Beruflich Dokumente
Kultur Dokumente
Feedback Systems
Routh-Hurwitz criterion
Relative stability
Examples
1
The Concept of Stability
2
Why Stability Is Important
https://www.youtube.com/watch?v=XggxeuFDaDU
Y ( s ) Pi ( s ) i ( s )
T (s) = = ,
R( s) ( s)
1 M Ai N Bk s + C k
Y (s) = + + 2 ,
s i =1s + i k =1 s + 2 k s + ( k + n )
2 2
M i t N
y (t ) = 1 + Ai e + Dk e k t sin( k t + k ),
i =1 k =1
5
The Laplace Transform
s +
=
s 2 + 2s + ( 2 + 2 )
6
S-Plane and Transient Response
Marginal stable
7
Why Stability Is Important
a a
b c
r1 + r2 = r1r2 = ?
a a
For a stable system, the roots of the characteristic equation must have
negative real parts, or Re(r1 + r2 ) < 0, or the roots are real r1 + r2 < 0.
This implies: b/a>0 and c/a>0.
Is it possible to find the similar conditions for higher order system? 12
Relationship between the coefficients and roots of
the characteristic equation
Consider the nth-order characteristic equation as:
( s ) = q( s ) = a n s n + a n 1 s n 1 + ... + a1 s + a 0 = 0.
To ascertain the stability of the system it is necessary to
determine whether any one of the roots of q(s)=0 lies in
the right half of the s-plane (RHP). (If none, stable)
Rewrite the above equation in factored form, we have
a n ( s r1 )( s r2 ).....( s rn ) = 0,
where ri is the ith root of the characteristic equation.
13
Relationship between the coefficients and roots of
the characteristic equation
( s ) = q( s ) = a n s n + a n 1 s n 1 + ... + a1 s + a 0 = 0.
a n ( s r1 )( s r2 ).....( s rn ) = 0,
Note that all the coefficients of the polynomial much have the
same sign if all the roots are in the left-hand of the s-plane.
15
Relationship between the coefficients and
roots of the characteristic equation
q( s ) = a n s n a n ( sum of all the roots) s n 1
+ a n ( sum of the products of the roots taken 2 at a time) s n 2
a n ( sum of the products of the roots taken 3 at a time) s n 3 + ...
+ a n ( 1) n ( product of all n roots) = 0.
It is necessary that all the coefficients for a stable system be
nonzero.
Those requirements are necessary but not sufficient.
That is we immediately know the system is unstable if they are
not satisfied; yet if they are satisfied, we must proceed father to
ascertain the stability of the system . ?
E.g: q ( s ) = ( s + 2)( s s + 4) = s + s + 2 s + 8,
2 3 2
16
Relationship between the coefficients and
roots of the characteristic equation
q( s ) = ( s + 2)( s s + 4) = ( s + s + 2 s + 8),
2 3 2
17
Routh-Hurwitz Criterion
In the later 1800s, A. Hurwitz and E.J. Routh published
independently a method of investing the stability of a linear system.
20
Routh Array: Method (contd)
Then add subsequent
nn
rows to complete the ss :: aan n aan2
n 2
aan4
n 4
Routh array: s n 1 : a
n1
aan3 a
s : an1 n 1 n 3 an5
n 5
n 2
s : bn 1 bn 3 bn 5
n 3
s : cn 1 cn 3 cn 5
?! s :
2
1
s :
0
s : hn 1 21
Routh Array: Method (contd)
sn : an an 2 an 4 Compute elements for the rest
s n 1 : an 1 an 3 an 5 rows:
an 1an 2 an an
3 1 an1 aann 2 an 2
s n 2 : bn 1 bn 3 bn 5 bn 1 = bn 1 = = , ,
an 1 an 1 ann11 anan13 an 3
s n 3 : cn 1 cn 3 cn 5
1 an an 4
bn 3 = ,
an 1 an 1 an 5
s2 : 1 an 1 an 3
cn 1 = ,
s1 : bn 1 bn 1 bn 3
s0 : hn 1
23
Routh-Hurwitz Criterion
a4 a2 a4 a0 a4 0
a3 a1 a3 0 a3 0
= b3 for Routh table
Initial layout = b=
1 a0 =0
a3 a3 a3
a3 a1 a3 0 a3 0
b3 b1 b3 0 b3 0
= c3 =0 =0
b3 b3 b3
b3 b1 b 0 b3 0
3
c3 0 c3 0 c3 0
= b=
1 a0 =0 =0
c3 c3 c3
s 2 a 2 a0 s 3 a3 a1
s1 a1 0 s 2 a 2 a0
s 0 b1 0 s1 b1 0
s 0 c1 0
1 a2 a0 1 a3 a1 a2 a1 a0 a3
where b1 = = a0 where b1 = = and c1 =
b1a0
= a0
a1 a1 0 a 2 a 2 a0 a2 b1
Stable if all are of same sign Stable if all are of same sign and
a1a2 > a0 a3 .
26
Simple Systems
All zeros of polynomial a1s + a0 are in LHP if all
coefficients are positive.
All zeros of polynomial 2 + a1s + a0 are in LHP if all
2
a s
coefficients are positive.
All zeros of polynomial 3 a s 3
+ a 2 s 2
+ a1s + a0 are in LHP if all
coefficients are positive and a1a2 a0a3 > 0.
27
Routh Hurwitz Special Cases
For roots to be in LHP( the left half side of the s-plane) , it is
necessary (but sufficient only in second order case) that all
the coefficients be positive
Issues with awkward zero entries:
Case A: A zero-entry appears in the first column,
but other entries on the row are non-zero
-Solution: Take entry as small value >0 and
proceed, taking > 0 in subsequent calculations
Example: q( s ) = s 3 + 2 s 2 + 4 s + K ,
s3 1 4 s3 1 4
s2 2 K s2 2 K qa (s )
8 K k =8 1
qa ( s ) = 2 s 2 + Ks 0 = 2 s 2 + 8
1
s 0 s 0 0
0
2
s K 0 s0 K 0 = 2( s 2 + 4) = 2( s + j 2)( s j 2).
0 < K < 8. 31
Case 3. A zero-entry appears in the first column,
and all other entries in that row are also zero
-Solution: Return to the previous row and form the Auxiliary Polynomial, qa(s),
which will be a divisor of the original q(s), divide out qa(s), and proceed.
Example: q( s ) = s 3 + 2 s 2 + 4 s + K , q( s ) = qa ( s )(?)
1 / 2s + 1
s 3 1 4 s3 1 4
(?) = q( s ) / qa ( s ) 2 s 2 + 8 s 33 + 2 s 2 + 4 s + 8
s2 2 K
k =8
s2 2 K qa (s )
s + 4s
8 K
s1 0 0
+8
1
s 0 2s2
2 s0 K 0
s 0
K 0 2s 2 +8
=0
When K=8 qa ( s) = 2s 2 + Ks 0 = 2s 2 + 8 = 2( s 2 + 4) = 2( s + j 2)(s j 2).
q (s ) = qa (s )(?) = qa (s )(1/ 2 s + 1) = (s + 2)(s + j 2)(s j 2) (when K = 8)
Marginal stable 32
Example Fifth-order system with roots on the j-axis
q ( s ) = s 5 + s 4 + 4 s 3 + 24s 2 + 3s + 63
s5 1 4 3
s4 1 24 63
s 3 20 60 0
s 2 21 63 0
s1 0 0 0 Case 3
The auxiliary polynomial:
q( s)
= s 3
+ s 2
+ s + 21
s +3
2
s3 1 1
s2 1 21
s1 20 0
s0 21 0
The system is unstable. The roots in the right-hand plane are
s = +1 j 6 34
Case 4. Repeated roots of the characteristic equation on
the j-axis
Ifthe j-axis roots of the characteristic equation are simple (first order),
the system is neither stable nor unstable; it is instead called marginal
stable, since it has an un-damped sinusoidal mode.
- If the j-axis roots are repeated, the system response will be
unstable, with a form t[sin(t+)] in time domain.
- The Routh-Hurwitz criterion will not reveal this form of
instability.
q( s ) = ( s + 1)( s + j )( s j )( s + j )( s j ) = s 5 + s 4 + 2 s 3 + 2 s 2 + s + 1.
s5 1 2 1
s 4
+ 2 s 2
+ 1 = ( s 2
+ 1) 2
Note that the absence of sign changes,
s4 1 21
3 0 as 0 a condition that falsely indicates that
s
the system is marginally stable.
s 2 11
( s + 1)
2
36
The Relative Stability of Feedback Control Systems
The verification of stability using the Routh-Hurwitz
criterion provides only a partial answer to the question
of stability----whether the system is absolutely stable.
In practice, it is desired to determine the relative stability.
- The relative stability of a system can be defined
as the property that is measured by the relative
real part of each root or pair of roots.
37
The Relative Stability of Feedback Control Systems
Because the relative stability of a system is dictated by location
of the roots of the characteristic equation, we can extend the
?
Routh-Hurwitz criterion to ascertain relative stability.
This can be accomplished by utilizing a change of variable, which shifts
the s-plane vertical axis in order to utilize the Routh-Hurwitz criterion.
The correct magnitude of shift the vertical axis must be obtained on a
trial-and-error basis.
One may determine the real part of the dominant root without solving
the high order polynomial q(s).
j
j
38
Example: Axis Shift
Consider the simple third -order characteristic equation
q( s ) = s 3 + 4 s 2 + 6s + 4. (a1a2 > a0 a3 ) 4 6 1 4 > 0 stable
?
First try: let sn = s + 2 There is no zero in the first column of (shifting
Routh array, but with two unstable roots. too much)
Second try: sn = s + 1 We have
(sn 1)3 + 4(sn 1) 2 + 6(sn 1) + 4 = sn3 + sn2 + sn + 1.
3 1 1
sn
sn2 1 1 ( sn ) = sn2 + 1 = ( sn + j )( sn j ) = ( s + 1 + j )( s + 1 j )
s1
n
0 0
0 1 0
sn
The shifting of the s-plane axis to ascertain the relative stability of a
system is a very useful approach, particularly for higher-order
system with several pairs of closed-loop complex conjugate roots. 39
Tracked Vehicle Turning Control
Goal: find K and a such that system is stable and ess to ramp
command to be less than 24%.
Gc ( s )G ( s ) K (s + a)
T ( s) =
1 + Gc ( s )G ( s ) 1 + Gc ( s )G ( s ) = 0 1 + =0
s( s + 1)( s + 2)( s + 5)
ess = 1 / K v = 1 / lim sGcG = 1 / ( Ka / 10) = 10 / Ka
s 0
10 / Ka 0.24 Ka 10 / 0.24
= 41.667 42
40
Tracked Vehicle Turning Control
K (s + a)
1+ =0
s( s + 1)( s + 2)( s + 5)
s 4 + 8s 3 + 17 s 2 + ( K + 10) s + Ka = 0
s4 1 17 Ka 1 17
s3 8 ( K + 10) 0 8 K + 10 126 K
b3 = =
s2 b3 Ka 8 8
8 K + 10
s1 c3
b3 Ka b3 ( K + 10) 8 Ka
s0 Ka c3 = =
b3 b3
10 / Ka 0.24 Ka 10 / 0.24 =
41.7
Ka = 42 if K = 70 a = 0.6 43
Example Stability Region for Unstable Plant
A jump-jet aircraft has a control system as shown in
Note that since one term within the bracket has a negative
coefficient, the characteristic equation has at least one root
in the right-hand side of the s-plane. 44
Example Stability Region for Unstable Plant
45
Example Stability Region for Unstable Plant
46
Example Stability Region for Unstable Plant
47
Example Stability Region for Unstable Plant
48
Example Stability Region for Unstable Plant
Example Stability Region for Unstable Plant
Gc G K (s + z) 1
=T = , Gc = ,G
1 + Gc G s+ p s ( s 1)
b2
( p 1)( K p) Kz
>0
( p 1)( K p) Kz > 0 since (p 1)>0.
( p 1)
( p 1)( K p) Kz > 0 K [( p 1) z ] p( p 1) > 0
p( p 1)
K < if z > ( p 1)
( p 1) z
K [( p 1) z ] p ( p 1) > 0 p( p 1) > 0 if z = p 1
< 0 ? K > p( p 1) if z < ( p 1)
( p 1) z
( p 1) z = = 0 ?
> 0 ?
Consider two cases:
z ( p 1) Unstable region
Now:
53
Sequential Design Example: Disk Drive Read System
5K a
Y ( s) = R( s ) ,
s + 20 s + 5K a
2
Kv
The best compromise (Ka=40) still does not meet all the specifications. (Ts=400ms)
What to do?
54
Sequential Design Example: Disk Drive Read System
Without velocity feedback:
Y ( s) K a G1 ( s )G2 ( s ) G ( s ) = 5000
=
s + 1000
1
R ( s ) 1 + K a G1 ( s )G2 ( s )
55
Sequential Design Example: Disk Drive Read System
With velocity feedback:
Y ( s) K a G1 ( s )G2 ( s )
=
R ( s ) 1 + [ K a G1 ( s )G2 ( s )](1 + K1s )
5000 1
The characteristic equation is 1 + Ka (1 + K1s ) = 0
( s + 1000) s( s + 20)
s( s + 20)( s + 1000) + 5000 K a (1 + K1s ) = 0 or
s 3 + 1020s 2 + [20000 + 5000 K a K1 ]s + 5000 K a = 0
Condition for stability is ( a3s 3 + a2 s 2 + a1s + a0 = 0) 3rd Order system
a1a2 a0 a3 > 0 1020 [20000 + 5000 K a K1 ] 5000 K a > 0
58
Additional Example 1: Stability versus Parameter Range
Consider a feedback system such as:
59
Additional Example 1 (contd)
s +1
1+ K =0
s ( s 1)( s + 6)
60
Additional Example 1 (contd)
s +1
1+ K =0
s ( s 1)( s + 6)
s ( s 1)( s + 6) + K ( s + 1)
=0
s ( s 1)( s + 6)
Expressing the characteristic equation
in polynomial form, we obtain:
s + 5 s + ( K 6) s + K = 0
3 2
61
Additional Example 1 (contd)
q(s) = s + 5s + (K 6)s + K
3 2
s0 : K K > 7.5
62
Additional Example 1 (contd)
q ( s ) = s + 5s + ( K 6) s + K
3 2
63
64
64
Additional Example 2: Stability versus Two
Parameter Range
65
Additional Example 2 (contd)
66
Additional Example 2 (contd)
K1 1
1+ K + =0
s ( s + 1)( s + 2)
s 3 + 3s 2 + ( 2 + K ) s + K1 = 0
67
Additional Example 2 (contd)
q( s ) = s + 3s + ( 2 + K ) s + K1 = 0
3 2
s3 : 1 2+K
s2 : K1
3 K1 K1 > 0 and K > 2
s1 : (6 + 3K K1 ) 3 3
s0 : K1
68
Additional Example 2: Allowable region
for Stability
K
K = 1 2
3
K1
K1 > 0 and K > 2
3
69
Additional Example 2: Transient Response for the
System
70
Summary
Routh Hurwitz stability criterion allows to check for
stability without computing roots of characteristic
equation
71