Chapter 2

Design via State Space

1
Contents
State Space Representation

Pole Placement Technique

Controllability

Observer Design

Observability
State Feedback with Integral
Control
2
Introduction
Analysis and design
of feedback control system

Frequency domain Time domain

(classical control) (modern control)

Transfer function State space equations

♥ Provide stability and ▬ limited to linear, LTI ♥ applicable to
𝑌(𝑠) transient response systems and its 𝑥ሶ = 𝐴𝑥 + 𝐵𝑢 nonlinear, MIMO, LPV ▬ not intuitive
𝐺 𝑠 =
𝑈(𝑠) immediately approximate 𝑦 = 𝐶𝑥 + 𝐷𝑢 systems.

3
State Space Representation
For a linear, continuous
The smallest set of linearly
𝑥1 ; state variable
independent system variables time –invariant
systems, the state
The 𝑛-dimensional space whose axes space equations is
are the state variables.
𝑥ሶ = 𝐴𝑥 + 𝐵𝑢
𝑥2 ; state variable
𝑦 = 𝐶𝑥 + 𝐷𝑢

A vector whose elements are the

state variables.
A set of points in state space that are future
states resulting from a given initial state

4
State Space Representation

State equations : A set of simultaneous, first order differential equations with

𝑛 state variables
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢

Output equation : The algebraic equation that expresses the output variables
of a system as linear combinations of the state variables and the inputs

𝑦 = 𝑪𝒙 + 𝑫𝑢

where 𝑥 is the state vector, 𝑥ሶ is the derivative of 𝑥 with respect to 𝑡, 𝑦 is the

output vector and 𝑢 is the input or control vector. 𝑨, 𝑩, 𝑪, 𝑫 are compatible
system, input, output and feedforward matrices, respectively.
5
State Space Representation

For a linear and SISO system, the dynamics is

𝒙ሶ = 𝑨𝒙 + 𝑩𝑢
𝑦 = 𝑪𝒙

The block diagram representation

Plant / 𝐺(𝑠)

6
State Space Representation
Converting from transfer function (frequency domain) to state space (time domain)

𝑌 𝑠
𝐺 𝑠 =
𝑈 𝑠

𝐵(𝑠) 𝑏𝑚 𝑠 𝑚 + 𝑏𝑚−1 𝑠 𝑚−1 + ⋯ + 𝑏1 𝑠1 + 𝑏0

G 𝑠 = =
𝐴(𝑠) 𝑠 𝑛 + 𝑎𝑛−1 𝑠 𝑛−1 + ⋯ + 𝑎1 𝑠1 + 𝑎0

where 𝐵(𝑠) is the numerator and 𝐴(𝑠) is denominator polynomials. Roots of

polynomial 𝐵(𝑠) are zero and A(𝑠) are poles of the system. Order of the system is
𝑛 and system is proper if 𝑛 ≥ 𝑚, strictly proper if 𝑛 > 𝑚
7
State space model from transfer function
Case 2 – Polynomial in the numerator 𝑥ሶ = 𝐴𝑥 + 𝐵𝑢
Consider a transfer function 𝑦 = 𝐶𝑥 + 𝐷𝑢
𝐶(𝑠) 𝑏2 𝑠 2 + 𝑏1 𝑠 + 𝑏0
G 𝑠 = = 3 2
⋯ ⋯ (12)
𝑅(𝑠) 𝑠 + 𝑎2 𝑠 + 𝑎1 𝑠 + 𝑎0
Step 1: Separate the system into two cascaded blocks

𝑅(𝑠) 𝑋1 (𝑠) 𝐶(𝑠)

1
𝑏2 𝑠2 + 𝑏1 𝑠 + 𝑏0
𝑠 3 + 𝑎2 𝑠 2 + 𝑎1 𝑠 + 𝑎0

Internal variables:
𝑋2 𝑠 , 𝑋3 (𝑠)

8
Step 2: Find the state equations for the block containing the denominator.

𝑅(𝑠) 𝑋1 (𝑠)
1
𝑠 3 + 𝑎2 𝑠 2 + 𝑎1 𝑠 + 𝑎0

Internal variables: 𝑋2 𝑠 , 𝑋3 (𝑠)

Step 2.1: Find the associated differential equation .

Cross-multiplying
𝑠 3 + 𝑎2 𝑠 2 + 𝑎1 𝑠 + 𝑎0 𝑋1 𝑠 = 𝑅(𝑠) ⋯ ⋯ (1)

and taking the inverse Laplace transform assuming zero initial conditions yields
𝑥ഺ1 + 𝑎2 𝑥ሷ1 + 𝑎1 𝑥ሶ1 + 𝑎0 𝑥1 = 𝑟 ⋯ ⋯ 2

Rearrange
ഺ𝑥1 = −𝑎2 𝑥ሷ 1 − 𝑎1 𝑥ሶ 1 − 𝑎0 𝑥1 + 𝑟 ⋯ ⋯ 3
9
Step 2.1 : Select the state variables. Choose the state variables as successive
derivatives of the output, that is
𝑥1 = 𝑥1 ⋯ ⋯ 4𝑎 ; 𝑥2 = 𝑥ሶ 1 ⋯ ⋯ 4𝑏 ; 𝑥3 = 𝑥ሷ 1 ⋯ ⋯ (4𝑐)

Differentiating both sides and equates with eq. (3) and eq. (4), the combined state
and output equation are
𝑥ሶ 1 = 𝑥ሶ1 = 𝑥2 ⋯ ⋯ 5𝑎
𝑥ሶ 2 = 𝑥ሷ 1 = 𝑥3 ⋯ ⋯ 5𝑏
𝑥ሶ 3 = 𝑥ഺ1 = −(𝑎2 𝑥3 + 𝑎1 𝑥2 + 𝑎0 𝑥1 ) + 𝑟 ⋯ ⋯ 5𝑐
𝑦 = 𝑥1 ⋯ ⋯ (5𝑑)

Note that the number of the state variables is equal to the highest derivatives.
11
Eqs. (5) in vector-matrix form, 𝑥ሶ = 𝐴𝑥 + 𝐵𝑢
𝑦 = 𝐶𝑥 + 𝐷𝑢
𝑥ሶ 1 0 1 0 𝑥1 0
𝑥ሶ 2 = 0 0 1 𝑥2 + 0 𝑟
𝑥ሶ 3 −𝑎0 −𝑎1 −𝑎2 𝑥3 1

𝑥1
𝑦 ′ = 1 0 0 𝑥2
𝑥3
Note that the third row of the system has the same coefficient as the denominator of
the transfer function but negative and in reverse order.
This representation is called phase-variable form where we choose the output
𝑦(𝑡) and its (𝑛 − 1) derivatives as the state variables.

12
Step 3: Introduce the effect of the block with the numerator.
𝑋1 (𝑠) 𝐶(𝑠)
𝑏2 𝑠 2 + 𝑏1 𝑠 + 𝑏0

The second block states that

𝐶 𝑠 = (𝑏2 𝑠 2 + 𝑏1 𝑠 + 𝑏0 )𝑋1 (𝑠)

Taking the inverse Laplace transform with zero initial conditions yields
𝑐 = 𝑏2 𝑥ሷ 1 + 𝑏1 𝑥ሶ 1 + 𝑏0 𝑥1 ⋯ ⋯ (6)

From eq. (5),

𝑥1 = 𝑥1 ; 𝑥ሶ 1 = 𝑥2 ; 𝑥ሷ 1 = 𝑥3
13
Hence eq. (6) becomes
𝑦 = 𝑐(𝑡) = 𝑏2 𝑥3 + 𝑏1 𝑥2 + 𝑏0 𝑥1 ⋯ ⋯ (8)
And generates the output equation
𝑥1
𝑦 = 𝑏0 𝑏1 𝑏2 𝑥2
𝑥3
The state space equation in phase-variable form is

𝑥ሶ 1 0 1 0 𝑥1 0
𝑥ሶ 2 = 0 0 1 𝑥2 + 0 𝑟
𝑥ሶ 3 −𝑎0 −𝑎1 −𝑎2 𝑥3 1
𝑥1
𝑦 = 𝑏0 𝑏1 𝑏2 𝑥2
𝑥3
where 𝑢 = 𝑟
14
Figure 6: The block diagram for the phase-variable form.
15
State Space Representation:
PHASE VARIABLE FORM
𝑏𝑚 𝑠 𝑚 + 𝑏𝑚−1 𝑠 𝑚−1 + ⋯ + 𝑏1 𝑠1 + 𝑏0
G 𝑠 =
𝑠 𝑛 + 𝑎𝑛−1 𝑠 𝑛−1 + ⋯ + 𝑎1 𝑠1 + 𝑎0

𝑥ሶ1 0 1 0 ⋯ 0 𝑥1 0
𝑥ሶ 2 0 0 1 ⋱ ⋮ 𝑥2 0
⋮ = ⋮ ⋮ ⋱ ⋱ 0 ⋮ + ⋮ 𝑢
𝑥ሶ 𝑛−1 0 0 ⋯ 0 1 𝑥𝑛−1 0
𝑥ሶ 𝑛 −𝑎0 −𝑎1 ⋯ −𝑎𝑛−2 −𝑎𝑛−1 𝑥𝑛 1
𝑥1
𝑥2
𝑦 = 𝑏0 𝑏1 ⋯ 𝑏𝑚−1 𝑏𝑚 ⋮
𝑥𝑛−1
𝑥𝑛
16
State Space Representation: The controller
canonical form is
CONTROLLER CANONICAL FORM obtained from the
phase-variable form
𝑏𝑚 𝑠 𝑚 + 𝑏𝑚−1 𝑠 𝑚−1 + ⋯ + 𝑏1 𝑠1 + 𝑏0 by rearranging the
G 𝑠 =
𝑠 𝑛 + 𝑎𝑛−1 𝑠 𝑛−1 + ⋯ + 𝑎1 𝑠1 + 𝑎0 last row as the first
row. Then rename
the state variables.
𝑥ሶ 1 −𝑎𝑛−1 −𝑎𝑛−2 ⋯ −𝑎1 −𝑎0 𝑥1 1
𝑥ሶ 2 1 0 ⋯ 0 0 𝑥2 0
⋮ = 0 1 ⋱ 0 0 ⋮ + ⋮ 𝑢
𝑥ሶ 𝑛−1 ⋮ ⋱ ⋱ ⋱ ⋮ 𝑥𝑛−1 0
𝑥ሶ 𝑛 0 0 ⋯ 1 0 𝑥𝑛 0

𝑥1
𝑥2
𝑦 = 𝑏𝑚 𝑏𝑚−1 ⋯ 𝑏1 𝑏0 ⋮
𝑥𝑛−1
𝑥𝑛 17
State Space Representation:
OBSERVER CANONICAL FORM
𝑏𝑚 𝑠 𝑚 + 𝑏𝑚−1 𝑠 𝑚−1 + ⋯ + 𝑏1 𝑠1 + 𝑏0
G 𝑠 =
𝑠 𝑛 + 𝑎𝑛−1 𝑠 𝑛−1 + ⋯ + 𝑎1 𝑠1 + 𝑎0

𝑥ሶ1 −𝑎𝑛−1 1 0 ⋯ 0 𝑥1 𝑏𝑚
𝑥ሶ 2 −𝑎𝑛−2 0 1 ⋱ ⋮ 𝑥2 𝑏𝑚−1
⋮ = ⋮ ⋮ ⋱ ⋱ 0 ⋮ + ⋮ 𝑢
𝑥ሶ 𝑛−1 −𝑎1 0 ⋯ 0 1 𝑥𝑛−1 𝑏1
𝑥ሶ 𝑛 −𝑎0 0 ⋯ 0 0 𝑥𝑛 𝑏0
𝑥1
𝑥2
𝑦= 1 0 ⋯ 0 0 ⋮
𝑥𝑛−1
𝑥𝑛
18
State Space Representation
• Phase variable form : The coefficients of the denominator of the
transfer function appears in the last row of the matrix in negative
and reverse order.

• Controller canonical form : The coefficients of the denominator

of the transfer function appears in the first row of the matrix in
negative and forward order.

• Observer canonical form : The coefficients of the denominator of

the transfer function appears in the first column of the matrix in
negative and forward order.
19
 State Space Representation
Example 1:
Represent the following transfer functions in Phase Variable Form, Controller
Canonical Form and Observer Canonical Form.
a)
24
G s = 3
s + 9s 2 + 26s + 24
b)
s 2 + 7s + 2
G s = 3
s + 9s 2 + 26s + 24
c)
5(𝑠 + 10)
𝑇 𝑠 = 3
𝑠 + 10𝑠 2 + 20𝑠 + 50
20
State Space Representation

For a linear and SISO system, the dynamics is

𝒙ሶ = 𝑨𝒙 + 𝑩𝑢
𝑦 = 𝑪𝒙
The poles and stability of the system Steady State Error:
are obtained by solving the
characteristic equation: 𝑒𝑠𝑠 ∞ = 1 + 𝐶𝐴−1 𝐵

𝑑𝑒𝑡 𝑠𝐼 − 𝐴 = 0

𝑠 𝑛 + 𝑎𝑛−1 𝑠 𝑛−1 + ⋯ + 𝑎1 𝑠1 + 𝑎0 = 0
(characteristic equation)
21
 State Space Representation

Example 2:
Determine the stability and steady state error of the control system
represented in the following state space equations.
a)
−2 1 0
𝑥ሶ = 𝑥+ 𝑢; 𝑦 = 1 1 𝑥
0 −5 1

b)
0 1 0 0
𝑥ሶ = 0 0 1 𝑥 + 0 𝑢; 𝑦 = 1000 100 0𝑥
0 −36 −15 1
22
Pole Placement Technique

Assumptions:
• The system is completely state controllable.
• The state variables are measurable and available for feedback
• Control input is unconstrained.

Objective:
To place all the closed-loop poles at the desired locations.

Necessary and Sufficient Condition:

The system is completely state controllable.
23
Pole Placement Technique

For a linear and SISO system, the dynamics in state space is

𝒙ሶ = 𝑨𝒙 + 𝑩𝑢
𝑦 = 𝑪𝒙

Plant / 𝐺(𝑠) 24
Phase variable representation
 Pole Placement Technique
Concept: Introduce additional parameters (𝐾) into the system to control the
location of closed loop poles.

The control input 𝑢 is designed in state feedback form

𝑢 = −𝑲𝒙 + r.
This yields a closed-loop system
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢
= (𝑨𝒙 + 𝑩 −𝑲𝒙 + 𝑟 )
= (𝑨 − 𝑩𝑲)𝒙 + 𝑩𝑟

𝑦 = 𝑪𝒙
Plant with state variable feedback 25
 Pole Placement Technique
Concept: Introduce additional parameters (𝐾) into the system to control the
location of closed loop poles.

𝒙ሶ = (𝑨 − 𝑩𝑲)𝒙 + 𝑩𝑟
𝑦 = 𝑪𝒙

𝑲 is the 𝑛 adjustable
feedback gains

𝑲 = 𝑘1 𝑘2 ⋯ 𝑘𝑛

26
Phase variable representation
Pole Placement Technique
Concept: Introduce additional parameters (𝐾) into the system to control the
location of closed loop poles.

𝒙ሶ = 𝑨 − 𝑩𝑲 𝒙 + 𝑩𝑟 = 𝑨𝑪𝑳 𝒙 + 𝑩𝒓

where 𝑨𝑪𝑳 ≜ 𝑨 − 𝑩𝑲

The poles of the closed-loop system are given by the characteristic equation of
the matrix 𝑨𝑪𝑳 ,
𝑑𝑒𝑡 𝑠𝐼 − 𝑨𝑪𝑳 = 0.

Comparing the terms of this equation with the desired characteristic equation
yield the values of the feedback matrix 𝐾 which force the closed-loop poles to
the desired locations. 27
 Pole Placement for Plants in Phase-Variable Form
Design concept: Equate the characteristic equation of a closed loop system, to a
desired characteristic equation and then find the values of the feedback gains, 𝑘𝑖 .

The CLOSED-LOOP system, 𝒙ሶ = (𝑨 − 𝑩𝑲)𝒙 + 𝑩𝑟 where

0 1 0 ⋯ 0
0 0 1 ⋱ ⋮
𝐴 − 𝐵𝐾 = ⋮ ⋮ ⋱ ⋱ 0
0 0 ⋯ 0 1
−(𝑎0 + 𝑘1 ) −(𝑎1 + 𝑘2 ) ⋯ −(𝑎𝑛−2 + 𝑘𝑛−1 ) −(𝑎𝑛−1 + 𝑘𝑛 )

The characteristic equation is

𝑑𝑒𝑡 𝑠𝐼 − (𝐴 − 𝐵𝐾) = 0
⟹ 𝑠 𝑛 + (𝑎𝑛−1 +𝑘𝑛 )𝑠 𝑛−1 + ⋯ + (𝑎1 +𝑘2 )𝑠1 + (𝑎0 +𝑘1 ) = 0
29
 Pole Placement for Plants in Phase-Variable Form
Design concept: Equate the characteristic equation of a closed loop system, to a
desired characteristic equation and then find the values of the feedback gains, 𝑘𝑖 .

The desired characteristic equation is obtain from the design requirements that give
the desired pole locations.

For the desired pole locations at 𝑝 = [𝑝1 ; 𝑝2 ; ⋯ ; 𝑝𝑛 ] where 𝑝𝑖 < 0 (on the LHP)
for a stable closed-loop system.

The characteristic equation is

𝑠 − 𝑝1 𝑠 − 𝑝2 ⋯ 𝑠 − 𝑝𝑛 = 0
⟹ 𝑠 𝑛 + 𝑑𝑛 𝑠 𝑛−1 + ⋯ + 𝑑2 𝑠1 + 𝑑1 = 0
where 𝑑𝑖 contains the pole location parameters.
30
 Pole Placement for Plants in Phase-Variable Form
Design concept: Equate the characteristic equation of a closed loop system, to a
desired characteristic equation and then find the values of the feedback gains, 𝑘𝑖 .

Setting the closed-loop characteristic equation equal to the desired characteristic

equation as
𝑠 𝑛−1 : 𝑎𝑛−1 + 𝑘𝑛 = 𝑑𝑛
⋮ : ⋮
𝑠1 : 𝑎1 + 𝑘2 = 𝑑2
𝑠 0 : 𝑎2 + 𝑘3 = 𝑑1

The unknown parameters are 𝑘𝑖 s.

31
 Pole Placement for Plants in Phase-Variable Form
To apply the pole placement methodology to plants represent in phase-variable
form, we take the following steps:
1. Represent the plant in phase-variable form.
2. Feed-back each phase variable to the input of the plant through a gain, 𝑘𝑖 .
3. Find the characteristic equation for the closed-loop system represented in Step
2.
4. Decide upon all closed-loop pole locations and determine an equivalent
characteristic equation.
5. Equate like coefficients of the characteristic equations from Step 3 and 4 and
solve for 𝑘𝑖 .

32
 Pole Placement for Plants in Phase-Variable Form
Example 3: Given the plant
20(𝑠 + 5)
𝐺 𝑠 =
𝑠 𝑠+1 𝑠+4

Design the phase-variable feedback gains to yield 9.5% overshoot and a settling time
of 0.74 second.

ans: 𝐾 = 413.1 132.08 10.9

s^2+ 2(zeta)(omega_n)s + omega_n^2
33
 Pole Placement for Plants in Phase-Variable Form
Example 4: For the plant
100(𝑠 + 10)
𝐺 𝑠 =
𝑠 𝑠 + 3 𝑠 + 12

Design the phase-variable feedback gains to yield 5% overshoot and a peak time of
0.3 second.

ans: 𝐾 = 2094 373.1 14.97

34
 Controllability
To control the pole location the closed-loop system, the control signal,
𝑢, can control the behaviour of each state variable in 𝑥.

*Parallel form signal flow graphs 35

 Controllability
Pole placement is a viable design technique only for systems that are
controllable.
Parallel form state space representation:
Controllability by inspection,
i.e., from the state equation
−𝑎1 0 0 1 • A system with distinct
𝑥ሶ = 0 −𝑎2 0 𝑥+ 1 𝑢 eigenvalues and a diagonal
0 0 −𝑎3 1 system matrix is controllable
if the input matrix 𝐵 does
Or not have any rows that are
𝑥ሶ1 = −𝑎1 𝑥1 + 𝑢 zero
𝑥ሶ 2 = −𝑎2 𝑥2 + 𝑢
𝑥ሶ 3 = −𝑎3 𝑥3 + 𝑢
36
 Controllability
Definition: If an input to a system can be found that takes every state
variable from a desired initial state to final state, the system is said to
be controllable; otherwise the system is said to be uncontrollable.

For system other than diagonal or parallel form with distinct

eigenvalues controllability by inspection are not valid.
For a plant under any representation whose state equation is Rank of a matrix
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 equals the number of
linearly independent
is completely controllable if the matrix rows or columns.
System has full rank if
𝑪𝑴 = 𝑩 𝑨𝑩 𝑨𝟐 𝑩 ⋯ 𝑨𝒏−𝟏 𝑩 det 𝐶𝑀 ≠ 0
is of rank 𝑛, where 𝐶𝑀 is called the controllability matrix. (nonsingular)

37
 Controllability
Example 5: Given the following system, represented in signal flow
diagram, determine its controllability.
𝒙ሶ 𝟐 𝒙ሶ 𝟏

𝒙ሶ 𝟑

38
 Controllability
Example 6*: Determine whether the system

−1 1 2 2
𝑥ሶ = 0 −1 5 𝑥+ 1 𝑢
0 3 −4 1
is controllable.

39
 Observer Design

• State feedback design relies upon access to state variables.

• Not all state are measurable, due to:
- State variables may not be available at all (system
configuration)
- Some sensors are expensive
• How to reconstruct all the states from a few measurement?
Design an OBSERVER / ESTIMATOR
to calculate state variables that are not accessible from the
plant.
40
Observer Design

• Plant: • Observer:
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 ෝሶ = 𝑨ෝ
𝒙 𝒙 + 𝑩𝑢 + 𝑳(𝑦 − 𝑦)
ො
𝑦 = 𝑪𝒙 𝑦ො = 𝑪ෝ
𝒙
41
 Observer Design

• Observer canonical form yields easy solution for the observer

gains.
• The design of observer is separate from the design of the
controller.
• The design of observer consists of evaluating the constant
vector, 𝐿 where transient response of the observer is faster
than the response of the controlled-loop in order to yield a
rapidly updated estimate of the state vector.

42
Observer Design
• Plant: • Observer:
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢 ෝሶ = 𝑨ෝ
𝒙 𝒙 + 𝑩𝑢 + 𝑳(𝑦 − 𝑦)ො
𝑦 = 𝑪𝒙 𝑦ො = 𝑪ෝ
𝒙
• Subtracting the plant equations from the observer equations yield
ෝሶ = 𝑨 𝒙 − 𝒙
𝒙ሶ − 𝒙 ෝ − 𝑳(𝑦 − 𝑦);
ො
the error between the actual state vector and the estimated state vector
ෝ ;
𝑦 − 𝑦ො = 𝑪 𝒙 − 𝒙
the error between the actual output and the estimated output
• Substitute the output equation into the state equation yield
ෝሶ = (𝑨 − 𝑳𝑪) 𝒙 − 𝒙
𝒙ሶ − 𝒙 ෝ ;
43
 Observer Design
• Letting the observer error, 𝒙
෥=𝒙−𝒙
ෝ, then
෥ሶ = 𝑨 − 𝑳𝑪 𝒙
𝒙 ෥; 𝑦 − 𝑦ො = 𝑪෥
𝒙
• The characteristic equation with the observer is
𝑑𝑒𝑡 𝑠𝑰 − (𝑨 − 𝑳𝑪) = 0

−𝑎𝑛−1 1 0 ⋯ 0 𝑙1
−𝑎𝑛−2 0 1 ⋱ ⋮ 𝑙2
where 𝑨 − 𝑳𝑪 = ⋮ ⋮ ⋯ ⋱ 0 − ⋮ 1 0 0 ⋯ 0
−𝑎1 0 ⋯ 0 1 𝑙𝑛−1
−𝑎0 0 ⋯ 0 0 𝑙𝑛
Note that, the state space is represented in Observer Canonical Form 44
 Observer Design
−(𝑎𝑛−1 + 𝑙1 ) 1 0 ⋯ 0
−(𝑎𝑛−2 + 𝑙2 ) 0 1 ⋱ ⋮
𝑨 − 𝑳𝑪 = ⋮ ⋮ ⋯ ⋱ 0
−(𝑎1 +𝑙𝑛−1 ) 0 ⋯ 0 1
−(𝑎0 + 𝑙𝑛 ) 0 ⋯ 0 0

• The characteristic equation with the observer, 𝑑𝑒𝑡 𝑠𝑰 − (𝑨 − 𝑳𝑪) = 0

𝑠 𝑛 + (𝑎𝑛−1 +𝑙1 )𝑠 𝑛−1 + (𝑎𝑛−2 +𝑙2 )𝑠 𝑛−2 + ⋯ + (𝑎1 +𝑙𝑛−1 ) + (𝑎0 +𝑙𝑛 ) = 0

45
 Observer Design
• The characteristic equation of the plant (desired)

𝑠 𝑛 + 𝑑𝑛−1 𝑠 𝑛−1 + 𝑑𝑛−2 𝑠 𝑛−2 + ⋯ + 𝑑1 𝑠1 + 𝑑0 = 0

• Solve for 𝑙𝑖 ’s by equating the coefficients of the observer and desired

characteristic equations,

𝑙𝑖 = 𝑑𝑛−𝑖 − 𝑎𝑛−𝑖
for 𝑖 = 1, 2, ⋯ , 𝑛

46
Observer Design

PLANT

OBSERVER

CONTROLLER
47
 Observer Design
Example 7: Design an observer for the plant
𝑠+4
𝐺 𝑠 =
(𝑠+1)(𝑠+2)(𝑠+5)
which is represented in observer canonical form. The observer will
respond 10 times faster than the controlled loop designed for 20.8%
overshoot and a settling time of 4 seconds.

48
 Observer Design
Example 8: Design an observer for the plant
𝑠+6
𝐺 𝑠 =
(𝑠+7)(𝑠+8)(𝑠+9)
operating with 20% overshoot and a settling time of 2 seconds. Design
the observer to respond 10 times faster than the plant.
Assume the plant is represented in observer canonical form.

49
 Observability
To evaluate a state variable from the output of a system, the output
signal, y, can observe the effect of each state variable in 𝑥.

50
 Observability
Pole placement for an observer is a viable design technique only for
systems that are observable.

The output equation for diagonal Observability by inspection,

system: i.e., from the output equation
• A system with distinct
𝑦 = 𝑪𝒙 = 1 1 1 𝒙 (observable) eigenvalues and a diagonal
system matrix is observable
And if the output coupling matrix
𝐶 does not have any
𝑦 = 𝑪𝒙 = 0 1 1 𝒙 (unobservable) columns that are zero

51
 Observability
Definition: If the initial state vector, 𝑥(𝑡𝑜 ) can be found from 𝑢(𝑡) and
𝑦(𝑡) measured over a finite interval of time 𝑡𝑜 , the system is said to be
observable; otherwise the system is said to be unobservable.

For system other than diagonal or parallel form with distinct eigenvalues observability by
inspection are not valid.
For a plant under any representation whose state and output equations are, respectively
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢
𝑦 = 𝑪𝒙
is completely observable if the matrix
𝑶𝑴 = 𝑪 𝑪𝑨 𝑪𝑨𝟐 ⋯ 𝑪𝑨𝒏−𝟏 𝑻
is of rank 𝑛, where 𝑂𝑀 is called the observability matrix.
52
 Observability
Example 9: Determine if the system below is observable.

53
 Observability
Example 10: Determine if the system below is observable.

54
 Observability
Example 11: Determine whether the system

−2 −1 −3 2
𝑥ሶ = 0 −2 1 𝑥+ 1 𝑢
−7 −8 −9 2

𝑦= 4 6 8𝑥

is observable.

55
 State Feedback with Integral Control
Design objective: To design a system with zero steady-state error for a
step input and design the desired transient response.

INTEGRAL CONTROL

PLANT WITH STATE

FEEDBACK CONTROLLER

56
 State Feedback with Integral Control
Integral Control increases the system type
and reduces the previous finite error to
zero.
• The output, 𝑦 is feedback and compared
with the reference input, 𝑟 to form the
error, 𝑒, which is fed forward to the
controlled plant via an integrator.
• An additional state variable, 𝑥𝑁 has been
added at the output of the leftmost
integrator. The error is the derivative of
this variable.
57
 State Feedback with Integral Control
The error
𝑒 = 𝑥ሶ 𝑁 = 𝑟 − 𝑪𝒙
The state equations
𝒙ሶ = 𝑨𝒙 + 𝑩𝑢
𝑥ሶ 𝑁 = −𝑪𝒙 + 𝑟
𝑦 = 𝑪𝒙
can be written in augmented vectors and
matrices,
𝒙ሶ 𝑨 0 𝒙 𝑩 𝟎
= + 𝑢 + 𝑟
𝑥ሶ 𝑁 −𝑪 0 𝑥𝑁 0 1
𝒙
𝑦= 𝑪 0 𝑥
𝑁

58
 State Feedback with Integral Control
But the control input
𝒙
𝑢 = −𝑲𝑥 + 𝐾𝑒 𝑥𝑁 = − 𝑲 −𝐾𝑒 𝑥𝑁
Hence,
𝑥ሶ (𝑨 − 𝑩𝑲) 𝑩𝐾𝑒 𝒙 𝟎
= 𝑥 + 𝑟
𝑥ሶ 𝑁 −𝑪 0 𝑁 1
𝒙
𝑦= 𝑪 0 𝑥
𝑁
Thus, system type has been increased. To
design 𝑲 and 𝐾𝑒 for the desired transient
response, we can use the characteristic
equation for the above state space
equation.
59
 State Feedback with Integral Control
Example 12: Consider the plant
0 1 0
𝒙ሶ = 𝒙+ 𝑢
−3 −5 1
𝑦= 1 0𝒙
a. Design a controller without integral control to yield a 10% overshoot and
a settling time of 0.5 second. Evaluate the steady-state error for a unit
step input.
b. Repeat the design of (a) using integral control. Evaluate the steady-state
error for a unit step input.
60
 State Feedback with Integral Control

Example 13: Design an integral controller for the plant

0 1 0
𝒙ሶ = 𝒙+ 𝑢
−7 −9 1
𝑦= 4 1𝒙
to yield a step response with 10% overshoot, a peak time of 2 seconds
and zero steady-state error.

Ans: 𝑲 = 2.21 −2.7 , 𝐾𝑒 = 3.79

61
 State Feedback with Integral Control

Example 14: Design a state feedback with integral control for the plant
−1 6 1
𝒙ሶ = 𝒙+ 𝑢
1 0 0
𝑦= 0 1𝒙
to assign the closed-loop poles at -5, -8 and -10. Then show that the
steady state error is zero for a unit step input.

62
 State space design
• Specify the system’s desired
Advantages Limitation State pole location
• All poles location • Inability to Feedback • Design controller with state -
variable feedback gains
can be specified design the open-
• Estimate of loop or closed-
system variables loop zeros
can be used for • To match the plant and estimate
inaccessible the system variable
variables via Observer • The response of the observer is
observer designed to be faster than the
controller.

Controllability : The ability of the control

signals to control the behavior of each state
• Addition of state variable
variable by manipulating the inputs.
Integral increases the system type
Observability : The ability of the system to
measure the state variables from the
Control • Improvement in steady state
error
outputs. 63
Observer Design

PLANT

OBSERVER

CONTROLLER
64
OBSERVER

INTEGRAL CONTROL

65
Example: Observer

66