5/19/2024

1.0 STATE VARIABLE METHODS

1.5.3. Transfer Function from State Model
For controllable form,
Example 1.13: Obtain the transfer function for the system,
𝑥̇ 0 1 0 𝑥 0
𝑥̇ = 0 0 1 𝑥 + 0 𝑢 𝑡
𝑥̇ −2 −4 −5 𝑥 2
𝑥
𝑦(𝑡) = 1 0 0 𝑥
𝑥
Solution
( )
The transfer function is of the form, for 𝑛 = 3 𝑇 𝑠 = =
( )
From the matrix 𝐀, 𝐁 and 𝐂 we have
𝑎 = 5, 𝑎 = 4, 𝑎 = 2, 𝑏 = 0, 𝑏 = 0, 𝑏 = 0, 𝑏 = 2
Hence
𝑌(𝑠) 2
𝑇 𝑠 = =
𝑈(𝑠) 𝑠 + 5𝑠 + 4𝑠 + 2
EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 1

1.0 STATE VARIABLE METHODS

1.5.3. Transfer Function from State Model
If the state model is in Jordan form, the transfer function can be obtained as in Example 1.14.
Example 1.14
Obtain the transfer function for the system,
𝑥̇ −1 1 0 𝑥 0
𝑥̇ = 0 −1 0 𝑥 + 0 𝑢 𝑡
𝑥̇ 0 0 −2 𝑥 1
𝑥
𝑦(𝑡) = −1 3 3 𝑥
𝑥
Solution
From the matrix 𝐀, the three poles of the transfer function are −1, −1 and−2. The residues at
these poles are −1, 3 and 3 (from the matrix 𝑪). The transfer function is therefore given by

−1 3 3 2𝑠 + 6𝑠 + 1
𝑇 𝑠 = + + =
𝑠+1 𝑠+1 (𝑠 + 2) 𝑠 + 1 (𝑠 + 2)

EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 2

1
 5/19/2024

1.0 STATE VARIABLE METHODS

1.5.3. Transfer Function from State Model
If the matrix A is not in either companion form or Jordan form, the transfer
function can be derived as follows.
Let the transfer function be given by
𝑌(𝑠)
𝐺 𝑠 =
𝑈(𝑠)
The general state space representation of any given systems is
𝑥̇ = 𝐴𝑥 + 𝐵𝑢 (1.25 𝑎)

𝑦 = 𝐶𝑥 + 𝐷𝑢 (1.25 𝑏)

Taking the Laplace transforms of (1.25a) and (1.25b)

𝑠𝑋 𝑠 − 𝑋 0 = 𝐴𝑋 𝑠 + 𝐵𝑈 𝑠 (1.26𝑎)
𝑌 𝑠 = 𝐶𝑋 𝑠 + 𝐷𝑈 𝑠 (1.26𝑏)
EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 3

1.0 STATE VARIABLE METHODS

1.5.3. Transfer Function from State Model
Assuming zero initial conditions of the system i.e. 𝑋 0 = 0 and rearranging (1.26a) then

𝑋 𝑠 = 𝑆𝐼 − 𝐴 𝐵𝑈 𝑠 (1.27)
Substituting (1.27) into (1.26b) yields

𝑌 𝑠 = 𝐶 𝑆𝐼 − 𝐴 𝐵+𝐷 𝑈 𝑠 (1.28)
and the transfer function is obtained as

𝑌(𝑠)
𝐺 𝑠 = = 𝐶(𝑆𝐼 − 𝐴) 𝐵+𝐷 (1.29)
𝑈(𝑠)
The transfer function representation of a given system is unique and thus Eqn (1.29) is independent of the form of A. This
implies that despite different system matrices that can be obtained for a given system; there will always be a unique transfer
function. Since
Adj(SI − A)
(𝑆𝐼 − 𝐴) =
SI − A
the roots of SI − A = 0 (1.30)
are the poles of the transfer function T(s) and eqn. (1.30) is known as the characteristic equation of the matrix A.

EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 4

2
 5/19/2024

1.0 STATE VARIABLE METHODS

1.5.3. Transfer Function from State Model
Example 1.15: Obtain the transfer function of the system whose state space representation is given as
0 1 𝑥 0
𝑥̇ = + u
−2 −3 𝑥 1

𝑥
𝑦= 1 0 𝑥
Solution
( )
𝐺 𝑠 = = 𝐶(𝑆𝐼 − 𝐴) 𝐵 + 𝐷 where
( )
0
1 0
𝐴= ,𝐵 = , 𝐶 = 1 0 and 𝐷 = 0
−2
−3 1
𝑠+3 1
(𝑆𝐼 − 𝐴) =
−2 𝑠
𝑠+3 1 0
𝐶(𝑆𝐼 − 𝐴) 𝐵 + 𝐷 = 1 0
−2 𝑠 1
=
EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 5

1.0 STATE VARIABLE METHODS

1.5.3. Transfer Function from State Model
Example 1.16: Obtain the transfer function for the system,
𝑥̇ −1 1 0 𝑥 0
𝑥̇ = 0 −1 0 𝑥 + 1 𝑢 𝑡
𝑥̇ 1 −2 −3 𝑥 1
𝑥
𝑦(𝑡) = 1 0 1 𝑥
𝑥
Solution
𝑌(𝑠)
𝐺 𝑠 = = 𝐶(𝑆𝐼 − 𝐴) 𝐵+𝐷
𝑈(𝑠)

𝑠+1 0 1
𝑆𝐼 − 𝐴 = 0 𝑠 + 1 −1
−1 2 𝑠+3
𝑠 + 4𝑠 + 5 2 −(𝑠 + 1)
(𝑆𝐼 − 𝐴) = 1 𝑠 + 4𝑠 + 4 𝑠+1
𝑠+1 −2(𝑠 + 1) (𝑠 + 1)
EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 6

3
 5/19/2024

1.0 STATE VARIABLE METHODS

1.5.3. Transfer Function from State Model
Example 1.16:
Solution
The transfer function is given by
1 𝑠 + 4𝑠 + 5 2 −(𝑠 + 1) 0
𝑇 𝑠 = 1 0 1 1 𝑠 + 4𝑠 + 4 𝑠+1 1
∆(𝑠)
𝑠+1 −2(𝑠 + 1) (𝑠 + 1) 1

where ∆ 𝑠 = 𝑠 + 5𝑠 + 10𝑠 + 6

1 1−𝑠 0 𝑠(𝑠 − 1)
𝑇 𝑠 = 1 0 1 𝑠 + 4𝑠 + 4 1 =
∆(𝑠) (𝑠 + 1)(𝑠 − 1) 1 𝑠 + 5𝑠 + 10𝑠 + 6

EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 7

1.0 STATE VARIABLE METHODS

1.5.3. Transfer Function from State Model

Practice Exercise 1.4

Consider the system defined by
𝑥̇ = 𝑨𝑥 + 𝑩𝑢
𝑦 = 𝑪𝑥

−1 0 1 0
Where 𝑨 =
1 −2 0 𝑩 = 0 𝑎𝑛𝑑 𝑪 = 1 1 0
0 0 −3 1
Transform the system equation into
i. Controllable canonical form
ii. Observable canonical form
iii.Diagonal canonical form
EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 8

4
 5/19/2024

1.0 STATE VARIABLE METHODS

1.6. Similarity Transformation and Diagonalization
1.6.1. Similarity Transformation
It has been stated that the choice of states is not unique for a given system. Suppose that there exists
a set of state variables

𝑋= 𝑥 𝑥 . . . 𝑥 (1.31)
We may take another set of state variables

𝑍= 𝑧 𝑧 . . . 𝑧 (1.32)
So that a linear or similarity transformation exists
Let
𝑋 = 𝑃𝑍 𝑖. 𝑒 𝑍 = 𝑃 𝑋 (1.33)

where 𝑃 is a non-singular transformational matrix

EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 9

1.0 STATE VARIABLE METHODS

1.6. Similarity Transformation and Diagonalization
1.6.1. Similarity Transformation
Differentiating equation (1.27) yields
𝑍̇ = 𝑃 𝑋̇ (1.34)
Using the general state space equations
𝑋̇ = 𝐴𝑥 + 𝐵𝑢
𝑦 = 𝐶𝑥
Then equation (1.34) becomes
𝑍̇ = 𝑃 𝐴𝑥 + 𝑃 𝐵𝑢 (1.35)
From equation (1.27) 𝑋 = 𝑃𝑍
Equation (1.35) becomes
𝑍̇ = 𝑃 𝐴𝑃𝑧 + 𝑃 𝐵𝑢 (1.36)
And 𝑦 = 𝐶𝑥 = 𝐶𝑃𝑧 (1.37)

EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 10

5
 5/19/2024

1.0 STATE VARIABLE METHODS

1.6. Similarity Transformation and Diagonalization
1.6.1. Similarity Transformation
Equations (1.30) and (1.31) can be written as
𝑍̇ = 𝐴𝑧 + 𝐵 𝑢 (1.38𝑎)
𝑦 = 𝐶𝑥 (1.38𝑏)
where 𝐴 = 𝑃 𝐴𝑃, 𝐵 = 𝑃 𝐵 𝑎𝑛𝑑 𝐶 = 𝐶𝑃
Hence, under similarity transformation the transformed system can be represented in the vector-
matrix differential form as
𝑍̇ = 𝐴𝑧 + 𝐵 𝑢 and the output as
𝑦 = 𝐶𝑥
NB
1. The characteristic equations and hence the Eigen values of 𝐴 and 𝐴 are invariant under similarity
transformation
2. The transfer function remains invariant under similarity transformation.

EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 11

1.0 STATE VARIABLE METHODS

1.6. Similarity Transformation and Diagonalization
1.6.1. Similarity Transformation
Example 1.17
Using the principle of similarity transformations, deduce the expression for the state equation given
that:

𝑥̇ −3 1 𝑥 0
= + 𝑢(𝑡)
𝑥̇ −2 0 𝑥 1
𝑥 1 1
𝑦 𝑡 = [0 1] 𝑥 with the transform P =
1 2
Solution
Under similarity transformation the transformed system can be represented in the vector-matrix
differential form as
𝑍̇ = 𝐴𝑧 + 𝐵 𝑢 and the output as
𝑦 = 𝐶𝑥
where 𝐴 = 𝑃 𝐴𝑃, 𝐵 = 𝑃 𝐵 𝑎𝑛𝑑 𝐶 = 𝐶𝑃
EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 12

6
 5/19/2024

1.0 STATE VARIABLE METHODS

1.6.1. Similarity Transformation
Example 1.17: Solution Cont
−3 1 0 1 1 2 −1
Now, 𝐴 = ,𝐵 = , 𝐶 = [0 1], P = and therefore 𝑃 =
−2 0 1 1 2 −1 1

1 −3 1 2 −1 1 −11 6
𝐴=𝑃 𝐴𝑃 = =
2 −2 0 −1 1 1 −15 8
1 1 0 1 2 −1
𝐵=𝑃 𝐵 = = and 𝐶 = 𝐶𝑃 = 0 1 = −1 1
1 2 1 2 −1 1
Thus the equations for the transformed system become:
−11 6 𝑧 1
𝑍̇ = + 𝑢
−15 8 𝑧 2
𝑥
𝑦 = −1 1 𝑥
It is necessary to select the matrix 𝐏 such that 𝐏 𝟏 𝐀𝐏 is a Jordan matrix.
This requires knowledge
5/19/2024
of Engineering
EEE 2502 Control Eigenvalues
III and Eigenvectors.
Lecture Notes by Dr. Linus. A. Aloo
13

1.0 STATE VARIABLE METHODS

1.6.2. Characteristic Equation, Eigen Values and Eigen Vectors.

 Consider the matrix equation

𝐴𝑥 = 𝜆𝑥 (1.39)
 The values of the scalar 𝜆 for which non-trivial solutions exist are called Eigen
values and the corresponding solutions 𝑥 = 0 are called Eigen vectors.

 Equation (1.39) can be written in the form 𝜆𝐼 − 𝐴 𝑥 = 0 where I is the identity

matrix 𝜆𝐼 − 𝐴 = 0 is the characteristic equation of A.

 The roots of the characteristic equation are called Eigen values of the matrix A.

 Corresponding to each Eigen value is a non-zero solution of 𝑥=𝑒 . This is called

the Eigen vector of A corresponding to 𝜆 .
EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 14

7
 5/19/2024

1.0 STATE VARIABLE METHODS

1.6.3. Diagonalization
Consider 𝑛 × 𝑛 matrix 𝑨 having distinct eigenvalues given by

0 1 0 . . 0
0 0 1 . . 0
. . . .
𝐀= (1.40)
. . . .
0 0 0 . . 1
−𝑎 −𝑎 −𝑎 . . −𝑎
the transformation 𝑥 = 𝑃𝑧, where
1 1 ... 1
𝜆 𝜆 ... 𝜆
𝐏= 𝜆 𝜆 … 𝜆 (1.41)
. . . .
𝜆 𝜆 . 𝜆
𝜆 , 𝜆 , … . , 𝜆 = 𝑛 distinct eigenvalues of 𝐀 will transform 𝐏 𝟏 𝐀𝐏 into the diagonal matrix, or
𝜆 0
𝟏 𝜆
𝐏 𝐀𝐏 = (1.42)
.
0 𝜆
EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 15

1.0 STATE VARIABLE METHODS

1.6.3. Diagonalization
 If the matrix 𝐀 specified by Eqn (1.40), entails multiple eigenvalues, then diagonalization is
impossible.
 To illustrate this, consider a 3 × 3 matrix 𝐀, where
0 1 0
𝐀= 0 0 1
−𝑎 −𝑎 −𝑎
has the eigenvalues 𝜆 , 𝜆 , 𝜆 , then the transformation 𝐱 = 𝐒𝐳, where

1 0 1
𝐒= 𝜆 1 𝜆
𝜆 2𝜆 𝜆
will yield
𝜆 1 0
𝐒 𝟏 𝐀𝐒 = 0 𝜆 0
0 0 𝜆
This is in the Jordan canonical form.
NB: The modal matrix P, shown in Eqn (1.41) is known as Vander
EEE 2502 Control Engineering III
Monde matrix.
Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 16

8
 5/19/2024

1.0 STATE VARIABLE METHODS

Example 1.18
Consider the system whose state model is defined by

𝑥̇ 0 4 1 𝑥
= 𝑢(𝑡) +
𝑥̇ 1 3 2 𝑥
𝑥
𝑦 𝑡 = [0 1] 𝑥
i. Determine the Eigen values and Eigen vectors of the system.
ii. Obtain a transformation matrix 𝑃 and show that under similarity transformation the characteristic
equation for the system is invariant.
iii. Write state space representation for the transformed system.
Solution
𝜆𝐼 − 𝐴 = 0 → 𝜆 − 4 𝜆 − 2 − 3 = 0 → 𝜆 = 5 , 𝜆 = 1
For 𝜆 = 5, 𝐴𝑥 = 𝜆𝑥 becomes
4 1 𝑥 𝑥
= 5 𝑥 , 4𝑥 + 𝑥 = 5𝑥
3 2 𝑥
3𝑥 + 2𝑥 = 5𝑥
𝑥 =𝑥
𝑥 1
Eigen vector corresponding to 𝜆 = 5 is 𝑒 = 𝑥 = simplest form
1
EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 17

1.0 STATE VARIABLE METHODS

Example 1.18
For 𝜆 = 1, 𝐴𝑥 = 𝜆𝑥 becomes
4 1 𝑥 𝑥
= 1 𝑥 , 4𝑥 + 𝑥 = 𝑥
3 2 𝑥
3𝑥 + 2𝑥 = 𝑥
𝑥 = −3𝑥
𝑥 1
Eigen vector corresponding to 𝜆 =1 is 𝑒 = 𝑥 = simplest form
−3

We obtain the transformation matrix 𝑃 from the Eigen vectors as follows

1 1
𝑃= 𝑒 𝑒 =
1 −3
The matrix 𝐴 = 𝑃 𝐴𝑃 is then obtained
1 1 4 1 1 1 5 0
𝐴 = 𝑃 𝐴𝑃 = =
1 −3 3 2 1 −3 0 1

Under similarity transformation the characteristic equation does not change i.e.

𝜆𝐼 − 𝐴 = 𝜆𝐼 − 𝐴 = 𝜆 + 𝜆𝑠 + 5
 There some instances where the Eigen values of matrix A are repeated.
 The Eigen vectors are evaluated as illustrated in the next example.

EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 18

9
 5/19/2024

1.0 STATE VARIABLE METHODS

Example 1.19
3 −3 2
Determine the Eigen values and Eigen vectors of 𝐴𝑥 = 𝜆𝑥 where 𝐴 = −1 5 −2
−1 3 0
Solution
𝜆−3 3 −2
𝜆𝐼 − 𝐴 = 0 → 1 𝜆 − 5 2 = 0 → 𝜆 − 8𝜆 + 20𝜆 − 16 = 0
1 −3 𝜆

𝜆 = 4, 𝜆 =𝜆 =2
For 𝜆 = 4
3 −3 2 𝑥 𝑥
−1 5 −2 𝑥 =4 𝑥
−1 3 0 𝑥 𝑥
This yield 𝑥 =−𝑥 and 𝑥 = −𝑥 and the simplest form of the corresponding Eigen vector as
1
𝑒 = −1
−1
For 𝜆 = 𝜆 = 2
3 −3 2 𝑥 𝑥
−1 5 −2 𝑥 =2 𝑥
−1 3 0 𝑥 𝑥

EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 19

1.0 STATE VARIABLE METHODS

Example 1.19
The simultaneous equations obtained are 𝑥 − 3𝑥 + 2𝑥 = 0. We let 𝑥 = 𝛼 and 𝑥 = 𝛽 where are 𝛼 and 𝛽 constants

The above equation becomes 𝑥 − 3𝛼 + 2𝛽 = 0 → 𝑥 = 3𝛼 − 2𝛽 and the resulting Eigen vector as

3 −2
𝑒=𝛼 1 +𝛽 0
0 1

3 −2
For 𝛽 = 0 𝑎𝑛𝑑 𝛼 = 1 , 𝑒 = 1 and for 𝛽 = 1 𝑎𝑛𝑑 𝛼 = 0 , 𝑒 = 0
0 1

1 3 −2
The transformation matrix 𝑃 is obtained as 𝑃 = −1 1 0
−1 0 1

The matrix 𝐴 = 𝑃 𝐴𝑃 is then obtained as

1 3 −2 3 −3 2 1 3 −2 4 0 0
𝐴=𝑃 𝐴𝑃 = −1 1 0 −1 5 −2 −1 1 0 = 0 2 0
−1 0 1 −1 3 0 −1 0 1 0 0 2

Checking the characteristic equation

𝜆𝐼 − 𝐴 = 𝜆𝐼 − 𝐴 = 𝜆 − 8𝜆 + 20𝜆 − 16 = 0

EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 20

10
 5/19/2024

1.0 STATE VARIABLE METHODS

Example 1.20
Consider the following state-space representation of a system.
𝑥̇ 0 1 0 𝑥 0
𝑥̇ = 0 0 1 𝑥 + 0 𝑢(𝑡)
𝑥̇ −6 −11 −6 𝑥 6
𝑥
0 𝑥
𝑦 = 1 0
𝑥
i. Determine the eigenvalues and eigenvectors of the given system
ii. Obtain a similarity transformation matrix 𝑃, its inverse and elements of diagonal matrix if this exists.
iii. Use transformation matrix 𝑃 to transform the system into a new set of state space representation in state
variable 𝑧. Hence, show that the characteristic equation does not change.
Solution
The given equations can be put into standard format:

ẋ = 𝐀x + 𝐁u
y = 𝐂x
0 1 0 0
Where 𝑨 = 0 0 1 𝑩 = 0 and 𝑪 = 1 0 0
−6 −11 −6 6
EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 21

1.0 STATE VARIABLE METHODS

(i) Using the steps shown in example, 1.19, the eigenvalues and eigenvectors of matrix 𝐀 can be obtained as
Eigenvalues: 𝜆 = −1, 𝜆 = −2, 𝜆 = −3
Eigen vector as
1 1 1
𝑒 = −1 , 𝑒 = −2 , 𝑒 = −3
1 4 9
(ii) Noting that the three eigenvalues are distinct, let’s define a set of new state variables 𝑧 , 𝑧 and 𝑧 by the
transformation 𝐱 = 𝐏𝐳 i.e.
𝑥 1 1 1 𝑧
𝑥 = −1 −2 −3 𝑧
𝑥 1 4 9 𝑧
where
1 1 1 1 1 1
𝑃= 𝑒 𝑒 𝑒 = 𝜆 𝜆 𝜆 = −1 −2 −3
𝜆 𝜆 𝜆 1 4 9
The inverse of P (𝑃 ) can be obtained as:
3 2.5 0.5
𝑃 = −3 −4 −1
1 1.5 0.5
Under similarity transformation the transformed system can be represented in the vector-matrix differential form as
𝑍̇ = 𝐴𝑧 + 𝐵 𝑢
𝑦 = 𝐶𝑥
where 𝐴 = 𝑃 𝐴𝑃 , 𝐵 = 𝑃 𝐵 and 𝐶 = 𝐶𝑃
EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 22

11
 5/19/2024

1.0 STATE VARIABLE METHODS

 Substituting these, we can write:

𝑧̇ 3 2.5 0.5 0 1 0 1 1 1 𝑧 3 2.5 0.5 0
𝑧̇ = −3 −4 −1 0 0 1 −1 −2 −3 𝑧 + −3 −4 −1 0
𝑧̇ 1 1.5 0.5 −6 −11 −6 1 4 9 𝑧 1 1.5 0.5 6
and
1 1 1 𝑧
𝑦 = 𝐶𝑃𝑧 = 1 0 0 −1 −2 −3 𝑧
1 4 9 𝑧
 Simplifying gives the transformed state model as:
𝑧̇ −1 0 0 𝑧 3
𝑧̇ = 0 −2 0 𝑧 + −6 𝑢
𝑧̇ 0 0 −3 𝑧 3
𝑧
𝑦= 1 1 1 𝑧
𝑧
 The diagonal matrix exists and is obtained by using the transformation matrix P to modify the coefficient matrix of 𝑧.
 It can be seen that the diagonal elements of the matrix 𝑃 𝐴𝑃 are similar to the three eigenvalues of A.

(iii) To show that the characteristic equation does not change under similarity transformation, we need to prove the
invariance of the eigenvalues under the linear transformation P, by showing that the characteristic polynomials λ𝐈 − 𝐀
and λ𝐈 − 𝐏 𝟏 𝐀𝐏 are identical.

EEE 2502 Control Engineering III Lecture Notes by Dr. Linus. A. Aloo
5/19/2024 23

12