Professional Documents
Culture Documents
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
Physical
System
Modeling
Physical
Modeling
TU Delft
Control systems
Bond
Graph
Non Linear
State Space
Model
`Mechanical'
Analysis
http://blackboard.icto.tudelft.nl/courses/et3100/
Non Linear
State Space
Model
Linear
State Space
Model
In-Out
Model
TU Delft
Control systems
Linearisation
state space
to
in-out
Laplace Tr.
http://blackboard.icto.tudelft.nl/courses/et3100/
Linear
State Space
Model
In-Out
Model
Transfer
Functions
Contents
Modeling (2.1)-(2.5)
Linearisation (2.6)
Transfer Functions (3.1)
Block Diagrams (3.2)
Signal
ow graphs (3.2)
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
Modeling
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
..
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
10
Parts:
1 inductor (storage of
magnetic energy)
gen.kinetic energy =
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
11
Domain
Eort
Flow
Mech. Trans. force F
velocity v
Mech. Rot.
torque ang. vel. !
Elec.-Magn. voltage v
current i
Hydraulic
pressure p
ow rate Q
Thermic
temp. T entropy f. E_
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
12
Network structure
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
13
Bond graphs
Bond graphs is a language for physical modeling
which shows the interconnections of physical
elements and the energy exchanged between
them.
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
14
Bond
signs !
In general,
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
15
Linearisation
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
16
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
17
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
18
Linearisation
x_ = f (x; u)
y = h(x; u)
with one of the form
x_ = F x + Gu
y = Hx + Ju
using linearization or small signal analysis.
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
19
@f
)+
@x x ;u
0
TU Delft
Control systems
x
0
@f
+
u +
@u x ;u
o(x; u)
http://blackboard.icto.tudelft.nl/courses/et3100/
20
@f
:=
@x x ;u
@f
:=
@u x ;u
0
_ = F x + Gu +f (x ; u ) + o(x; u)
x
0 0
{z
}
|
Linear Part
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
21
_ = F x + Gu +f (x ; u ) + o(x; u)
x
0 0
{z
}
|
Linear Part
= 0
(x0; u0)
is an equilibrium
Control systems
_ = F x + Gu
x
http://blackboard.icto.tudelft.nl/courses/et3100/
22
Geometric idea of
_ = F x + Gu for dependence on x
x
f (x; u0)
x0
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
x
23
Geometric idea of
_ = F x + Gu for dependence on u
x
f (x0; u)
G
u0
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
u
24
Output equation
The same procedure can be used for the
function h(x; u):
y = h(x; u) = h(x0 + x; u0 + u) =
h(x0; u0
@h
)+
@x x ;u
0
TU Delft
Control systems
x
0
@h
+
u +
@u x ;u
o(x; u)
http://blackboard.icto.tudelft.nl/courses/et3100/
25
@h
:=
@x x ;u
0
@h
:=
@u x ;u
0
and
y := y
we have for the output equation:
y0 := h(x0; u0)
y0
y = Hx + J u
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
26
system
AROUND
x_ = F x + Gu
y = Hx + J u
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
27
S
-
u
+
u0
TU Delft
Control systems
x_ = f (x; u)
y = h(x; u)
y+
y0
x_ = F x + Gu
y = Hx + Ju
http://blackboard.icto.tudelft.nl/courses/et3100/
28
Example
m
r
Ft
Ft
x
TU Delft
Control systems
m1
m2
Fg
http://blackboard.icto.tudelft.nl/courses/et3100/
29
=
(m2l + m1r )
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
30
!
= f ((!; ); m) =
1
m l +m r
2
TU Delft
Control systems
f1((!; ); m)
f2((!; ); m)
http://blackboard.icto.tudelft.nl/courses/et3100/
31
0=
0=!
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
32
where
0 := arcsin(
TU Delft
Control systems
rm1g + 0
m2gl
http://blackboard.icto.tudelft.nl/courses/et3100/
33
= +
@f1 @f1
@! @
@f2 @f2
@! @
G=
!0;0;0)
@f1
@
@f2
@ (!0;0;0)
_
!
0
= 1
!
=
F
TU Delft
Control systems
where is a
0
m2gl cos 0
m2l2+m1r2
m2l2+m1r2
1
+ G
http://blackboard.icto.tudelft.nl/courses/et3100/
34
_!
!
=
F
+
G
is the dynamics around (0; ) for an input equal to
+ .
0
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
35
Conclusions on Linearization
_ = F x + Gu
x
y = Hx + Ju
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
36
_ = F x + Gu
x
y = Hx + Ju
where x is a SMALL displacement from x and u is
a varying part of the input which is supplied together
with the bias u .
0
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
37
x_ = F x + Gu
y = Hx + Ju
For what follows, it is handy to have a discription in
input-output dierential terms:
n
X
i=0
TU Delft
Control systems
di
ai i y(t) =
dt
m
X
j =0
dj
bj j u(t)
dt
http://blackboard.icto.tudelft.nl/courses/et3100/
38
a I +a F +a F
0
R nn
+ : : : an F n = 0
where
TU Delft
+ a + a + : : : + ann = det(I
1
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
F)
39
y = Hx + Ju
y_ = H x_ + J u_ = H (F x + Gu) + J u_ =
HF x + HGu + J u_
:::
y n = HF nx + f (u; u;
_ : : :)
(
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
40
yi
( )
= HF ix + : : :
a y + a y_ + : : : + any n =
H (|a I + a F{z+ : : : anF n}) x+
(
=0 for Cayley-Hamilton. !!
b u + b u_ + : : : + bnu n
(
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
41
di
ai i y(t) =
dt
TU Delft
Control systems
m
X
j =0
dj
bj j u(t)
dt
y
http://blackboard.icto.tudelft.nl/courses/et3100/
42
Laplace Transform
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
43
Solution
Problem
Starting
Problem
Time domain
s domain
Transf.
Problem
TU Delft
Control systems
Solution
Transf.
http://blackboard.icto.tudelft.nl/courses/et3100/
44
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
45
! F (s) :=
f (t)
f (t) :=
Note:
1
2j
+1
Z j
0
0 j 1
f (t)e
F (s)estds
st
dt
F (s)
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
46
Convolution:
L ff (t) f (t)g = F (s)F (s)
1
Linearity (superposition)
L ff (t) + f (t)g = F (s) + F (s)
1
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
47
Dierentiaton
L
Integration
L
L ftne atg =
TU Delft
Control systems
df
dt
Z t
0
n!
n+1
(s+a)
= sF (s) f (0 )
f ( ) d
1
= F (s)
s
(Fundamental!)
http://blackboard.icto.tudelft.nl/courses/et3100/
48
f (0
) = slim
sF
(
s
)
!1
lim
f
(
t
)
=
lim
sF
(
s
)
t!1
s!
0
Time delay
L ff (t )g = L ff (t)g e
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
49
(t)
1(t)
s+a
s
am
..
( + )
TU Delft
Control systems
at
tm e
1
1)!
at
..
http://blackboard.icto.tudelft.nl/courses/et3100/
50
F (s)
s+a
2 2
(s+a) +b
at
cos bt
b
2 2
(s+a) +b
at
sin bt
sin at
cos at
s2+a2
s2+a2
..
TU Delft
Control systems
f (t) t 0
..
http://blackboard.icto.tudelft.nl/courses/et3100/
51
bmsm + bm sm + : : : + b
G(s) = n
=
n
s + an s + : : : + a
C
s p
1
we get:
Cn
+ : : : + s pn
Cn
L fG(s)g = L s p + : : : + L s p
n
which we know by the previous table. Review Partial
fraction Expansion (pag.98).
1
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
52
Convolution Integral
y(t) =
Z 1
g(t )u( ) d
Z 1
g( )u(t ) d
y(t) =
TU Delft
Control systems
Z 1
g( )(t ) d
= g(t)
http://blackboard.icto.tudelft.nl/courses/et3100/
53
Consider
Z 1
g(t ) u( ) d
Then
Y (s)
or
TU Delft
y(t) = L
Control systems
fL fg(t)g L fu(t)gg
http://blackboard.icto.tudelft.nl/courses/et3100/
54
Transfer Function
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
55
For
1<t<1
y =gu
and using the convolution result of the Laplace transform:
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
56
U (s)
TU Delft
Control systems
G(s)
Y (s)
http://blackboard.icto.tudelft.nl/courses/et3100/
57
For t > 0
Furthermore, if we consider the linear time invariant
system in the dierential form:
n
X
i=0
di
ai i y(t) =
dt
m
X
j =0
dj
bj j u(t)
dt
aisiY (s)
TU Delft
Control systems
n
X
i=1
ai
i 1
X
j =0
sj y i
(
1)
(0 ) =
http://blackboard.icto.tudelft.nl/courses/et3100/
m
X
j =0
bj sj U (s)
58
n
X
i=0
aisiY (s) =
gives
m
X
j =0
bj sj U (s) +
n
X
i=1
ai
i 1
X
j =0
sj y i
(
1)
(0 )
where
b0 + b1s + : : : + bms
G(s) =
a0 + a1s + : : : + ansn
m
en
Y0(s) =
n
i=1 i
i 1 j (i j 1)
(0
j =0
n
+ n
1 +
sy
a0 + a s : : : a s
Note:
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
59
Y0(s)
If y (0
then we have:
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
60
Zeros, poles
=1
where
zi roots of b
G(s)
zi )
pj )
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
61
DC gain
For a stable system, the nal value we get after a unitary
input step is called the DC gain.
Using Final Value Theorem (pag.103)
lim
y
(
t
)
=
lim
sY
(
s
)
=
t!1
s!
1
lim
sG
(
s
)
U
(
s
)
=
lim
sG
(
s
)
=
s!
s!
s
b
lim
G
(
s
)
=
s!
a
0
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
62
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
63
U(s)
S
+
Y(s)
R(s)
U(s)
TU Delft
Control systems
G(s)
Y(s)
U(s)
http://blackboard.icto.tudelft.nl/courses/et3100/
U(s)
U(s)
U(s)
64
Z(s)
-1
Y(s)
1
R(s)
G(s)
U(s)
TU Delft
Control systems
U(s)
Y(s)
http://blackboard.icto.tudelft.nl/courses/et3100/
65
G1(s)
Z(s)
Y(s)
G2(s)
G1(s)
U(s)
G2(s)
Z(s)
Y(s)
G(s)
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
66
Parallel connection
G1(s)
U(s)
Y(s)
G2(s)
G1(s)
Y(s)
U(s)
G2(s)
G(s)
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
67
Feedback connection
R(s)
E(s)
G1(s)
Y(s)
G2(s)
R(s)
E(s)
G1(s)
Y(s)
- G2(s)
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
68
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
69
Y (s) 1 X
=
G(s) =
Gi(s)i
U (s) i
Gi(s)= Path gain of the i th forward path
P
=
The system
determinant=
1
(all
individual
P
loop gains)+
(gain products
of all possible two
P
loops that do not touch)
(gain products of all
possible three loops that do not touch)+...
i= i
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
70
Example
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
71
Various Paths
Forward paths
1236
12346
123456
G
G
G
1
2
3
Loops
=1
=1
=1
232
2342
23452
TU Delft
l
l
l
1
2
3
=
=
=
Control systems
a =s
a =s
a =s
1
( )(1)
( )(1)
(b )(1)
Determinants
= 1 as1
=1 0
=1 0
=1 0
s b1
1
1
s s b2
1
1
1
s s s
a2
s2
a3
s3
+0
1
2
3
http://blackboard.icto.tudelft.nl/courses/et3100/
72
Y (s)
U (s)
which is equal to
Y (s)
U (s)
TU Delft
Control systems
b1s2 + b2s + b3
s3 + a1s2 + a2s + a3
http://blackboard.icto.tudelft.nl/courses/et3100/
73
Conclusions
For
We
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
74
A
The
linear systems
TU Delft
Control systems
http://blackboard.icto.tudelft.nl/courses/et3100/
75