1.4.3.

3 Design of Regulator Systems with Observer:

In this section we shall consider a problem of designing regulator systems by using the pole-
placement-with-observer approach.
Consider the regulator system shown in Figure 11. (The reference input is zero).
The plant transfer function is:
10(𝑠 + 2)
𝐺(𝑠) =
𝑠(𝑠 + 4)(𝑠 + 6)

Figure 11 Regulator System

Using the pole placement approach, design a controller . Assume that we use the minimum-
order observer. (We assume that only the output y is measurable).
We shall use the following design procedure:
1. Derive a state-space model of the plant.
2. Choose the desired closed-loop poles for pole placement. Choose the desired observer
poles.
3. Determine the state feedback gain matrix 𝒌 and the observer gain matrix 𝒌𝒆 .
4. Using the gain matrices 𝒌 and 𝒌𝒆 obtained in step 3, derive the transfer function of the
observer controller. it must be stable controller.
Design Step 1: the state space equation and the output equation of the given system is
obtained as:
𝑥̇ 1 0 1 0 𝑥1 0
[𝑥̇ 2 ] = [0 0 1 ] [ 𝑥2 ] + [ 10 ] 𝑢
𝑥̇ 3 0 −24 −10 𝑥3 −80
𝑥1
𝑦 = [1 0 0] [𝑥2 ] + [0]𝑢
𝑥3
Design Step 2: assume that the desired closed loop poles at:
𝑠1,2 = −1 ∓ 𝑗2, 𝑠3 = −5
Then, the sate feedback gain matrix is given as (using Ackermann’s Formula):
𝑘 = [1.25 1.25 0.19375]

33
Design Step 3: Assume that the desired closed loop poles for observer pole locations as
follows: 𝑠1,2 = −4.5
Then, the observer gain matrix is given as (using Ackermann’s Formula):
−1
𝑘𝑒 = [ ]
6.25
Design Step 3: We shall determine the transfer function of the observer controller from the
following equation:
𝑈(𝑠) 𝑛𝑢𝑚
𝐺𝑐 (𝑠) = = = − [𝑪 ̃ )−𝟏 𝑩
̃ (𝒔𝑰 − 𝑨 ̃+𝐷
̃]
−𝑌(𝑠) 𝑑𝑒𝑛
Therefore;
1.2109𝑠 2 + 11.2125𝑠 + 25.3125 1.2109(𝑠 + 5.3582)(𝑠 + 3.0012)
𝐺𝑐 (𝑠) = =
𝑠 2 + 6𝑠 + 2.1406 (𝑠 + 5.619)(𝑠 + 0.381)
As seen, the poles of the controller are located at lift hand side of s-plane; it means the
transfer function of the designed controller is stable.

1.5 Design of Control System with Observers:

In Section 1.4 we discussed the design of regulator systems with observers. (The systems did
not have reference or command inputs.) In this section we consider the design of control
systems with observers when the systems have reference inputs or command inputs.
In Section 1.4 we discussed regulator systems, whose block diagram is shown in Figure 11.
This system has no reference input, or r = 0. When the system has a reference input, several
different block diagram configurations are conceivable, each having an observer controller.
Two of these configurations are shown in Figures 12 (a) and (b); we shall consider them in
this section.

Figure 12 a) Control system with observer controller in the feed-forward path.

b) Control system with observer controller in the feed-back path.
34
Configuration 1: Consider the system shown in Figure 13. In this system the reference input
is simply added at the summing point. We would like to design the observer controller such
that in the unit-step response the maximum overshoot is less than 30% and the settling time is
about 5 sec.
In what follows we first design a regulator system. Then, using the observer controller
designed, we simply add the reference input r at the summing point.
Before we design the observer controller, we need to obtain a state-space representation of the
plant. Since
𝑌(𝑠) 1
= 2
𝑈(𝑠) 𝑠(𝑠 + 1)

Figure 13 Observer Controller in the feed-forward bath

The state equation of the given system is:

𝑥̇ 1 0 1 0 𝑥1 0 𝑥1
[𝑥̇ 2 ] = [0 0 1] [𝑥2 ] + [0] 𝑢; 𝑦 = [1 0 𝑥
0] [ 2 ]
𝑥̇ 3 0 −1 0 𝑥3 1 𝑥3
Assume the desired closed loop poles for pole placement at: 𝑠1,2 = −1 ∓ 𝑗; 𝑠3 = −8. And
the desired observer poles at: 𝑠1,2 = −4.
Using Ackermann’s Formula, the state feedback gain matrix 𝑘 and the observer gain matrix
𝑘𝑒 are given as follows:
8
𝑘 = [16 17 10]; 𝑘𝑒 = [ ]
15
The transfer function of the observer controller is obtained by use the following equation:
𝑈(𝑠) 𝑛𝑢𝑚
𝐺𝑐 (𝑠) = = = − [𝑪 ̃ )−𝟏 𝑩
̃ (𝒔𝑰 − 𝑨 ̃+𝐷
̃]
−𝑌(𝑠) 𝑑𝑒𝑛
Therefore;
302𝑠 2 + 303𝑠 + 256 302(𝑠 + 0.5017 + 𝑗0.5017)(𝑠 + 0.5017 − 𝑗0.5017)
𝐺𝑐 (𝑠) = =
𝑠 2 + 18𝑠 + 113 (𝑠 + 9 + 𝑗5.6569)(𝑠 + 9 − 𝑗5.6569)
The unit-step response curve for this control system is shown in Figure 14. The maximum
overshoot is about 28% and the settling time is about 4.5 sec. Thus, the designed system
satisfies the design requirements.

35
 Figure 14 Unit step input of the controlled system

Figure 15 shows the system without controller:

Figure 15 the response of the system without controller

The Simulink tools of MATLAB program has been used to plot figure 14 and 15 as shown in
Figure 16 below:

Figure 16
36
Configuration 2: A different configuration of the control system is shown in Figure 17. The
observer controller is placed in the feedback path. The input r is introduced into the closed-
loop system through the box with gain N. From this block diagram, the closed-loop transfer
function is obtained as

Figure 17 Control system with observer controller in the feedback

𝑌(𝑠) 𝑁(𝑠 2 + 18𝑠 + 113)

=
𝑅(𝑠) 𝑠(𝑠 2 + 1)(𝑠 2 + 18𝑠 + 113) + 302𝑠 2 + 303𝑠 + 256
We determine the value of constant N such that for a unit-step input r, the output y is unity as t
approaches infinity.
Note: if
𝑠 𝑚 + 𝑎1 𝑠 𝑚−1 + 𝑎2 𝑠 𝑚−2 + ⋯ + 𝑎𝑚−1 𝑠 + 𝑎𝑚
𝐺𝑐 (𝑠) =
𝑠 𝑛 + 𝑏1 𝑠 𝑛−1 + 𝑏2 𝑠 𝑛−2 + ⋯ + 𝑏𝑛−1 𝑠 + 𝑏𝑛
Then:
𝑎𝑚
𝑁=
𝑏𝑛
Therefore,
256
𝑁= = 2.2655
113
The unit-step response of the system is shown in Figure 18. Notice that the maximum
overshoot is very small; approximately 4%. The settling time is about 5 sec.

Figure 18 Unit step response

37
The Simulink tools of MATLAB program has been used to plot Figure 18 as shown in Figure
19 below:

Figure 19

1.6 Quadratic Optimal Regulator System:

An advantage of the quadratic optimal control method over the pole-placement method is that
the former provides a systematic way of computing the state feedback control gain matrix.
We shall now consider the optimal regulator problem that, given the system equation:
𝒙̇ = 𝑨𝒙 + 𝒃𝒖 (1.37)
determines the matrix 𝒌 of the optimal control vector:
𝑢(𝑡) = −𝒌𝒙(𝒕) (1.38)
so as to minimize the performance index:
∞

𝐽 = ∫ (𝒙𝑻 𝑸𝒙 + 𝒖𝑻 𝑹𝒖)𝑑𝑡 (1.39)

0

where Q is a positive-definite (or positive-semidefinite) Hermitian or real symmetric matrix

and R is a positive-definite Hermitian or real symmetric matrix.
As will be seen later, the linear control law given by Equation (1.38) is the optimal control
law. Therefore, if the unknown elements of the matrix 𝒌 are determined so as to minimize the
performance index, then 𝑢(𝑡) = −𝒌𝒙(𝒕) is optimal for any initial state x(0).
The design steps may be stated as follows:
1. Solve the following Equation for the matrix P. [If a positive-definite matrix P (𝑛 ×
𝑛 𝑚𝑎𝑡𝑟𝑖𝑥) exists (certain systems may not have a positive definite matrix P), the
system is stable, or matrix A - Bk is stable.].
𝑨𝑻 𝑷 + 𝑷𝑨 − 𝑷𝑩𝑹−𝟏 𝑩𝑻 𝑷 + 𝑸 = 𝟎 (1.40)

38