State-space feedback 3 transformation to get a canonical...
Transcript of State-space feedback 3 transformation to get a canonical...
State-space feedback 3 transformation to get a canonical
form
J A Rossiter
1
Slides by Anthony Rossiter
Introduction
• The previous video showed how state feedback can place poles precisely when the system is in control canonical form.
• More generally, the system is not in canonical form, but we may still wish to place the closed-loop poles.
• This video shows how the a transformation can create the canonical form, and thus implicitly allows explicit and straightforward pole placement.
Slides by Anthony Rossiter
2
xBKAxKxu
BuAxx
)(
Control canonical form
Slides by Anthony Rossiter
3
zbbby
uz
aaaa
zdt
d
C
nn
BA
nn
021
0121
;
0
01
0100
0001
Remark: A transformation to control canonical form only exists if the system is
fully controllable.
Observation
In control canonical form, the eigenvalues are determined solely by the parameters along the top row of the A matrix.
Slides by Anthony Rossiter
4
Corollary: One can place the poles precisely if one can select the top row of A precisely without changing any other coefficients.
0
1
1
0121
0100
0001
aaAI
aaaa
A
n
n
n
nn
Similarity transformation (chapter 1)
Consider how the state feedback varies for two alternative representations linked by a similarity transform.
Slides by Anthony Rossiter
5
BuzATzTBuAxx 11
uBzAz
uTBzTATzTTBA
ˆˆ
ˆˆ
11
zCTy
Cxy
C
1
zTxTxz 1;
Similarity transform continued
Let the state feedback for system A, B be K.
In the transformed state, an equivalent control law is given by:
Slides by Anthony Rossiter
6
xBKAxKxu
BuAxx
)(
zKBAzzKTKxu
uBzAz
K
ˆˆ
1 )ˆˆˆ(
ˆˆ
zTxTxz 1;
Compare controllability matrices
For the original system.
For the transformed system.
Slides by Anthony Rossiter
7
],,,,[ 12 BABAABBM n
cx
]ˆˆ,,ˆˆ,ˆˆ,ˆ[ 12 BABABABM n
cz
])(,,)(,,[
ˆ,ˆ
11211
1
TBTATTBTATTBTATTBM
TATATBB
n
cz
cx
n
cz TMBABAABBTM ],,,,[ 12
Controllability matrices are linked by the
transformation matrix!
REMARK
We can get the control canonical form directly from the transfer function so this is known for any A,B,C,D.
Slides by Anthony Rossiter
8
zbbby
uz
aaaa
zdt
d
C
nn
BA
nn
ˆ
021
ˆˆ
0121
;
0
01
0100
0001
)()(01
2
2
1
1
01
1
1 sUasasasas
bsbsbsY
n
n
n
n
n
UDBAsICY ])([ 1
INSIGHT
If we know the system in control canonical form, then we can find the transformation that takes us from an arbitrary form to the control canonical form!
Slides by Anthony Rossiter
9
BuAxx
Given form Control canonical form
uBzAz ˆˆ zTxTxz 1;
cxcz M
n
M
n BABAABBTBABABAB ],,,,[]ˆˆ,,ˆˆ,ˆˆ,ˆ[ 1212
INSIGHT CONTINUED If we know the TRANSFORMATION that takes us from control canonical form to another form, we can do pole placement in control canonical form, and then find the equivalent state feedback for an alternative form.
Slides by Anthony Rossiter
10
KxuBuAxx ;
Given form Control canonical form
zKuuBzAz ˆ,ˆˆ
zTxTxz 1; TMM cxcz 1][
TKKKTK ˆˆ 1
Pole placement algorithm
1. Find transfer function representation.
2. Find control canonical form.
3. Find pole placement state feedback for control canonical form.
4. Find transformation matrix using controllability matrices.
5. Find state feedback for original state space system.
Slides by Anthony Rossiter
11
uBzAz ˆˆ
zKu ˆ
TMM cxcz 1][
TKK ˆ
BAsIC 1)(
EXAMPLES
Slides by Anthony Rossiter
12
Example 1: Choose K to set the closed-loop poles at -1 and -2.
Slides by Anthony Rossiter
13
First find the control canonical form.
43;2
1;
4.01
21
CBA
Next find the similarity transform relating the two.
2.95ˆ;0
1ˆ;
01
4.24.1ˆ
CBA
4.24.1
2.95)(
2
1
ss
sBAsIC
Example 1: Choose K to set the closed-loop poles at -1 and -2.
Slides by Anthony Rossiter
14
Find the controllability matrices.
8.12
31
],[
;2
1;
4.01
21
cx
cx
M
ABBM
BA
Next find the similarity transform relating the two.
10
4.11
]ˆˆ,ˆ[
;0
1ˆ;
01
4.24.1ˆ
cz
cz
M
BABM
BA
TMMTxz cxcz 1][
Example 1
Slides by Anthony Rossiter
15
Define feedback for control canonical form, Desired pole polynomial is s2+2s+1.
Find feedback for original system.
01
12
01
4.24.1ˆˆˆ
;0
1ˆ;
01
4.24.1ˆ
21 kkKBA
BA 4.16.0ˆ K
52.044.0
13.026.0
56.013.0][;ˆ 1
K
TMMTKK cxcz
Confirm results using MATLAB
Slides by Anthony Rossiter
16
Example 2: Choose K to set the closed-loop poles at -0.5, -1 and -1.5.
Slides by Anthony Rossiter
17
First find the control canonical form.
Next find the similarity transform relating the two.
39.018.0
21.02.12)(
23
21
sss
ssBAsIC
431;
0
1
1
;
4.01.00
6.04.02.0
5.021
CBA
21.02.12ˆ;
0
0
1
ˆ;
010
001
39.018.01
ˆ
CBA
Example 2
Slides by Anthony Rossiter
18
Find the controllability matrices.
02.01.00
1.06.01
25.211
],,[ 2
cx
cx
M
BAABBM
Next find the similarity transform relating the two.
100
110
82.011
]ˆˆ,ˆˆ,ˆ[ 2
cz
cz
M
BABABM
TMMTxz cxcz 1][
Example 2
Slides by Anthony Rossiter
19
Define feedback for control canonical form, Desired pole polynomial is s3+3s2+2.75s+0.75.
Find feedback for original system.
14.157.22ˆ K
57.2018.218.0
][;ˆ 1
K
TMMTKK cxcz
010
001
75.075.23
010
001
39.018.01
ˆˆˆ321 kkk
KBA
Confirm results using MATLAB
Slides by Anthony Rossiter
20
Summary Introduced concepts of pole placement state feedback without a control canonical form.
1. Show that, assuming full controllability, there exists a transformation matrix to generate the equivalent control canonical form.
2. Pole placement design can be done using the canonical form.
3. Feedback parameters for the original states are obtained using the corresponding similarity transformation which is defined from the controllability matrices (see algorithm).
4. Not paper/pen exercise in general.
Slides by Anthony Rossiter
21
Kxu
© 2016 University of Sheffield This work is licensed under the Creative Commons Attribution 2.0 UK: England & Wales Licence. To view a copy of this licence, visit http://creativecommons.org/licenses/by/2.0/uk/ or send a letter to: Creative Commons, 171 Second Street, Suite 300, San Francisco, California 94105, USA. It should be noted that some of the materials contained within this resource are subject to third party rights and any copyright notices must remain with these materials in the event of reuse or repurposing. If there are third party images within the resource please do not remove or alter any of the copyright notices or website details shown below the image. (Please list details of the third party rights contained within this work. If you include your institutions logo on the cover please include reference to the fact that it is a trade mark and all copyright in that image is reserved.)
Anthony Rossiter Department of Automatic Control and
Systems Engineering University of Sheffield www.shef.ac.uk/acse