UNIT 1: STATE SPACE REPRESENTATION

1.Introduction

A modern complex system may have many inputs, many outputs, and
these may be interrelated in a complicated manner. To analyze such a
system, it is important to reduce the complexity of the mathematical
expressions, as well as to resort to computers for most of the tedious
computations necessary in the analysis. The state-space approach to the
system analysis is best suited for this view point.
While conventional control theory is based on the input-output relationship
or transfer function. Modern control theory is based on the description of
system equations in terms of ‘n’ first order differential equation which may
be combined into first order vector matrix differential equation. The use of
vector matrix notation greatly simplifies the mathematical representation
of systems of equations.
• The increase in the number of state variables, the number of inputs,
or the number of outputs does not necessary increase the complexity
of the equation depending on the intended solution approach to be
used.
• State-space approach to control system analysis and design is the
time domain method.

State-space: It is the ‘n’ dimensional space whose coordinate axes consists

of the 𝑥1 axis, 𝑥2 axis …. 𝑥𝑛 axis. A state can be represented by a point in
the state space. State space is essential concept of modern control theory.
The state space is the simultaneous derivation of the state-vector
differential equation with the SISO differential equation for a chosen
physical system.

State vector: if ‘n’ variables are needed to completely describe the

behaviour of a given system, the ‘n’ components of a vector ‘x’, such a
vector is called a state vector. This is a vector that determines uniquely the
system state x(t) for any time 𝑡 ≥ 𝑡𝑜 , once the state at 𝑡 = 𝑡𝑜 is given and
the input u(t) for 𝑡 ≥ 𝑡𝑜 is specified.

2. State Space Equations

Three types of variables are involved in modeling dynamic systems: input variables,
output variables and state variables.

1
The state space model is represented by,

𝒙̇ = 𝑨𝒙 + 𝑩𝒖
𝒚 = 𝑪𝒙 + 𝑫𝒖
Where
x is state vector
u is input vector
y is output vector

A: state matrix. It shows how x(t+1) depends on x(t) state transition

B: Input matrix (control matrix) . shows how u(t) influences x(t+1)
C: output matrix. Shows how y(t) depends on x(t)
D: direct transmission matrix. Shows how u (t) influences y(t) (static input -output
relation)

Example 1

Consider a linear, time invariant system, if it can be described by n states, r inputs

and m outputs then the states equations will have the following form:
𝑥̇ 1 = 𝑎11 𝑥1 + 𝑎12 𝑥2 + ⋯ + 𝑎1𝑛 𝑥𝑛 + 𝑏11 𝑢1 + 𝑏12 𝑢2 + ⋯ + 𝑏1𝑟 𝑢𝑟
𝑥̇ 2 = 𝑎21 𝑥1 + 𝑎22 𝑥2 + ⋯ + 𝑎2𝑛 𝑥𝑛 + 𝑏21 𝑢1 + 𝑏22 𝑢2 + ⋯ + 𝑏2𝑟 𝑢𝑟
⋮ ⋮
𝑥̇ 𝑛 = 𝑎𝑛1 𝑥1 + 𝑎𝑛2 𝑥2 + ⋯ + 𝑎𝑛𝑛 𝑥𝑛 + 𝑏𝑛1 𝑢1 + 𝑏𝑛2 𝑢2 + ⋯ + 𝑏𝑛𝑟 𝑢𝑟
and the output equations will have the following form:
𝑦̇ 1 = 𝑐11 𝑥1 + 𝑐12 𝑥2 + ⋯ + 𝑐1𝑛 𝑥𝑛 + 𝑑11 𝑢1 + 𝑑12 𝑢2 + ⋯ + 𝑑1𝑟 𝑢𝑟
𝑦̇ 2 = 𝑐21 𝑥1 + 𝑐22 𝑥2 + ⋯ + 𝑐2𝑛 𝑥𝑛 + 𝑑21 𝑢1 + 𝑑22 𝑢2 + ⋯ + 𝑑2𝑟 𝑢𝑟
⋮ ⋮
𝑦̇𝑚 = 𝑐𝑚1 𝑥1 + 𝑐𝑚2 𝑥2 + ⋯ + 𝑐𝑚𝑛 𝑥𝑛 + 𝑑𝑚1 𝑢1 + 𝑑𝑚2 𝑢2 + ⋯ + 𝑑𝑚𝑟 𝑢𝑟
Note that the coefficients 𝑎𝑖𝑗 , 𝑏𝑖𝑗 , 𝑐𝑖𝑗 , 𝑎𝑛𝑑 𝑑𝑖𝑗 are constants and that some of them
may be zero. Using vector-matrix expressions allows the equations to be written as:
𝒙̇ = 𝑨𝒙 + 𝑩𝒖
𝒚 = 𝑪𝒙 + 𝑫𝒖
where:
𝑥1 𝑎11 𝑎12 … 𝑎1𝑛 𝑏11 𝑏12 … 𝑏1𝑟 𝑢1
𝑥2 𝑎21 𝑎22 … 𝑎2𝑛 𝑏21 𝑏22 … 𝑏2𝑟 𝑢2
𝒙=[ ⋮ ] 𝑨=[⋮ ⋮ ⋮ ] 𝑩 = [⋮ ⋮ ]𝒖 = [ ⋮ ]
⋮
𝑥𝑛 𝑎𝑛1 𝑎𝑛2 … 𝑎𝑛𝑛 𝑏𝑛1 𝑏𝑛2 … 𝑏𝑛𝑟 𝑢𝑟

𝑦1 𝑐11 𝑐12 … 𝑐1𝑛 𝑑11 𝑑12 … 𝑑1𝑟

𝑦2 𝑐21 𝑐22 … 𝑐2𝑛 𝑑21 𝑑22 … 𝑑2𝑟
𝒚=[ ⋮ ] 𝑪=[⋮ ⋮ ⋮] 𝑫=[ ⋮ ⋮ ]
⋮
𝑦𝑚 𝑐𝑚1 𝑐𝑚2 … 𝑐𝑚𝑛 𝑑𝑚1 𝑑𝑚2 … 𝑑𝑚𝑟

Example 2
For example, suppose that the physical system being modelled can be
described by an nth – order differential equation:
𝑑𝑛 𝑐(𝑡) 𝑑𝑛−1 𝑐(𝑡) 𝑑𝑛−2 𝑐(𝑡) 𝑑𝑐(𝑡)
+ 𝑎1 + 𝑎2 + ⋯ + 𝑎𝑛−1 + 𝑎𝑛 𝑐(𝑡) = 𝑟(𝑡)
𝑑𝑡 𝑛 𝑑𝑡 𝑛−1 𝑑𝑡 𝑛−2 𝑑𝑡
(1)

2
The variables c(t) and r(t) are the output and input variables, respectively.
The above differential equation can be reduced to a set of first-order
differential equations by defining the state variables as follows:

𝑥1 (𝑡) = 𝑐(𝑡)
𝑑𝑐(𝑡)
𝑥1̇ (𝑡) = 𝑑𝑡
.
. (2)
.
𝑑𝑛−1 𝑐(𝑡)
𝑥𝑛̇ (𝑡) = 𝑑𝑡 𝑛−1

The state equations can be written as

̇ = 𝑥2 (𝑡)
𝑥1 (𝑡)
̇ = 𝑥3 (𝑡)
𝑥2 (𝑡)
.
.
𝑥2 (𝑡) = −𝑎𝑛 𝑥1 (𝑡) − 𝑎𝑛−1 𝑥2 (𝑡) … − 𝑎1 𝑥𝑛 (𝑡) + 𝑟(𝑡)

The last equation is obtained by solving for the highest-order derivative in

the original differential equation. Rewriting the equation in state vector
form yields
𝑥̇ = 𝐴𝑥 + 𝐵𝑢
Where
0 1 0 0 0 . 0
0 0 1 0 0 . 0
0 0 0 1 0 . 0
𝐴= . . . . . . 0
. . . . . . 0
0 0 0 0 0 . 1
[−𝑎𝑛 −𝑎𝑛−1 −𝑎𝑛−2 −𝑎𝑛−3 −𝑎𝑛−4 . −𝑎1 ]
0
0
0
𝐵= 0
.
.
[1]
The output equation 𝑦 = 𝐶𝑥
Where 𝐶 = [1 0 0 . . 0]

3.State Space Formulation

Steps in modelling
1. Understand the physical system and its components
3
 2. Make appropriate simplifying assumptions
3. Use basic principles to formulate the mathematical model
4. Write differential and algebraic equations describing the model
5. Check the model for validity

Example
Consider a mass M, on a frictionless surface connected to a rigid wall by a spring with
stiffness K. Obtain the mathematical model for the system.
Spring
y (t)

Mass
M(kg)
K (N/m)

Solution:
1. Choose a sign convention for the position variable y (t). NB that the sign
convention for velocity and acceleration are the same as that for displacement.
2. Use fundamental physical principles to model the system (Newton’s law)
3. Draw free body diagrams of the system. In this example, the spring force is the
only force acting on the mass.

y (t)

Ky(t) M(kg
N )

4. Spring exerts a force proportional to and in opposition to movement of the

mass.
From Newton’s Law ∑ 𝐹 = 𝑚𝑎
2
𝑑 𝑦
∴ −𝐾𝑦(𝑡) = 𝑀 2
𝑑𝑡
𝑑2𝑦
𝑀 2 + 𝐾𝑦(𝑡) = 0
𝑑𝑡
5. Obtain differential equation

𝑀 𝑑2𝑦
+ 𝑦(𝑡) = 0
𝐾 𝑑𝑡 2
1 𝑑2𝑦
+ 𝑦(𝑡) = 0
𝜔𝑛 2 𝑑𝑡 2
𝐾
where 𝜔𝑛 = √𝑀 , 𝜔𝑛 is the undamped natural frequency
1 𝑑2 𝑦
6. System characteristics + 𝑦(𝑡) = 0.
𝜔𝑛 2 𝑑𝑡 2
The system has no input. No external force acts on the mass. In the differential
equation, this is indicated by the zero on the right-hand side. The system has
no damping.

4
Similarly, consider the mechanical system with mass-spring-Damper configuration
with system equation of:
𝑚𝑦̈ + 𝑏𝑦̇ + 𝑘𝑦 = 𝑢
This system is second order therefore we need two state variables to describe the
system dynamics. As 𝑦̇ (0), 𝑦(0)𝑎𝑛𝑑 𝑢(𝑡) ≥ 0 completely determines the system
behavior for 𝑡 ≥ 0:

This is of the second order. This means that the system involves 2
integrators.

Step 1
Define the state variables as
𝑥1 (𝑡) = 𝑦(𝑡) (5)
𝑥2 (𝑡) = 𝑦̇ (𝑡) (6)
Then we obtain
𝑥̇ 1 = 𝑥2
1 1
𝑥̇ = 𝑚 (−𝑘𝑦 − 𝑏𝑦̇ ) + 𝑚 𝑢
Or
𝑥̇ = 𝑥2 (7)
𝑘 𝑏 1
𝑥2̇ = − 𝑚 𝑥1 − 𝑚 𝑥2 + 𝑚 𝑢 (8)
The output equation is
𝑦 = 𝑥1 (9)
In vector matrix equation (7) and (8) can be written as
𝑥̇ 0 1 𝑥1 0
[ 1 ] = [− 𝑘 − 𝑏 ] [𝑥 ] + [ 1 ] 𝑢
𝑥2̇ 𝑚 𝑚 2 𝑚
(10)
The output equation (9) can be written as
𝑥1
𝑦 = [1 0] [𝑥 ] (11)
2
Equation (10) is the state space equation and equation (11) is the output
equation for the system.
Equation (10) and (11) are in the standard form:
𝑥̇ = 𝐴𝑥 + 𝐵𝑢
𝑦 = 𝐶𝑥 + 𝐷𝑢
Where
0 1 0
𝐴 = [− 𝑘 − 𝑏 ] 𝐵 = [ 1 ] 𝐶 = [1 0] 𝐷 = 0
𝑚 𝑚 𝑚

Example 3

Obtain the state equation in phase variable form for the following
differential equation:

5
 𝑑3 𝑦 𝑑2𝑦 𝑑𝑦
2 3 +4 2 +6 + 8𝑦 = 10𝑢(𝑡)
𝑑𝑡 𝑑𝑡 𝑑𝑡
Solution
The differential equation is third order, and thus there are three state
variables:
𝑥1 = 𝑦, 𝑥2 = 𝑦̇ , and 𝑥3 = 𝑦̈ .
The first derivatives are:
𝑥̇ 1 = 𝑥2
𝑥̇ 2 = 𝑥3
𝑥̇ 3 = −4𝑥1 − 3𝑥2 − 2𝑥3 + 5𝑢(𝑡)
Or, in matrix form:
𝑥̇ 1 0 1 0 𝑥1 0
[ 𝑥̇ 2 ] = [ 0 0 1 ] [ 𝑥 2 ] + [ 0] 𝑢(𝑡)
𝑥̇ 3 −4 −3 −2 𝑥3 5
𝑥1
𝑦 = [1 0 0] [ 𝑥2 ]
𝑥3

Example

Obtain the state equation for the circuit below, where i2 is considered to be the
output of the system with these state variables 𝑥1 = 𝑖1 , 𝑥2 = 𝑖2 and 𝑥3 = 𝑣𝑐

Figure : An electric circuit

Expected Answer
𝑅 1
− 𝐿1 0 −𝐿 1
1 1
𝑅2 1 𝐿1
𝑥̇ = 0 −𝐿 x+ [ 0 ] 𝑢
2 𝐿2
1 −1 0
[ 0 ]
𝑐 𝑐

Y=[0 1 0]x

6
4 STATE TRANSITION MATRIX

The state transition matrix defined as a matrix that satisfies the linear homogeneous
state equation, i.e.
𝒙̇ = 𝑨𝒙 Homogeneous state equation
𝑥1 (0)
𝒙(0) = [ ⋮ ] Initial state at time t=0
𝑥𝑛 (0)
𝒙(𝑡) = 𝚽(𝑡)𝒙(0)
Where Φ(𝑡) is the state transition matrix.

State transition matrix is used for the following:

1. Solution of Cayley -Hamilton theorem
2. Laplace transform method

A linear system is called homogeneous if the righthand of the equation is equal to

𝑑𝑦
zero, for example, 𝑑𝑥 − 5𝑦 = 𝑜

Work Example

Find a physical homogeneous system and identify areas of application of state

transition matrix.

4.1 State transition matrix by Laplace transformation technique

𝑥1 (0)
𝒙̇ = 𝑨𝒙 and 𝒙(0) = [ ⋮ ]
𝑥𝑛 (0)
Taking the Laplace transformation of the above equation yields
𝑠𝒙(𝑠) − 𝒙(0) = 𝑨𝒙(𝑠)
Or
𝒙(𝑠) = [𝑠𝐈 − 𝐀]−1 𝑥(0)
The state transition matrix is obtained by taking the inverse Laplace Transform of the
equation above
𝚽(𝑡) = ℒ −1 [(𝑠𝐈 − 𝐀)−1 ]

Find the state transition matrix of the system described by the following
equations:
𝑥̇ 0 1 𝑥1
[ 1] = [ ][ ]
𝑥2̇ −3 −4 𝑥2
Where
𝑥 (0) 0
[ 1 ] = [1]
𝑥2 (0)
𝑠 1 −1
ф(𝑠) = [𝑠𝐼 − 𝐴]−1 = [ ] (find adjoint and determinant)
3 𝑠+4

7
 𝑠+4 1
2 𝑠2 +4𝑠+3
[𝑠 +4𝑠+3
−3 𝑠 ]
𝑠2 +4𝑠+3 𝑠2 +4𝑠+3
Using the partial fraction expansion, the elements of the transition matrix
can be written as
3 1 1 1
− 2(𝑠+3) − 2(𝑠+3)
2(𝑠+1) 2(𝑠+1)
ф(𝑠) = [ −3 3 −1 3 ]
+ +
2(𝑠+1) 2(𝑠+3) 2(𝑠+1) 2(𝑠+3)
the state transition matrix is determined by taking the inverse Laplace
transform of ф(s):
∴ф (t) =L-1(ф(s))
3 1 1 −𝑡 1
𝑒 −𝑡 − 2 𝑒 −3𝑡 𝑒 + 2 𝑒 −3𝑡
2 2
ф(𝑡) = [ 3 3 1 3 ]
− 2 𝑒 −𝑡 + 2 𝑒 −3𝑡 − 2 𝑒 −𝑡 + 2 𝑒 −3𝑡

4.2 State Transition Matrix by Classical Technique

The state transition matrix can also be found in the following manner
𝒙(𝑡) = 𝑒 𝑨𝑡 𝒙(0)
Where 𝑒 is a matrix exponential and𝑑𝑒 𝐴𝑡 ⁄𝑑𝑡 = 𝐴𝑒 𝐴𝑡 . Substituting the above
𝑨𝒕

equation into the homogeneous state equation shows that it is a solution:

𝐴𝑒 𝐴𝑡 𝑥(0) = 𝐴𝑒 𝐴𝑡 𝑥(0)
𝑒 𝐴𝑡 can be represented by a power series as follows:
1 1
𝑒 𝐴𝑡 = 𝑰 + 𝑨𝑡 + 𝑨2 𝑡 2 + 𝑨3 𝑡 3 + ⋯
2 3
The state transition matrix is:
1 1
Φ(𝑡) = 𝑰 + 𝑨𝑡 + 𝑨2 𝑡 2 + 𝑨3 𝑡 3 + ⋯
2 3
Since the state transition matrix satisfies the homogeneous state equation, it
represents the free response of the system.

4.3 Properties of the State Transition Matrix

Some of the properties of thee state transition matrix are represented below.
1. Φ(0) = 𝑒 𝐴0 = 𝐼
2. [Φ(𝑡)]−1 = [Φ(−t)]
3. Φ(𝑡1 + 𝑡2 ) = 𝑒 𝐴(𝑡1+𝑡2 ) = 𝑒 𝐴𝑡1 𝑒 𝐴𝑡2 = Φ(𝑡1 )Φ(𝑡2 ) = Φ(𝑡2 )Φ(𝑡1 )
4. [Φ(𝑡)]𝑘 = Φ(𝑘𝑡), 𝑤ℎ𝑒𝑟𝑒 𝑘 𝑖𝑠 𝑎𝑛 𝑖𝑛𝑡𝑒𝑔𝑒𝑟

Once the state transition matrix has been found, the solution to the
nonhomogeneous equation can be determined as follows:
𝒙̇ = 𝑨𝒙 + 𝑩𝜼
Taking the Laplace transform of the above equation yields
𝑠𝑥(𝑠) − 𝑥(0) = 𝐴𝑥(𝑠) + 𝐵𝜂(𝑠)
Solving for 𝑥(𝑠)
𝑥(𝑠) = [𝑠𝐼 − 𝐴]−1 𝑥(0) + [𝑠𝐼 − 𝐴]−1 𝐵𝜂(𝑠)
𝑥(𝑡) = ℒ −1 [𝑠𝐼 − 𝐴]−1 𝑥(0) + ℒ −1 [[𝑠𝐼 − 𝐴]−1 𝐵𝜂(𝑠)]

8
Or
1

𝑥(𝑡) = Φ(𝑡)𝑥(0) + ∫ Φ(𝑡 − 𝜏)𝐵𝜂(𝜏)𝑑𝜏

0

5. Transient-Response Analysis of State-Space Systems

Step Response using MatLab
First the system is defined with
𝑠𝑦𝑠 = 𝑠𝑠(𝐴, 𝐵, 𝐶, 𝐷)
For a unit step response, the command
𝑠𝑡𝑒𝑝(𝑠𝑦𝑠)or 𝑠𝑡𝑒𝑝(𝐴, 𝐵, 𝐶, 𝐷)
will generate plots of the responses. There will be one plot for each state and a set of
plots for each input.

5.1 Correlation between transfer functions and state – space equations.

How to derive the transfer functions of a single-input-single-output system from the
+state – space equations.
Let us consider the system whose transfer function is given by
𝑦(𝑠)
= 𝐺(𝑠) (5)
𝑢(𝑠)
This system may be represented in state space by the following equations
𝑥̇ = 𝐴𝑥 + 𝐵𝑢 (6)
𝑦 = 𝐶𝑥 + 𝐷𝑢 (7)
Where x is the state vector, u is the input and y is the output. The Laplace transform
of equation (6) and (7) are given by
𝑠𝑥(𝑠) − 𝑥(0) = 𝐴𝑥(𝑠) + 𝐵𝑢(𝑠) (8)
𝑦(𝑠) = 𝐶𝑥(𝑠) + 𝐷𝑢(𝑠) (9)
Since the transfer function was previously defined as the ratio of the Laplace transform
of the output to the Laplace transform of the input when the initial conditions were zero,
we assume that x(0) in equation (8) is zero.

Then we have
𝑠𝑥(𝑠) − 𝐴𝑥(𝑠) = 𝐵𝑢(𝑠)
(𝑠𝐼 − 𝐴)𝑥(𝑠) = 𝐵𝑢(𝑠)
−1
By pre multiplying (𝑠𝐼 − 𝐴) to both sides of this last equation we obtain
𝑥(𝑠) = (𝑠𝐼 − 𝐴)−1 𝐵𝑢(𝑠) (10)
By substituting equation (10) into equation (9) we get
𝑦(𝑠) = [𝐶 (𝑠𝐼 − 𝐴)−1 𝐵 + 𝐷]𝑢(𝑠) (11)
By comparing equation (11) with equation (5) we see that
𝐺(𝑠) = 𝐶 (𝑠𝐼 − 𝐴)−1 𝐵 + 𝐷 (12)
This is the transfer function expression in terms of A, B, C, and D.

Consider a multiple-input-multiple-output(MIMO) system with r input and m outputs.

Thus:

9
 𝑢1 𝑦1
𝑢2 𝑦2
𝑢 = [ ⋮ ]𝑦 = [ ⋮ ]
𝑢𝑟 𝑦𝑚
Now the transfer function matrix G(s) relates the output Y(s) to the input U(s), that is
𝑌(𝑠) = 𝐺(𝑠)𝑈(𝑠)
Given that:

𝒙̇ = 𝑨𝒙 + 𝑩𝒖
𝒚̇ = 𝑪𝒙 + 𝑫𝒖
Then; taking Laplace transforms gives:
𝑠𝑋(𝑠) − 𝑥(0) = 𝐴𝑋(𝑠) + 𝐵𝑈(𝑠)
𝑌(𝑠) = 𝐶𝑋(𝑠) + 𝐷𝑈(𝑠)
Assuming zero initial condition then:
𝑋(𝑠) = (𝑠𝐼 − 𝐴)−1 𝐵𝑈(𝑠)
And:
𝑌(𝑠) = [𝐶(𝑠𝐼 − 𝐴)−1 𝐵 + 𝐷]𝑈(𝑠)
Therefore:
𝐺(𝑠) = 𝐶(𝑠𝐼 − 𝐴)−1 𝐵 + 𝐷

Worked Examples- Transfer Matrix

Consider the following system:
𝑥̇ −1 −1 𝑥1 1 1 𝑢1
[ 1] = [ ] [𝑥 ] + [ ][ ]
𝑥̇ 2 6.5 0 2 1 0 𝑢2
𝑦1 1 0 𝑥1 0 0 𝑢1
[𝑦 ] = [ ] [𝑥 ] + [ ][ ]
2 0 1 2 0 0 𝑢2
Determine the transfer matrix (𝐺(𝑠)) for this system.
Solution:
𝐺(𝑠) = 𝐶(𝑠𝐼 − 𝐴)−1 𝐵 + 𝐷
But:
𝑠+1 1
𝑠𝐼 − 𝐴 = [ ]
−6.5 𝑠
Therefore:
1 𝑠 −1
(𝑠𝐼 − 𝐴)−1 = 2 [ ]
𝑠 + 𝑠 + 6.5 6.5 𝑠 + 1
And:
1 1 0 𝑠 −1 1 1
𝐶(𝑠𝐼 − 𝐴)−1 𝐵 = 2 [ ][ ][ ]
𝑠 + 𝑠 + 6.5 0 1 6.5 𝑠 +1 1 0
1 𝑠−1 𝑠
= 2 [ ]
𝑠 + 𝑠 + 6.5 𝑠 + 7.5 6.5
Hence:
𝑠−1 𝑠
𝑌 (𝑠) 2 2 𝑈 (𝑠)
𝐺(𝑠) = [ 1 ] = [𝑠 + 𝑠 + 6.5 𝑠 + 𝑠 + 6.5] [ 1 ]
𝑌2 (𝑠) 𝑠 + 7.5 6.5 𝑈2 (𝑠)
2 2
𝑠 + 𝑠 + 6.5 𝑠 + 𝑠 + 6.5
Or:
𝑌1 (𝑠) 𝑠−1 𝑌1 (𝑠) 𝑠
= 2 = 2
𝑈𝐼 (𝑠) 𝑠 + 𝑠 + 6.5 𝑈2 (𝑠) 𝑠 + 𝑠 + 6.5
𝑌2 (𝑠) 𝑠 + 7.5 𝑌2 (𝑠) 6.5
= 2 = 2
𝑈𝐼 (𝑠) 𝑠 + 𝑠 + 6.5 𝑈2 (𝑠) 𝑠 + 𝑠 + 6.5

10
Note the importance of |𝑠𝐼 − 𝐴| in determining the dynamic response of the system.
The force individual step response curves can be plotted in MatLab by using the
command step(A,B,C,D) or step(sys) where 𝑠𝑦𝑠 = 𝑠𝑠(𝐴, 𝐵, 𝐶, 𝐷).

Example

Find the transfer function of the mass/spring/damper state space model

𝑥̇ 0 1 𝑥1 0
[ 1] = [ ] [ ] + [ ] [𝑢]
𝑥2̇ −2 −3 𝑥2 1
𝑥1
[𝑦] = [1 0] [𝑥 ]
2
Solution:
1 0 0 1 𝑠 −1
𝑠𝐼 − 𝐴 = 𝑠 [ ]−[ ]=[ ]
0 1 −2 −3 2 𝑠+3
𝑠 −1 𝑠+3 1 𝑠+3 1
𝑎𝑑𝑗[ ] [ ] [ ]
(𝑠𝐼 − −1 2 𝑠+3 −2 𝑠 −2 𝑠
𝐴) = 𝑠 −1 = =
𝑑𝑒𝑡[ ] 𝑠(𝑠+3)+2 𝑠2 +3𝑠+2
2 𝑠+3
𝑠+3 1 0 1
[ ][ ] [ ]
(𝑠𝐼 − 𝐴)−1 𝐵 = −2 𝑠 1
= 𝑠
𝑠2 +3𝑠+2 𝑠2 +3𝑠+2
1
[1 0][ ] [1]
𝑠
𝐶(𝑠𝐼 − 𝐴)−1 𝐵 = 𝑠2 +3𝑠+2 = 𝑠2 +3𝑠+2
−1 1
𝐺(𝑠) = 𝐶(𝑠𝐼 − 𝐴) 𝐵 + 𝐷 = 𝑠2 +3𝑠+2

Obtain the transfer function for the system described in the state space
model below.
𝑥̇ 0 1 𝑥1 0
[ 1] = [ ] [ ] + [ ] 𝑢(𝑡)
𝑥̇ 2 −6 −5 𝑥2 1
𝑥1
𝑦 = [8 1] [𝑥 ]
2
Solution
0 1 0
𝐴=[ ], 𝐵 = [ ], 𝐶 = [8 1], 𝐷=0
−6 −5 1
𝑠 −1
𝑠𝐼 − 𝐴 = [ ]
6 𝑠+5
𝑠+5 1
[ ]
Φ(𝑠) = (𝑠𝐼 − 𝐴) = 2−1 −6 𝑠
𝑠 + 5𝑠 + 6
𝑠+5 1 0
[ ][ ]
𝐺(𝑠) = 𝐶(𝑠𝐼 − 𝐴)−1 𝐵 = [8 1] −6 𝑠 1
𝑠 2 + 5𝑠 + 6
[8 1] [1]
= 2 𝑠
𝑠 + 5𝑠 + 6

Therefore, the transfer function is

𝑠+8
𝐺(𝑠) =
𝑠 2 + 5𝑠 + 6

11
Obtain the transfer function of the system given by
𝑥̇ 1 −1 1 0 𝑥1 0
[ 𝑥̇ 2 ] = [ 0 −1 1 ] [ 𝑥2 ] + [0] 𝑢
𝑥̇ 3 0 0 −2 𝑥3 1
𝑥1
𝑦 = [1 0 0] [ 𝑥2 ]
𝑥3
Solution
The transfer function G(s) is given by
𝐺(𝑠) = 𝑪(𝑠𝑰 − 𝑨)−1 𝑩 + 𝐷
In this problem, matrices A, B, C, and D are
−1 1 0 0
𝑨 = [ 0 −1 1 ] , 𝑩 = [0] , 𝑪 = [1 0 0], 𝐷=0
0 0 −2 1
Hence
𝑠 + 1 −1 0 −1 0
𝐺(𝑠) = [1 0 0] [ 0 𝑠 + 1 −1 ] [0]
0 0 𝑠+2 1
1 1 1
𝑠 + 1 (𝑠 + 1)2 (𝑠 + 1)2 (𝑠 + 2)
1 1 0
= [1 0 0] 0 [0]
𝑠+1 (𝑠 + 1)(𝑠 + 2) 1
1
[ 0 0 ]
𝑠+2
1 1
= =
(𝑠 + 1)2 (𝑠 + 2) 𝑠 3 + 4𝑠 2 + 5𝑠 + 2

Assignment
A state space model for the longitudinal motion of a helicopter near hover
is
𝑥1̇ −0.4 0 −0.01 𝑥1 6.3
[𝑥2̇ ] = [ 1 0 0 ] [𝑥2 ] + [ 0 ] 𝑢
𝑥3̇ −1.4 9.8 −0.02 𝑥3 9.8
𝑥1
𝑦 = [0 0 1] [𝑥2 ]
𝑥3
Determine the transfer function matrix, characteristic equation and state
whether the helicopter is stable.

CONTROLLABILITY AND OBSERVABILITY

CONTROLLABILITY AND OBSERVABILITY
The two concepts that play an important role in modern control theory are
the concepts of controllability and observability.
Controllability is concerned with whether the states of the dynamic system
are affected by the control input. A system is said to be controllable if there
exists a control that transfers any initial state 𝑥𝑖 (𝑡) to any final state 𝑥𝑓 (𝑡)

12
in some finite time. If one or more of the states are unaffected by the
control, the system is not controllable. A mathematical definition of
controllability for a linear dynamical system can be expressed as follows. If
the dynamic system can be described by the state equation
𝑥1̇ = 𝐴𝑥 + 𝐵𝑢
where 𝑥 and 𝑢 are the state and control vectors of order ‘n’ and ‘m’
respectively, then the necessary and sufficient condition for the system to
be completely controllable is that the rank of the matrix P is equal to the
number of states. The matrix P is constructed from the A and B matrices in
the following way:
𝑃 = [𝐵, 𝐴𝐵, 𝐴2 𝐵, … , 𝐴𝑛−1 𝐵]
The rank of a matrix is defined as the largest non – zero determinant. The
mathematical test was developed by kalman in the 1960’s.
Consider the system given by
𝑥̇ 1 1 𝑥1 1
[ 1] = [ ] [𝑥 ] + [ ] [𝑢]
𝑥2̇ 0 −1 2 0
Solution
𝑃 = [𝐵, 𝐴𝐵, 𝐴2 𝐵, … , 𝐴𝑛−1 𝐵]
1 1 1
Matrix A =[ ] and B=[ ]
0 −1 0
1 1 1 1
AB=[ ][ ] = [ ]
0 −1 0 0
1 1
∴ P=[ ] =singular
0 0
Hence the system is not completely state controllable. Since the inverse
and the determinant are zero.
Example 2
Consider the system given by
𝑥̇ 1 1 𝑥1 0
[ 1] = [ ] [ ] + [ ] [𝑢]
𝑥2̇ 2 −1 𝑥2 1
is controllable.

OBSERVABILITY
It deals with whether the states of the system can be identified from the
output of the system. A system is said to be completely observable if every
state 𝑥 can be determined by the measurement of the output 𝑦(𝑡) over a
finite time interval. If one or more states cannot be identified from the
output of the system, then the system is not observable. A mathematical
test for observability of an nth-order dynamic system governed by the
equations.
𝑥̇ = 𝐴𝑥 + 𝐵𝑢
𝑦 = 𝐶𝑥 + 𝐷𝑢
is given as follows. The necessary and sufficient condition for a system to
be completely observable is that the matrix 𝑀𝑜 is defined as
𝑀𝑜 = [𝐶 𝑇 , A𝐶 𝑇 , … , (𝐴𝑇 )𝑛−1 𝐶 𝑇 ] is of the rank n.
Consider the system described by
𝑥̇ 1 1 𝑥1 0
[ 1] = [ ] [ ] + [ ] [𝑢]
𝑥2̇ −2 −1 𝑥2 1

13
 𝑥1
[𝑦] = [1 0] [𝑥 ]
2
Is this system controllable and observable?
Try controllability.
Solution
1 1 0
Matrix A=[ ] B=[ ] C=[1 0]
−2 −1 1
𝑇 𝑇 1 −2 1 1
[𝐴 𝐶 ] = [ ] [ ]=[ ]
1 −1 0 1

Observability Matrix
𝑀𝑜 = [𝐶 𝑇 , A𝐶 𝑇 , … , (𝐴𝑇 )𝑛−1 𝐶 𝑇 ]
1 1
𝑀𝑜 = [ ]
0 1
The rank of [𝐴𝐶 𝑇 : 𝐴𝑇 𝐶 𝑇 ] is 2, hence the system is observable. The rank of a
matrix A is the order of the largest square array contained in A which has
a non-zero determinant.

Example

Exercise.
1. Which of the systems are fully controllable

1 4 0
a) 𝑥̇ = [ ]𝑥 + [ ]𝑢
−2 −5 1

0 4 1
b) 𝑥̇ = [ ]𝑥 + [ ]𝑢
−2 6 1

1 0 0
c) 𝑥̇ = [ ]𝑥 + [ ]𝑢
−8 −5 1

14
 0 4 1
d) 𝑥̇ = [ ]𝑥 + [ ]𝑢
−2 −6 1

15