Pole Placement Control Design: Dr. Radhakant Padhi

Lecture – 21
Pole Placement Control Design
Dr. Radhakant Padhi

Asst. Professor
Dept. of Aerospace Engineering
Indian Institute of Science - Bangalore
Assumptions:
The system is completely state controllable.
The state variables are measurable and are

available for feedback.
Control input is unconstrained.
ADVANCED CONTROL SYSTEM DESIGN 2

Dr. Radhakant Padhi, AE Dept., IISc-Bangalore
Objective:
The closed loop poles should lie at μ1 ,… , μn , which are their
‘desired locations’.
Difference from classical approach:

Not only the “dominant poles”, but “all poles” are forced to lie at
specific desired locations.
Necessary and sufficient condition:

The system is completely state controllable.

Closed Loop System Dynamics
X = AX + BU
The control vector U is designed in the following
state feedback form
U = − KX
This leads to the following closed loop system
X = ( A − BK ) X = ACL X
where ACL ( A − BK )

Philosophy of
The gain matrix K is designed in such a

way that
sI − ( A − BK ) = ( s − μ1 )( s − μ2 ) ( s − μn )
where μ1 , , μn are the desired pole locations.

Pole Placement Design Steps:
Method 1 (low order systems, n ≤ 3)
z Check controllability
z Define K = [ k1 k2 k3 ]
z Substitute this gain in the desired
characteristic polynomial equation
sI − A + B K = ( s − μ 1 ) (s − μ n )
z Solve for k1 , k2 , k3 by equating the like
powers on both sides

Pole Placement Control Design:
Method – 2
X = AX + Bu
u = −K X , K = [ k1 k2 kn ]
Let the system be in first companion (controllable canonical) form
⎡ 0 1 0 0 0 ⎤ ⎡0⎤
⎢ 0 0 1 0 0 ⎥⎥ ⎢0⎥
⎢ ⎢ ⎥
A=⎢ ⎥,B = ⎢ ⎥
⎢ ⎥ ⎢ ⎥
⎢ 0 0 0 0 1 ⎥ ⎢0⎥
⎢⎣ −an −an−1 −an−2 −an−3 −a1 ⎥⎦ ⎢⎣1⎥⎦

After applying the control, the closed loop system
dynamics is given by
X = ( A − BK ) X = ACL X
⎡ 0 1 0 0 ⎤ ⎡0 0 0 0⎤
⎢ 0 0 1 0 ⎥⎥ ⎢⎢ 0 0 0 0 ⎥⎥
⎢
⎢ ⎥ ⎢ ⎥
ACL = ⎢ ⎥−⎢ ⎥
⎢ 1 ⎥ ⎢ ⎥
⎢ 0 0 0 0 ⎥ ⎢0 0 0 0⎥
⎢ ⎥ ⎢ ⎥
⎢⎣ − a n − a n −1 − a n − 2 − a1 ⎥⎦ ⎢⎣ k1 k2 k3 k n ⎥⎦
⎡ 0 1 0 0 0 ⎤
⎢ 0 0 1 0 0 ⎥
⎢ ⎥
⎢ 0 0 0 1 0 ⎥
ACL = ⎢ ⎥ (1)
⎢ 0 0 1 0 ⎥
⎢ 0 0 1 ⎥
⎢ ⎥
⎢⎣( −an − k1 ) ( −an −1 − k2 ) ( 1 n )⎥⎦
− a − k
Method – 2
If μ1 , , μ n are the desired poles. Then the desired
characteristic polynomial is given by,
( s − μ 1 ) ( s − μ n ) = s n + α 1 s n −1 + α 2 s n − 2 + + α n
This characteristic ⎡ 0 1 0 0⎤
⎢ 0 0 1 0 ⎥⎥
polynomial, will lead to ⎢
the closed loop system ⎢ ⎥
⎢ ⎥
matrix as ACL = ⎢ ⎥ ( 2)
⎢ 1⎥
⎢ ⎥
⎢ 0 0 ⎥
State space form ⎢ −α −α −α n − 2 −α1 ⎥⎦
⎣ n n −1

Method – 2
Comparing Equation (1) and (2), we arrive at:
⎡ an + k1 = αn ⎤ ⎡ k1 = (αn − an ) ⎤
⎢ ⎥ ⎢ ⎥
⎢ an −1 + k2 = α n −1 ⎥⇒ ⎢ k2 = ( α n −1 − an −1 ) ⎥
⎢ ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥
⎣ a1 + kn = α1 ⎦ ⎢⎣ kn = (α1 − a1 ) ⎥⎦
K = (α − a ) ( Row vector form)
What if the system is not given
in the first companion form?
Define a transformation X = TXˆ
Xˆ = T −1 X
Xˆ = T −1 ( AX + Bu )
( ) (
Xˆ = T −1 AT Xˆ + T −1B u )
Design a T such that T-1AT will be in first companion form.
Select T = MW
where M ⎡⎣ B AB An−1B ⎤⎦ is the controllability matrix

Method – 2
⎡ an −1 an − 2 a1 1⎤
⎢a 0 ⎥⎥
⎢ n−2
W =⎢ ⎥
⎢ ⎥
⎢ a1 1 ⎥
⎢⎣ 1 0 0 ⎥⎦
Next, design a controller for the transformed system (using the
technique for systems in first companion form).
u = − Kˆ Xˆ = − ( Kˆ T −1
)X = −KX
Note: Because of its role in control design as well as the use of M
(Controllability Matrix) in the process, the ‘first companion
form’ is also known as ‘Controllable Canonical form’.
Method 2: Bass-Gura Approach
z Check the controllability condition
z Form the characteristic polynomial for A
| sI − A |= s n + a1s n −1 + a2 s n − 2 + + an −1s + an
find ai’s
z Find the Transformation matrix T
z Write the desired characteristic polynomial
( s − μ1 ) ( s − μn ) = s n + α1s n−1 + α 2 s n−2 + + α n
and determine the αi’s
z The required state feedback gain matrix is
K = [(α n − an ) (α n −1 − an −1 ) (α1 − a1 )] T −1
Method 3 (Ackermann’s formula)
Define A = A − BK
desired characteristic equation is
sI − ( A − BK ) = ( s − μ1 ) ( s − μn )
| sI − A |= s n + α1s n −1 + α 2 s n −2 + + α n −1s + α n = 0
Caley-Hamilton theorem states that every matrix A
satisfies its own characteristic equation
φ ( A) = An + α1 An−1 + α 2 An−2 + + α n −1 A + α n = 0
For the case n = 3 consider the following identities.

Method 3: (Ackermann’s formula)
Multiplying the identities in order by α3, α2, α1 respectively and
adding we get
……….(1)
From Caley-Hamilton Theorem for A
And also we have for A

Substituting ϕ ( A) and ϕ ( A) in equation (1) we get
Since system is completely controllable inverse of the controllability

matrix exists we obtain
………(2)

Pre multiplying both sides of the equation (2) with [0 0 1]
z For an arbitrary positive integer n ( number of states)

Ackermann’s formula for the state feedback gain
matrix K is given by
−1
K = [0 0 0 1 ] ⎡⎣ B AB A B2
A B ⎤⎦
n −1
φ ( A)
w h e re φ ( A ) = A n + α 1 A n − 1 + + α n −1 A + α n I
α i ' s a re th e c o e ffic ie n ts o f th e
d e s ire d c h a ra c te ris tic p o ly n o m ia l
Choice of closed loop poles :
Guidelines
z Do not choose the closed loop poles far away
from the open loop poles, otherwise it will
demand high control effort
z Do not choose the closed loop poles very

negative, otherwise the system will be fast
reacting (i.e. it will have a small time constant)
• In frequency domain it leads to large bandwidth, and
hence noise gets amplified

Guidelines
z Use “Butterworth polynomials”
n +1
s n +1
⎛ ( j ( 2 k + 1) π ) ⎞
2n
( ) = ( − 1) 2n
= ⎜e ⎟ k = 0,1, 2 - - -
wo ⎝ −1 ⎠
w o = a constant (like " natural frequency")
n = system order (num ber of closed loop poles)
choose only stable poles.
Im
Example: 1
let n = 1 only one pole
use k = 1 -ω0
Re
s=ω0 (cos π + j sin π ) = −ω0
Guidelines
Example 2: S1
Im
Let n = 2 we know ( cos θ + j sin θ ) m =( cos mθ + j sin mθ )

n+1 3
=
2n 4 Im
Re
s = ω 0 [cos ( (2 k + 1)3π / 4 ) + j sin ( (2 k + 1)(3π / 4) )] S2
k = 0 ⇒ s1 = ω 0 [cos ( 3π / 4 ) + j sin ( (3π / 4) )]

stabilizing: acc ep t . Re
k = 1 ⇒ s 2 = ω 0 [cos ( 9π / 4 ) + j sin ( (9π / 4) )]

Re
Destabiliz i ng: reject.
S3 Im
k = 2 ⇒ s 3 = ω 0 [cos (15π / 4 ) + j sin ( (15π / 4) )]Im
Re
Destabilizing: reject.
S4
k = 3 ⇒ s 4 = ω 0 [cos ( 21π / 4 ) + j sin ( (21π / 4 ) )]
stabilizi ng: accept .

Example: Inverted Pendulum
⎡ 0 1 0 0⎤ ⎡0 ⎤
⎢ 20.601 0 ⎥⎥ ⎢ −1 ⎥
0 0 ⎡1 0 0 0 ⎤
A=⎢ ; B = ⎢ ⎥ ;C = ⎢ ⎥
⎢ 0 0 0 1⎥ ⎢0 ⎥ ⎣ 0 1 0 0 ⎦
⎢ ⎥ ⎢ ⎥
⎣ −0.4905 0 0 0⎦ ⎣0.5⎦
Step 1: Check controllability
Rank = 4 ; Controllable
M ≠0
Hence, the system is controllable.

Step 2: Form the characteristic equation and get ai’s

Step 3: Find Transformation T = MW and its inverse

Step 4: Find αi’s from desired poles μ1 , μ2 , μ3 , μ4

Step 5: Find State Feed back matrix K and input u

Time Simulation: Inverted Pendulum

Guidelines
z Do not choose the closed loop poles far away
from the open loop poles. Otherwise, it will
demand high control effort.
z Do not choose the closed loop poles very

negative. Otherwise, the system will be fast
reacting (i.e. it will have a small time constant)
In frequency domain it leads to a large bandwidth,
which in turn leads to amplification of noise!

Multiple input systems
z The gain matrix is not unique even for
fixed closed loop poles.
z Involved (but tractable) mathematics
X = AX + BX
U = − KX
⎡ k11 k12 k1 n ⎤
⎢ ⎥
K =⎢ ⎥
⎢ ⎥
⎢ ⎥
⎣ k m1 k m 2 k mn ⎦
Multiple input systems:
Some tricks and ideas
z Eliminate the need for measuring some xj by appropriately
choosing the closed loop poles.
E xam ple : u = μ 1 x1 + ( μ 2 − β ) x 2
select μ 2 = β provided β < 0
z Relate the gains to proper physical quantities
⎡ x⎤
⎡ux ⎤ ⎡ g11 0 g13 0 ⎤ ⎢⎢ y ⎥⎥
⎢u ⎥ = ⎢ ⎥ ⎢ x⎥
⎣ ⎦
y ⎣ 0 g22 0 g24 ⎦
⎢ ⎥
⎣ y⎦
z Shape eigenvectors: “Eigen structure assignment control”
z Introduce the idea of optimality: “optimal control”
Control Allocation:

References
z K. Ogata: Modern Control Engineering,

3rd Ed., Prentice Hall, 1999.
z B. Friedland: Control System Design,

McGraw Hill, 1986.


Pole Placement Control Design: Dr. Radhakant Padhi

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Pole Placement Control Design: Dr. Radhakant Padhi

Uploaded by

Copyright:

Available Formats

Lecture – 21

Pole Placement Control Design

Dr. Radhakant Padhi

 The system is completely state controllable.

 The state variables are measurable and are

 Control input is unconstrained.

ADVANCED CONTROL SYSTEM DESIGN 2

Difference from classical approach:

Necessary and sufficient condition:

ADVANCED CONTROL SYSTEM DESIGN 3

ADVANCED CONTROL SYSTEM DESIGN 4

The gain matrix K is designed in such a

ADVANCED CONTROL SYSTEM DESIGN 5

ADVANCED CONTROL SYSTEM DESIGN 6

Let the system be in first companion (controllable canonical) form

ADVANCED CONTROL SYSTEM DESIGN 7

ADVANCED CONTROL SYSTEM DESIGN 9

Comparing Equation (1) and (2), we arrive at:

ADVANCED CONTROL SYSTEM DESIGN 11

ADVANCED CONTROL SYSTEM DESIGN 14

From Caley-Hamilton Theorem for A

And also we have for A

ADVANCED CONTROL SYSTEM DESIGN 15

Since system is completely controllable inverse of the controllability

ADVANCED CONTROL SYSTEM DESIGN 16

z For an arbitrary positive integer n ( number of states)

z Do not choose the closed loop poles very

ADVANCED CONTROL SYSTEM DESIGN 18

Let n = 2 we know ( cos θ + j sin θ ) m =( cos mθ + j sin mθ )

s = ω 0 [cos ( (2 k + 1)3π / 4 ) + j sin ( (2 k + 1)(3π / 4) )] S2

k = 0 ⇒ s1 = ω 0 [cos ( 3π / 4 ) + j sin ( (3π / 4) )]

k = 1 ⇒ s 2 = ω 0 [cos ( 9π / 4 ) + j sin ( (9π / 4) )]

ADVANCED CONTROL SYSTEM DESIGN 20

ADVANCED CONTROL SYSTEM DESIGN 23

ADVANCED CONTROL SYSTEM DESIGN 24

ADVANCED CONTROL SYSTEM DESIGN 25

ADVANCED CONTROL SYSTEM DESIGN 26

ADVANCED CONTROL SYSTEM DESIGN 27

ADVANCED CONTROL SYSTEM DESIGN 28

z Do not choose the closed loop poles very

ADVANCED CONTROL SYSTEM DESIGN 29

ADVANCED CONTROL SYSTEM DESIGN 32

z K. Ogata: Modern Control Engineering,

z B. Friedland: Control System Design,

ADVANCED CONTROL SYSTEM DESIGN 33

You might also like

The system is completely state controllable.

The state variables are measurable and are

Control input is unconstrained.