Professional Documents
Culture Documents
Gi-Woo Kim
Fall 2022
Lecture 2
1
Today’s Course Outline
2
Review : General Control System
Sensor
Noise
Closed-loop (negative feedback) control system Automatic Control
3
• Examples of Automotive Control Systems
a. Automobile steering
control system.
b. The driver uses the
difference between
the actual and the
desired direction of
travel to generate a
controlled adjustment
of the steering wheel.
c. Typical direction-of-
travel response.
4
Six Step Approach to Dynamic System Problems
5
Transfer Function in Mechanical Systems
• Example 1: 1990 ACC Benchmarking Problem
• 2DOF
sensing • Semi-definite system
actuating • ω1=0 (rigid body)
d m1 x1 = u − k ( x1 − x2 )
m2 x2 = −d + k ( x1 − x2 )
x1 ( s ) (m1s 2 + k ) x2 ( s ) k
= = 2 2 = ,d 0 = ,d 0
u ( s ) s (m1m2 s + k (m1 + m2 )) u ( s ) s 2 (m1m2 s 2 + k (m1 + m2 ))
x1 ( s ) −k x2 ( s ) −(m1s 2 + k )
= = = = , u 0 ,u 0
d ( s ) s 2 (m1m2 s 2 + k (m1 + m2 )) d ( s ) s 2 (m1m2 s 2 + k (m1 + m2 ))
6
• Example 2: 2-DOF with damping
7
Θ1 ( s ) J 2 s 2 + D2 s + K Θ 2 ( s ) K
= , =
T (s) ∆ T (s) J1 J 2 s 4 + ( D1 J 2 + D2 J1 )s 3 + ( D1 D2 + KJ 1 + KJ 2 )s 2 + K ( D1 + D2 )s
Step Response
0.45
K = 10; 0.4
G1
J1 = 20; G2
J2 = 20; 0.35
D1 = 0.3;
0.3
D2 = 100;
%num = [J2 D2 K]; 0.25
Amplitude
t = 0:0.01:30;
0.2
num = K;
den=[J1*J2 0.15
D1*J2+D2*J1
D1*D2+K*J1+K*J2 0.1
Open-loop system is
K*(D1+D2) 0]
step(num,den,t)
0.05
unstable (divergent!)
0
0 5 10 15 20 25 30
Time (seconds)
8
• Example 3: 2-DOF
K1 K2 K3
M 1 x1 + ( f v1 + f v 2 ) x1 − f v 2 x 2 + ( K1 + K 2 ) x1 − K 2 x2 = u
M 2 x2 − f v 2 x1 + ( f v 2 + f v 3 ) x2 − K 2 x1 + ( K 2 + K 3 ) x2 = 0
[ M ] x + [ f v ]x + [ K ]x =
f
M1 0 f v1 + f v 2 − f v2 K1 + K 2 − K2
[M ] = , [ f ] = −f , [ K ] =
M 2 f v 2 + f v 3 −K K 2 + K 3
v
0 v2 2
9
M 1 x1 + ( f v1 + f v 2 ) x1 − f v 2 x 2 + ( K1 + K 2 ) x1 − K 2 x2 = f
M 2 x2 − f v 2 x1 + ( f v 2 + f v 3 ) x 2 − K 2 x1 + ( K 2 + K 3 ) x2 = 0
[ M 1s 2 + ( f v1 + f v 2 )s + ( K1 + K 2 )] X 1 − ( f v 2 + K 2 ) X 2 = F
− ( K 2 + f v 2 s ) X 1 + [ M 2 s 2 + ( f v 2 + f v3 )s + ( K 2 + K 3 )] X 2 = 0
X 1 ( s) M 2 s 2 + ( f v 2 + f v3 ) s + ( K 2 + K 3 )
=
F (s) ∆
10
Using Mathematica
X 1 ( s) M 2 s 2 + ( f v 2 + f v3 ) s + ( K 2 + K 3 )
=
F (s) ∆
M 2 s 2 + ( f v 2 + f v3 ) s + ( K 2 + K 3 )
=
M 1M 2 s 4 + ( f v3 M 1 + f v1M 2 + ( M 1 + M 2 ) f v 2 ) s 3 + ( K 2 K 3 M 1 + K1 K 2 M 2 + f v1 f v 2 + f v1 f v3 + f v 2 f v3 ) s 2 + Cs +
11
Transfer Function of Electrical Systems
• RC circuit
1 1
s → jω H ( jω ) = =
RCjω + 1 1 + (RCω )
2
low-pass filter ω = 1 1
• LRC circuit c =
RC τ
Slope=-
20/decade
12
• Analogy between Mechanical and Electrical Quantities
• Force Voltage Analogy
∝
mx + bx + kx =
F
13
TF of Electro-Mechanical Systems
• Armature controlled DC servomotor
14
Transfer Function of Hydro-mechanical Systems [Kim]
• Hydraulic Servo Valve System
β Orifice Q =CtP
P = (Q − Ct P − Ax ), Mx = PA − Cx − kx − F k
Vo P
M P
β A
sP ( s ) = (Q( s) − Ct P( s) − AsX ( s) ), X ( s) = P( s)
Vo Ms 2 + Cs + k
X (s) A
∴ =
Q( s ) Vo M VoC V k
s3 + ( + Ct M ) s 2 + o + Ct C + A2 s + Ct K
β β β
15
Summary
• 선형 시불변 (LTI) 시스템의 블록 선도는 네 가지 요소 (신호, 시스템, 합점 및
분기점)로 구성되어 TF G(s), Sensor H(s)등으로 구성된다.
• 장점: 복잡한 Closed-loop 블록 선도도 입력과 출력의 관계를 나타내는 단일
전달 함수로 단순화시킬 수 있다
16
State Space Model Approach
• State variable & state space model
• The set of variables that describe the system
• The state variable does not represent physical
quantities big freedom to get the State Space Model
0 1 0
Many SS Representations
= x (t ) k A c x + 1B u
TF : (canonical forms) − −
m m m
y = [1 C0] x
The only one converting
Many State-Space models
17
• State-Space Model
• Example 1
18
Example 2 : Quarter Car Suspension System
19
Phase variable (controllability) canonical form
• Example 1: numerator is no longer constant
20
• Method 3 (optional)
21
• MATLAB
[A,B,C,D] = tf2ss(num,den)
Example
Y ( s) 25.04 s + 5.008
= 3
U ( s ) s + 5.03247 s 2 + 25.1026 s + 5.008
.
x.1 − 5.0325 − 25.1026 − 5.008 x1 1
x = 1 0 0 x2 + 0u
.2
x3 0 1 0 x3 0
x1
y = [0 25.04 5.008] x2 + [0]u
x3
22
SS Model to TF
H (s ) = C(sI − A ) B
−1
Converting from SS to TF is
the only one!
23
• Example 1: Mass-spring-damper system
Numeric example
A ⇒ λ1 =
−1, λ2 =
−2
CE : s 2 + 3s + 2 = ( s + 1)( s + 2) = 0,
• Eigenvalues of A are same as the root
(poles) of characteristic equation ∴ s =−1, s =−2
24
• MATLAB Example
[num,den] = ss2tf(A,B,C,D)
Example
25
State Space Representation is NOT the only one!
• Example 1
Y (s) 10 s + 10
= 3
U ( s ) s + 6 s 2 + 5s + 10
. .
x.1 0 1 0 x1 0 x.1 − 6 −5 − 10 x1 1
x = 0 0 1 x2 + 10 u x = 1 0 0 x2 + 0u
.2 .2
x3 − 10 −5 − 6 x3 − 50 x3 0 1 0 x3 0
A B
x1 x1
y = [1 0 0] x2 + [0]u y = [0 10 10] x2 + [0]u
C x3 D x3
26
>>sys1 = ss(A,B,C,D) >>sys2 = ss(A,B,C,D)
>>step(sys1) >>step(sys2)
case1 case2
1.5 1.5
Amplitude
Amplitude
1 1
0.5 0.5
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Time (seconds) Time (seconds)
27
Why Do We Need Linear Approximation?
• Watt’s centrifugal governor : James Watt first introduced the
rotational steam engine automatic control system[1788]
x1 = x2
x2 = x32 (sin x1 cos x1 ) − sin x1 − γx2 State-space model
x3 = k (cos x1 − ρ ) [A,B,C,D]?
Nonlinear system !
28
• Linearization of Nonlinear System Using Taylor Approximation
• Step 1: Find equilibrium (operation) point static system
• Step 2: Applying Taylor Series at equilibrium (operation) point x0
where δ x= x − x0 δ y= y − y0
dg
a=
dx x = x0
δv(t ) + 2av0δv(t ) = δf (t )
δ v( s ) 1
sδ v( s ) + 2av0δ v( s ) =
δ f (s) ∴ G ( s )= =
δ f ( s ) s + 2av0
1st order linear TF
used for auto cruise control
30
• Example 2: Magnetic levitation System
• T. H. WONG Design of a Magnetic Levitation Control System Undergraduate, IEEE
TRANSACTIONS ON EDUCATION, VOL. E-29, NO. 4, NOVEMBER 1986 Project
2
i
EOM: mx = mg − Fm (i, x) = mg − C
2 x
i
At equilibrium point: mg = C
x
Linearization :
2 2
i i 2i 2 2i
Fm C ≈ C − C 3 xˆ + C 2 iˆ
∴=
x x x x
2 2 2
i 2i 2 2i ˆ i i 2i 2 2i
∴ mxˆ = mg − C + C 3 xˆ − C 2 i ⇒ mxˆ = C − C + C 3 xˆ − C 2 iˆ
x x x x x x x
2i 2 2i
⇒ mxˆ − C 3 xˆ = −C 2 iˆ ⇒ mxˆ − k1 xˆ =
− k2iˆ x2 xˆ=
x1 xˆ ,=
= , u iˆ
0 1 0
x x
⇒ x =xˆ =x
1 2 = A = k ,B −k
2i 2 2i 1
0 2
SS k k2 ˆ m m
where k1 C=
= 3 , k2 C 2 x2= xˆ= 1 x1 − i
x x m m C = [1 0]
y = xˆ
31
• Using Jacobian Matrix
• Example: nonlinear system x + 0.6 x + 3x + x 2 = 0
x1 = x2
x2 = −0.6 x2 − 3 x1 − x12
Jacobian
matrix
32
Concept of Stability
)
Pole-Zero Map 0.6 0.5
-1
1
0.4
Axis (seconds
Y (s) = 2
0
0
0.2
s +4 -2
Imaginary
-4 0 -0.5
-2 -1 0 1 2 0 2 4 6 8 10 0 2 4 6 8 10
-1
Real Axis (seconds )
)
1
0.2
0.2
Y (s) = 2
Axis (seconds
0.1
0
s + 3s + 4
0.1
0
Imaginary
-2 0 -0.1
-2 -1 0 1 2 0 2 4 6 8 10 0 2 4 6 8 10
-1
Real Axis (seconds )
)
1
0.2
0.1
Y (s) = 2
Imaginary Axis (seconds
0.1
0
s + 2 s + 10
0.05
0
-5
-2 -1
Convergent and Stable !
0
)
2
0
0 2 4 6 8 10
-0.1
0 2 4 6 8 10
)
10
0.015 0.05
1
Imaginary Axis (seconds
0.01 0
Y (s) =
0
s 2 + 2 s + 100
0.005 -0.05
-10 0 -0.1
-2 -1 0 1 2 0 2 4 6 8 10 0 2 4 6 8 10
-1
Real Axis (seconds ) 7 7
10 10
)
P l Z M 8 15
1
33
Pole-Zero Map 8 15
-1
1
1
6
10
s −4
2
-1 0 0
-2 -1 0 1 2 0 2 4 6 8 10 0 2 4 6 8 10
-1
Real Axis (seconds )
Pole-Zero Map 10 40
-1
1
5
20
Y (s) = 2
0
Imaginary Axis (seconds
s − s + 10
0
-10
-5 -20 -40
-2 -1 0 1 2 0 2 4 6 8 10 0 2 4 6 8 10
-1
Real Axis (seconds )
1 1000
1 40
Y (s) = 2
Imaginary Axis (seconds
Amplitude
Amplitude
0
s
500
20
-1 0 0
Time (seconds)
30 40 0 10 20
Time (seconds)
30 40
34
The Concept of Stability
• Stability is of utmost importance in designing and analysis of feedback
control systems.
– Absolute stability: either stable or not stable
– Relative stability : Considering the degree of stability
• A stable system is a dynamic system with a bounded response to a
bounded input (BIBO)
• If the system poles (eigenvalues) are negative, the system is stable in
the sense of Routh-Hurwitz Stability
35
Ubiquitous Examples
36
Historical Review on Stability
• Maxwell’s paper “On Governors” on feedback control and
stability analysis of James Watt’s Steam Engine in 1769
x1 x1 = x2
x2 = x32 (sin x1 cos x1 ) − sin x1 − γx2
x3
x3 = k (cos x1 − ρ )
Equilibrium points
x20 = 0
2 1 1
x30 cos x10 − 1 =0 ⇒ x30 =± =±
cos x10 ρ
k (cos x10 − ρ ) =0 ⇒ x10 =cos −1 ( ρ )
0 1 0
1− ρ2
Linearization J = ρ − γ ρ −γ 2
2 ρ (1 − ρ ) ∆( s ) = s 3 + γ s 2 + 2
s + 2k (1 − ρ ) = 0
ρ
2
− k 1 − ρ 0 0 3
γC = 2k ρ 2
37
The Routh-Hurwitz Stability Criterion
• Motivation
CE: ∆ ( s ) : an s n + an −1s n −1 + ... + a1s + a0 =
0
− a ± a 2 − 4b
( s ) : s 2 + as + b =0 s=
2
( s ) : s 3 + as 2 + bs + c =0
( s ) : s 4 + as 3 + bs 2 + c =0
s=?
38
CE: ∆ ( s ) : an s n + an −1s n −1 + ... + a1s + a0 =
0
39
• Routh Array
an s n + an −1s n −1 + an − 2 s n − 2 + an − 3 s n −3 + an − 4 s n − 4 ... + a1s + a0 =
0
• Place an , an −1 , an −2 , ..., a0
in the 1, 2 row of Routh array
• Compute b1 , b2 , ..., bn
n−3 0
• Repeat the computation of coefficient until s , ..., s
40
Special Case : multiple poles on the jw-axis
• If the jw-axis roots of the characteristic equation are simple, the system is
neither stable nor unstable - called marginally stable
• The Routh-Hurwitz criteria will not reveal the instability disadvantage
pole =
• Example 1
q( s) = s 5 + s 4 + 2s 3 + 2s 2 + s + 1 -1.0000
0.0000 + 1.0000i
0.0000 - 1.0000i
-0.0000 + 1.0000i
s5 1 2 1 s5 1 2 1 -0.0000 - 1.0000i
s4 1 2 1 s4 1 2 1 2
Pole-Zero Map
s3 ε ε 0 s3 4 4 0
s 2 1 1 p( s) = s + 2s
-1
s2 1 1
4 2
+1 -2
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4
s1 ε dP( s )
Real Axis (seconds -1)
s1 ε 0 ← Step Response
ds
200
s0 1 s0 1 100
Amplitude
0
41
[IEEE Control systems, 1997]
16
G (s) =
s 5 + s 4 + 8s 3 + 8s 2 + 16 s + 16
s = −1, j 2, j 2,−2 j , and − j 2
Unstable !
No unstable poles
42
Controller Design Using R-H Stability Criteria
• Example : Proportional control problem
K
G(s) ( s + 2)( s + 3)( s + 5)
=
1 + G(s) 1 + K
( s + 2)( s + 3)( s + 5)
K K
K = = 3
( s + 2)( s + 3)( s + 5) + K s + 10 s 2 + 31s + K + 30
30+K
s3 1 31
s2 10 K + 30 280 − K > 0, K + 30 > 0
280 − K
s1 0 ∴stable gain: -30 <K <280
10
s0 K + 30 0
43
0
10
28 Step Response K =-50 0
Step Response K =-20
-1
-0.5
-2
unstable stable
Amplitude
Amplitude
-3 -1
-4
-1.5
-5
-6 -2
0 20 40 60 80 100 120 0 5 10 15
1.5
Step Response
K =200 2
10
26 Step Response
K =282
1
1 0
Amplitude
Amplitude
-1
stable unstable
0.5 -2
-3
0 -4
0 5 10 15 0 2000 4000 6000 8000
44
Application Examples
• Example 1: Mass-spring-damper system
1
G(s) =
s 2 + cs + k
RH array s2 1 k
s1 c 0
s0 k
45
• Example 2: Inverted Pendulum (Unstable System)
• Transfer function approach after linear approximation
Inverted
controller
pendulum
θr = 0 − K ( s + 6) F (s) −0.011 θ
- s + 12 s 2 − 11
G(s) 0.011K ( s + 6)
=Gc ( s ) =
1 + G ( s ) H ( s ) ( s + 12)( s 2 − 11) + 0.011K ( s + 6)
1
( M − m )
G(s) = − CLCE : ∆ ( S ) =s 3 + 12 s 2 − (11 − 0.011K ) s + (0.066 K − 132) =0
Mg
s2 −
( M − m )
where
= , M 100 = kg , m 10kg , 11 − 0.011K < 0, 0.066 K − 132 > 0
= m, g 9.8m / s 2
1= ⇒ 2000 < K
46
Open-loop step response K = 2000 K = 2100
Pole-Zero Map Pole-Zero Map Pole-Zero Map
1 1 1
)
)
)
-1
-1
-1
0.5 0.5 0.5
-1 -1 -1
-4 -3 -2 -1 0 1 2 3 4 -12 -10 -8 -6 -4 -2 0 -12 -10 -8 -6 -4 -2 0
-1 -1 -1
Real Axis (seconds ) Real Axis (seconds ) Real Axis (seconds )
25
4 4000
20
3 3000
Amplitude
Amplitude
Amplitude
15
2 2000
10
1
Unstable 1000
Unstable
5
Stable
0 0 0
0 2 4 6 8 10 12 14 16 18 0 50 100 150 200 250 300 350 400 0 2 4 6 8 10
47
• State-Space Model Approach
−λ 1 0 0 M +m
M +m −λ 0 0 g 0 0
g −λ 0 0 M
M = −λ 0 −λ 0 − 0 −λ 1
0 0 −λ 1 mg
mg 0 0 −λ − 0 −λ
− 0 0 −λ M
M
M +m 2 M +m
= λ4 − gλ = 0, ⇒ λ = 0(double), λ = ± g
M M
∴ unstable due to right plane pole
48
Numeric Example
0 1 0 0 θ 0
M + m
1
M g 0 0 0 θ M
x (t ) = + F (t )
0 0 0 1 x 0
mg 1
− 0 0 0 x 22 Step Response
M M
10
5
y(t ) = [0 0 1 0]x (t )
0
M = 1, m = 0.1, = 1, g = 9.8
-5
Amplitude
>> eig(A)
-10
ans =
-15
0 -20
0 0 2 4 6 8 10 12 14 16 18
Time (seconds)
3.2833 Unstable !
-3.2833
49
• Example 3: Space Aircraft Attitude Control (Unstable System)
controller
θ
θr = 0 s+3 δ K ( s + 2) θ
- s s2 − 1
G(s) K ( s + 3)( s + 2)
=Gc ( s ) =
1 + G ( s ) H ( s ) s ( s 2 − 1) + K ( s + 3)( s + 2)
CLCE : ∆ ( S ) = s 3 + Ks 2 + [6 K − 1]s + 5 K = 0
δ s3 1 6K − 1
s2 K 5K
K ( s + 2) s1 6( K − 1) 0
G(s) =
s2 − 1 s0 − 0
∴ K −1 > 0 ⇒ K > 1
50
K > 1 for stable
Open-loop at K = 0.5 K = 0.5 K = 1.1
Pole-Zero Map Pole-Zero Map
Pole-Zero Map
1 2 3
)
)
-1
)
-1
-1
0.5 1
1
-1
-0.5 -1
Unstable pole -2
-1 -3
-2
-3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5
-1 -1 -1
Real Axis (seconds Real Axis (seconds )
Real Axis (seconds )
)
8 0.5 1.5
Amplitude
0
Amplitude
Amplitude
4
-0.5
0.5
2
Unstable
Unstable -1 Unstable Stable
0 0
0 10 20 30 40 50 60 0 50 100 150 200 250 0 20 40 60 80 100 120
51
• Example 4: Friction Clutch Shudder Instability
θ
k
Engine J Load
Friction
clutch c
Jθ + cθ + kθ = Tt
∂µ
Tt = µ (v ) FR µ (v ) = µ o + v
∂v
where F is apply force on the clutch,
where µ o is the static coefficient,
R is effective radius of the clutch
v = − Rθ
52
Define state variables x1 = θ , x2 = θ
x1 0 1 x 0
= − k − (c + mR 2 F ) 1 + µ 0 FR
x 2 J J
x2
J
⇒ (c + mR 2 F ) > 0
Stable
c ∂µ c µ range
⇒m>− ⇒ > −
R2F ∂v R2F
Velocity (v)
53