Week 9-10 Part2

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

ESTIMATOR DESIGN

Bushra Tahir
An Estimator Design for a Simple
Pendulum
• Compute estimator gain matrix that will place error poles
at -10ω0 (five times as fast as the controller poles -2ω0 ),
for ω0=1.
An Estimator Design for a Simple
Pendulum
• Solution:
• We are asked to place the two estimator error poles at
-10ω0. The corresponding characteristic equation is

• Estimator characteristics equation:

• Comparing coefficients:
An Estimator Design for a Simple
Pendulum
The response of this closed-loop system with wo = 1 to an initial condition
𝑥0 = [1,0]𝑇 and 𝑥ො0 = [0,0]𝑇

Initial-condition response of oscillator showing 𝑥 and 𝑥ො


An Estimator Design for a Simple
Pendulum
• Block Diagram of the setup (control law + estimator)
Estimator & Observer Canonical Form
• Observer Canonical form, the estimator gain design
equations are particularly simple and the existence of a
solution is obvious.
𝑥ሶ 𝑜 = 𝐴𝑜 𝑥𝑜 + 𝐵𝑜 𝑢
𝑦 = 𝐶𝑜 𝑥𝑜
• where,
Estimator & Observer Canonical Form
• A block diagram for the third-order case.
• In observer canonical form, all the feedback loops come
from the output, or observed signal.

Click to add text


Estimator & Observer Canonical Form
• Matrix 𝐴𝑜 is called a left companion matrix to the
characteristic equation because the coefficients of the
equation appear on the left side of the matrix.
• Advantage:
• Estimator gains can be obtained from it by inspection.
• The estimator error closed-loop matrix for the third-order case is

• which has the characteristics equation

𝑠 3 + 𝑎1 + 𝑙1 𝑠 2 + 𝑎2 + 𝑙2 𝑠 + 𝑎3 + 𝑙3 = 0
Estimator & Observer Canonical Form
• We can find a transformation to take a given system to observer
canonical form if and only if the system has a structural property that
in this case we call observability.

• Observability: ability to deduce information about all the modes of the


system by monitoring only the sensed outputs.

• Mathematical test for determining observability: observability matrix


be nonsingular or have a nonzero determinant.
Estimator & Observer Canonical Form
• As with control-law design, we could find the
transformation to observer form, compute the gains from
the equivalent characteristic equation in OCF, and
transform back.

• An alternative method of computing L is to use


Ackermann's formula in estimator form, which is
Duality
• Considerable resemblance between estimation and
control problems
• The two problems are mathematically equivalent.
• This property is called duality.

• Control problem: select K for satisfactory


pole placement of the system matrix A - BK;

• Estimator problem: select L for satisfactory


placement of the poles of A - LC.
Duality
• Poles of A − LC = poles of (A − LC)𝑇

• (A − LC)𝑇 = 𝐴𝑇 − 𝐶 𝑇 𝐿𝑇

• Therefore. where we used Ackermann's formula or the


place algorithm in the forms for the control problem
K=acker(A, B, pc), K=place(A, B, pc),
• we use
Lt=acker(A', C', pe), Lt=place(A‘ ,C', pe),
L=Lt',
• for estimator problem

You might also like