m m m

State-Space Representation: A = [0 1; -0.4 -0.2], B = [0;0.2], C = [1 0], D = 0.

0.2
Transfer Function in Matlab: G( s) = .
s + 0.2 s + 0.4
2

Simulink Diagram:

ODE

X initial

valusa

~
no initial

velocity
·
present to O

(with initial values

Transfer Function Block:

3
mass
spring damper system example

? A X * U

= xm
y (x Bu
Fr
+
=
kym =

y
2
-
=

+ [037
Y
= 20 1) .

adiCS On

.
u(s)

= [m Ym] a]
= 20 ]. [*]
all 922-Q1292l

[0][ -
·

20 (
Ym[*]
+
=

im 7
transfora
4m
5 +ms +

So +
Tms + 4m

Y(s) (s2 +
%ms +
Ym) = YmU()

take inverse laplace Y

+ Ymy =

Ym]00z
Handout – Ya-Jun Pan Chapter 8 - Page 4

State-Space Representation Block:

Step Response: (three curves are identical)

of motion
equation
52 +
0 25.
+ 0 4 .

3-O = As

=
b

anti

Wh =2 = 0 632
.

=
F/m
SS-value = 0 S
.

Question: What are the damping ratio, undamped natural frequency, and steady state value?

4
Handout – Ya-Jun Pan Chapter 8 - Page 5

8.3 Pole-Zero Plot F(t) = 0

Consider the unforced response of a linear SISO (Single-Input-Single-Output) system

with a set of initial conditions (IC’s):
+ 0
↑ (a) n
yh (t) = ∑ ci e pit
i=1

where
ci : constantsdetermined from givenset of IC's
pi : exponent are roots of characteristic equation (n poles)
For the ith pole :
pi = - +
ju
Yi(t) A
=
*
sin(w+ + b)
1. pi = −σ in the left hand side of the s-plane: real pole.
where AI and

Exponentially decaying component: ce −σ t

in homogeneous response.
are
determined
by initial
2. pi = 0 (at origin) conditions
Component constant in amplitude defined by IC’s.

3. pi = σ in the right hand side of the s-plane: real pole.

Exponentially increasing component: ceσ t in homogeneous response (unstable).

4. pi = −σ ± jω in the left hand side of the s-plane: complex conjugate pair.

Combine to generate a response component that is decaying sinusoid of the form as:
Ae−σ t sin(ω t + Φ)
where A and Φ are determined by IC’s.
Note: The rate of decay is specified by σ ;
The frequency of oscillations is specified by ω .

5. pi = ± jω : imaginary pole pair.

Oscillatory component in constant amplitude.

5
 Overview of the DifferentCases

case a

Im

& Re

+
(t)

-
t

Case 2

x(t)

I
Im

↓
-

* Re

case
+
(t)
Im

- Re
-
Case &
y(t)
Im

- - .
ju
I

,
In
I

Re
I

=....
-ju
case S
y(t)
Im

&

Wit
&

case C
+ (t)
Im

&

·
N +

-
 Handout – Ya-Jun Pan Chapter 8 - Page 6

6. pi = σ ± jω : complex pole pair in the right hand side of the s-plane.

Exponentially increasing oscillatory component.

C
A

O
-o + ju/imaginary
(
real

112
poles
? 1) ;
- -

Zeros : -2

GCs) (transfer function) from pole-zero plot

Determine

b(s) = k(s +
2) =
k(s +
2)
(S + 1)((S + 1) 4)
=

2j)(S
1
2j) +
-
+

(S r)(s +
+
+ note !
-
- a je j
=

j = Fi

s + 5
6
I could be determined if other conditions
au
given
 Handout – Ya-Jun Pan Chapter 8 - Page 7

Aerotsin (wt b)
+

u
41 Ansin (w 4) ++

Vy(t) -
t

Ce-ot Cert

ju
⑦
-

O +
: constant

Suntjens F
I =
En (wi

within-wa constant
Wh
②
=

-Funn
circiature

③ 30 From
will
polo
approach
the

ax
-
imaginary
7 4
① just-jun
⑤ RHS : unstable
 Determine response from plot (4th order)
pole-zero

X - - -

2j
&
3 - 0 . 1+
-

(e Ce
y(t)
= +

- A &
Re
3
·

+ Ae sin(2+ +
b)
E
= .. 2)
,
=

C, A
↓, determined by initial conditions
Handout – Ya-Jun Pan Chapter 8 - Page 8

Fig. 8.3.6 from the textbook by [Link].

from A-6 : 3d I had e

·
move overshoot
· Faster oscillations

From A- : Wh ,
34 ,
was

overshoot
less
of oscillation
· same
period
from A-B-C :
und ,5
· the same overshoot

8
Handout – Ya-Jun Pan Chapter 8 - Page 9

As shown in the pole-zero plot above, analyze the influence of the poles on the system response.
Discussions:
1). p1 = −6 :

2). p1 = −0.2 :

3). p1 = −3± j :

8.4 Coordinate Transformations (Section 10.4.7 in the textbook by R&W)

Suppose we make the coordinate transformation:
x(t) = Tz(t), T ∈ R n×n
where T is a nonsingular constant matrix and det(T ) ≠ 0 . Note that T is a coordinate
transformation as being a rotation of the coordinate axes with respect to which everything is
being measured. Let’s look at the system:
! x = Ax + Bu
" with x(t) = Tz(t)
# y = Cx + Du
Derivation to the new set of state-space equations: (to be done in the class).
NOTE: the new set of state-space equations z(t) = T −1x(t) written in terms of rather than x(t) .
Both representations have the same G(s).

9
 Lucas started by providing a brief overview

representing real world

systems with mechanical models
·

?
why
analize and simulate
· to
design ,

we to conversions between 3 Forms

State-Space Form ODE

S S
define x1 =
y a
+
2
=
4 Inverse
G(s) = CCSI A) D - + N Laplace
Transform Laplace
Transform
Transfer Function (
[

Example

Car suspension
+(t)
a mi(H) bj(+) hy(t) +

-
+
u(t) =

t
-o
- y(t)a
car over a
bump
going

& Mn(t) Sus ,

Three reasons why we use transfer functions

# ↓fficult to work with CDEs in sories

y(t)
'unedt
u(t

3
2) it makes it easy to
analyze a
system

Stable ?

poles

3) we can
design Feedback controllere)