Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
CONTROL SYSTEMS
Radar Dish Mechanical Engineering Department,
Middle East Technical University
Armature Inside
controlled Outside θr input
dc motor θD output
θm
G
Gearbox
b
Control
transmitter
θD
dc amplifier Control
transformer P f Dr.
Prof. D Y
Y. S
Samim
i ÜÜnlüsoy
lü
ME 304 CONTROL SYSTEMS Prof. Dr. Y. Samim Ünlüsoy STABILITY - 1
CH IV
COURSE OUTLINE
I. INTRODUCTION & BASIC CONCEPTS
II. MODELING DYNAMIC SYSTEMS
III. CONTROL SYSTEM COMPONENTS
IV
IV. STABILITY
V. TRANSIENT RESPONSE
VI. STEADY S
S STATE RESPONSE
S O S
VII. DISTURBANCE REJECTION
VIII. BASIC CONTROL ACTIONS & CONTROLLERS
IX
IX. FREQUENCY RESPONSE ANALYSIS
X. SENSITIVITY ANALYSIS
XI. ROOT LOCUS ANALYSIS
Bounded-Input,
Bounded- p , Bounded-
Bounded-Outputp Stabilityy
(Zero state response) : A linear time
invariant system is said to be stable if it
produces a bounded response to a
bounded input.
Zero Input and Asymptotic Stability (Zero
input response) : A system is stable, if
the zero input response due to finite
initial conditions returns to zero
asymptotically as time goes to infinity.
First
Order
Second
Order
Time
q r ⎡ ⎤
+ ∑ ⎢ Ak e−ξk ωk t cos ⎜⎛ ωk 1 − ξk2 t − φk ⎟⎞ ⎥
−p jt
y(t) = a + ∑ a je
j= 1 k =1 ⎣ ⎝ ⎠⎦
Re
Re
Im
Re
Routh’s stability
y criterion allows the
determination of
• whether there are any roots of the
characteristic equation with positive
real parts
and,
d if there
th are,
• the number of these roots
without actually finding the roots.
D(s)=ansn+an-1sn-1+...+a1s+a0
the Routh
Routh’s
s array is a special
arrangement of the coefficients in a
certain pattern.
sn an an-2 an-4 . . .
an-1an-2 - anan-3
sn-1 an-1 an-3 an-5 . . . b1 =
an-1
sn-2 b1 b2 b3 ... an-1an-4 - anan-5
b2 =
sn-3 c1 c2 c3 ... an-1
. .
. . b a - b2 an-1
c1 = 1 n-3
. . b1
s2 e1 b1an-5 - b3 an-1
c2 =
s1 f1 b1
s0 g1
Example : s3+20s2+9s+100
D(s)=s
D(s)
Passes Hurwitz test !
s3 1 9
s2 20 100
(20 )( 9 ) - (1)(100 )
s1 4 b1 =
20
=4
s0 100
00 ( 4 )(100 ) - (20 )(0 )
c1 = =100
4
s1 = -3.0000
Example : D(s)=s3+s2+2s+24
s2 = 1.0000
1 0000 + 2.6458i
2 6458i
Passes Hurwitz test ! s3 = 1.0000 - 2.6458i
s3 1 2
s2 1 24
(1)(2 ) - (1)(24 )
b1 = = -22
s1 -22 0 1
s0 24 ( -22 )(24 ) - (1)(0 )
c1 = = 24
Knight’s move -22
Special Cases :
There are some cases in which
problems appear
p pp in completing
p g the
Routh’s array.
They are encountered in the case of
systems that are not stable, and means
are devised to allow the completion of
the Routh’s
Routh s array.
Special Case 1 :
When a first column term in a row
becomes zero with all other terms
being nonzero,
nonzero the calculation of the
rest of the terms becomes impossible
due to division by zero.
In such a case the system is unstable
and the procedure is continued just to
determine the number of roots with
positive real parts.
Example : D(s)=s5+2s4+2s3+4s2+11s+10
s5 1 2 11
(2 )(2 ) - (1)( 4 )
s4 2 4 10 b1 = =0
2
s3 0 6 0 (2 )(11) - (1)(10 )
b2 = =6
2
s2 ???
(0 )( 4 ) - (2 )(6 )
c1 =
s1 0
s0
ME 304 CONTROL SYSTEMS Prof. Dr. Y. Samim Ünlüsoy STABILITY - 24
ROUTH’S STABILITY CRITERION
Example : D(s)=s5+2s4+2s3+4s2+11s+10
s5 1 2 11 In such a case,
s4 2 4 10 replace zero term
s3 0 6 0 by a very small
s1 number
b ε.
s0
ME 304 CONTROL SYSTEMS Prof. Dr. Y. Samim Ünlüsoy STABILITY - 25
ROUTH’S STABILITY CRITERION
Example : D(s)=s5+2s4+2s3+4s2+11s+10
s5 1 2 11 ( 4 )( ε ) - (2 )(6 )12
c1 = =4-
ε ε
s4 2 4 10
12
ε → 0 ⇒ c1 ≅ -
ε
s3 0→ε 6 0
⎛ 12 ⎞
⎜- ⎟ ( 6 ) - (10 )( ε ) 10 2
ε ⎠
s2 -
12
10 0 d1 = ⎝ =6+ ε
ε 12 12
-
ε
s1 6 0 ε → 0 ⇒ d1 ≅ 6
s0 10
s5 1 2 11 2 sign changes :
s4 2 4 10 2 roots with
Special Case 2 :
If all
ll the
th terms
t on a derived
d i d row are zero,
this means that the characteristic equation
has roots which are symmetric
y with
respect to the origin,
origin,
1. Two real roots with equal magnitudes
but opposite signs, and/or
2. Two conjugate imaginary roots, and/or
3. Two complex roots with equal real and
imaginary parts of opposite signs.
Re
s3 1 1
s2 2 2 Q(s)=(2)s2 +(2)s0
s1 0 0
Replace
row of
dQ(s)
Q( )
zeroes =(4)s +(0)s0
4 0 with ds
dQ/ds
s0 2
No s
sign
ign changes in the 1st column :
No roots with positive real parts.
s2 2 2 s3 +2s2 +s +2 s
= +1 s3 = -2
2s2 +2 2
s1 0 0
4 0 s +j
-2
s0 2
1
s
3 s -j
2
D(s)=s3+4s2+3s+Kp
ME 304 CONTROL SYSTEMS Prof. Dr. Y. Samim Ünlüsoy STABILITY - 35
EXAMPLE 1b
D(s)=s3+4s2+3s+Kp
s3 1 3 For stability :
s2 4 Kp Kp
3- > 0 and Kp > 0
4
Kp
s1 3- 0
4
s0 Kp
0 < Kp <12
12
Kp For Kp =12 :
s1 3- 0
4 ( s +4 ) ( s2 +3 ) = 0
s0 Kp s1 = -4, s2 = + 3j, s3 = - 3j
Marginally stable!
Oscillatory response,
Undamped oscillations.
What is the frequency of undamped oscillations ?
ME 304 CONTROL SYSTEMS Prof. Dr. Y. Samim Ünlüsoy STABILITY - 37
STABILITY BOUNDARIES – Example 2a
Proportional+Integral (PI)
Controller
R(s) ⎛ 1 ⎞ 1 C(s
Kp ⎜1 + ⎟
+ s ( s +1)( s +3 ) )
_
⎝ Ti ⎠
s
R(s) ⎛ 1 ⎞ 1 C(s
Kp ⎜1 + ⎟
+ s ( s +1)( s +3 ) )
_
⎝ Ti ⎠
s
G1(s)
( ) G2(s)
( )
⎛ 1 ⎞⎛ 1 ⎞
Kp ⎜ 1 + ⎟ ⎜⎜ ⎟⎟
G1 (s)G 2 (s) ⎝ Ti s ⎠ ⎝ s ( s + 1)( s + 3 ) ⎠
G(s) = =
1 + G1 (s)G 2 (s) ⎛ 1 ⎞⎛ 1 ⎞
1 + Kp ⎜ 1 + ⎟ ⎜⎜ ⎟⎟
⎝ Ti s ⎠ ⎝ s ( s + 1)( s + 3 ) ⎠
R(s) ⎛ 1 ⎞ 1 C(s
Kp ⎜1 + ⎟
+ s ( s +1)( s +3 ) )
_
⎝ Ti ⎠
s
G1(s)
( ) G2(s)
( )
K p ( Ti s + 1)
G(s) =
Ti s 2 ( s + 1)( s + 3 ) + K p ( Ti s + 1)
s4 Ti 3Ti Kp Ti > 0
Kp > 0
s3 4Ti KpTi 0
Kp
3- > 0 ⇒ K p < 12
s2 ⎛ Kp
⎜⎜ 3 -
⎞
⎟⎟ T i
Kp 0
4
⎝ 4 ⎠
⎛ ⎞
s1 ⎜ i
⎜
T -
16
⎟ Kp
⎟
0 0 0 < K p < 12
⎝ 12 - K p ⎠
s0 Kp Ti -
16
>0 Ti >
16
12 - K p 12 - K p
D(s)=s3+12s2+46s+52
can provide a stability margin of 2.1.
21
The approach in this problem is to shift the
origin of the complex plane towards the left
by 2.1.
If the system
y is still stable,, then the system
y
has a stability margin greater than 2.1. If
the system becomes unstable, then it has a
stability margin of less than 2.1.2 1.
2.1
ME 304 CONTROL SYSTEMS Prof. Dr. Y. Samim Ünlüsoy STABILITY - 46
STABILITY MARGIN – Example 3b
Positions of the roots :
Original Im After Im
shifting by 2
Re Re
-5 -4 -3 -2 -1 -3 -2 -1 0 +1
s1 = - 5.0000
5 0000 + 1.0000i
1 0000i s1 = - 3.0000
3 0000 + 1.0000i
1 0000i
s2 = - 5.0000 - 1.0000i s2 = - 3.0000 - 1.0000i
s3 = -2.0000 s3 = 0
z3 1 8 83
8.83 Polynomial
z1 8.995 0 test !
z0 -0.941
Since th
Si there is
i only
l one sign
i change,
h
there is one root with a positive real part
g from 0 to 2.1.
in the range