State Space Analysis

State variable approach is a powerful tool for analysis and design of control
systems.

A modern approach capable of handling

1. Linear system
2. Non-Linear system
3. Time Invariant System
4. Time varying System
5. MIMO system
Conventional method of transfer function approach has the following
drawbacks.

1. Transfer function is defined under zero initial conditions

2. Applicable to only LTI systems
3. Restricted to SISO systems only
4. Does not provide information regarding internal state of the system.
State Variables
Minimum set of variables 𝑥1 (𝑡), 𝑥2 (𝑡),……., 𝑥𝑛 (𝑡) such that knowledge of
these variables at time 𝑡=𝑡0 and the information about the input excitation
subsequently applied is sufficient to determine the state of the system at any
time 𝑡 > 𝑡0

State variables are used to represent the dynamic characteristics of the

system.
• At time 𝑡=𝑡0 , 𝑥1 (𝑡0 ), 𝑥2 (𝑡0 ),……., 𝑥𝑛 (𝑡0 ) refers to the initial state of the
system

• With the initial states specified, once the input is applied to the system,
state variables should completely define the future behaviour of the system
State space
N dimensional space whose co-ordinate axis consist of 𝑥1 axis, 𝑥2 axis, …. 𝑥𝑛
axis.
Any state can be uniquely represented by a point in the state space
In state variable formulation of system, a system consists of m inputs, p
outputs, and n state variables
Let
State variables be 𝑥1 (t), 𝑥2 (t), 𝑥3 (t)……….. 𝑥𝑛 (t)
Input variables be 𝑢1 (t), 𝑢2 (t), 𝑢3 (t)……….. 𝑢𝑚 (t)
Output variables be 𝑦1 (t), 𝑦2 (t), 𝑦3 (t)……….. 𝑦𝑝 (t)
State Equations
State variable representation can be arranged in the form of a n number of
first order differential equations

𝑑𝑥1
= 𝑥1ሶ = 𝑓1 (𝑥1 , 𝑥2 , 𝑥3 … . 𝑥𝑛 ; 𝑢1 , 𝑢2 , 𝑢3 … . 𝑢𝑚 )
𝑑𝑡
𝑑𝑥2
= 𝑥2ሶ = 𝑓2 𝑥1 , 𝑥2 , 𝑥3 … . 𝑥𝑛 ; 𝑢1 , 𝑢2 , 𝑢3 … . 𝑢𝑚
𝑑𝑡
.
.
𝑑𝑥𝑛
= 𝑥𝑛ሶ = 𝑓𝑛 (𝑥1 , 𝑥2 , 𝑥3 … . 𝑥𝑛 ; 𝑢1 , 𝑢2 , 𝑢3 … . 𝑢𝑚 )
𝑑𝑡
In vector notation,
ሶ
𝑋(𝑡)= f 𝑋 𝑡 , 𝑈(𝑡) ……… [1]
The set of all possible values which input vector can have at time t forms the
input space of the system.

The set of all possible values which output vector can have at time t forms the
output space of the system.

The set of all possible values which state vector can have at time t forms the
state space of the system.
State Model of a LTI system
State Model of a system has state equation and output equation.
State equation is a function of state variables and inputs as defined in [1]
For LTI systems, first derivatives of state variables can be expressed as a
linear combination of state variables and inputs
𝑥1ሶ = 𝑎11 𝑥1 + 𝑎12 𝑥2 + …. +𝑎1𝑛 𝑥𝑛 +𝑏11 𝑢1 +𝑏12 𝑢2 + …. +𝑏1𝑚 𝑢𝑚
𝑥2ሶ = 𝑎21 𝑥1 + 𝑎22 𝑥2 + …. +𝑎2𝑛 𝑥𝑛 +𝑏21 𝑢1 +𝑏22 𝑢2 + …. +𝑏2𝑚 𝑢𝑚
.
.
𝑥𝑛ሶ = 𝑎𝑛1 𝑥1 + 𝑎𝑛2 𝑥2 + …. +𝑎𝑛𝑛 𝑥𝑛 +𝑏𝑛1 𝑢1 +𝑏𝑛2 𝑢2 + …. +𝑏𝑛𝑚 𝑢𝑚
In the matrix form the above equation can be represented as
 ሶ
The equation X(𝑡)= AX t +B U(t) is called the State equation of an LTI system

The output at any point of time are the functions of state variables and inputs
Therefore,
Output vector Y(𝑡)= f 𝑋 𝑡 , 𝑈(𝑡)
Output variables can be expressed as the linear combination of state
variables and inputs.
𝑦1 = 𝑐11 𝑥1 + 𝑐12 𝑥2 + …. +𝑐1𝑛 𝑥𝑛 +𝑑11 𝑢1 +𝑑12 𝑢2 + …. +𝑑1𝑚 𝑢𝑚
𝑦2 = 𝑐21 𝑥1 + 𝑐22 𝑥2 + …. +𝑐2𝑛 𝑥𝑛 +𝑑21 𝑢1 +𝑑22 𝑢2 + …. +𝑑2𝑚 𝑢𝑚
.
.
𝑦𝑝 = 𝑐𝑝1 𝑥1 + 𝑐𝑝2 𝑥2 + …. +𝑐𝑝𝑛 𝑥𝑛 +𝑑𝑝1 𝑢1 +𝑑𝑝2 𝑢2 + …. +𝑑𝑝𝑚 𝑢𝑚
In the matrix form the above equation can be represented as

The equation Y(𝑡)= CX t +D U(t) is called the output equation of an LTI system
State equation together with output equation is called State Model of a
system
ሶ
X(𝑡)= AX t +B U(t)
Y(𝑡)= CX t +D U(t)
Let 𝑥1 𝑡 and 𝑥2 (𝑡) be any two variables defined such that

𝑑𝑥1 (𝑡)
= 𝑥2 (𝑡)
𝑑𝑡

Integrating on both the sides and taking Laplace transform,

𝑋2 𝑆 𝑋1 𝑡0
𝑋1 𝑆 = + for 𝑡 > 𝑡0 initial value theorem
𝑆 𝑆
Corresponding SFG can be written as
State Diagram ,State equation and Transfer function from differential
equations

Consider a differential equation describing dynamic behaviour of a system

𝑑2 𝑦 𝑡 𝑑𝑦 𝑡
+ 3 + 2𝑦(𝑡) = 𝑟 𝑡
𝑑𝑡 2 𝑑𝑡

Keep the highest order derivative on LHS,

𝑑2 𝑦 𝑡 𝑑𝑦 𝑡
2
= −3 − 2𝑦 𝑡 + 𝑟 𝑡
𝑑𝑡 𝑑𝑡

𝑆 2 𝑌 𝑆 = −3 𝑆𝑌 𝑆 − 2𝑌 𝑆 + 𝑅(𝑆)
𝑆 2 𝑌 𝑆 = −3 𝑆𝑌 𝑆 − 2𝑌 𝑆 + 𝑅(𝑆)

Let 𝑥1 (𝑡)= 𝑦(𝑡)

and be the two state variables
State variables
𝑑𝑥1 (𝑡)
𝑥2 (𝑡) = = 𝑥1ሶ (1)
𝑑𝑡

Then
𝑑𝑥2 (𝑡)
𝑥2ሶ = = −3𝑥2 (𝑡) −2𝑥1 (𝑡) + 𝑟(𝑡) (2)
𝑑𝑡

𝑥1ሶ 0 1 𝑥1 (𝑡) 0
= + 𝑟(𝑡) ; State equation
𝑥2ሶ −2 −3 𝑥2 (𝑡) 1
𝑆 2 𝑌 𝑆 = −3 𝑆𝑌 𝑆 − 2𝑌 𝑆 + 𝑅(𝑆)
Specify initial conditions for state variables 𝑋1 (𝑆)= Y(𝑆) and 𝑋2 (𝑆)= 𝑆𝑌(𝑆)
Corresponding State Diagram can be written as
Transfer function can be obtained by applying gain formula between input
and output node and setting the initial condition to zero.

𝑆 2 𝑌 𝑆 = −3 𝑆𝑌 𝑆 − 2𝑌 𝑆 + 𝑅(𝑆)

𝑌 𝑆 1
= ; Transfer function
𝑅(𝑆) 𝑆 2 +3𝑆+2
Transfer function from State Model
Consider a standard state model
ሶ
X(𝑡)= AX t +B U(t)
Y(𝑡)= CX t +D U(t)

Taking Laplace transform on both sides,

[SX 𝑆 - X 0 ]=AX S +B U S ………………..(1)
Y(𝑆)= CX S +D U(S) ………………..(2)
ABCD are the constants, System is Time invariant, definition of transfer
function is based on the assumption of zero initial conditions ie X 0 =0
SX 𝑆 =AX S +B U S
SX 𝑆 −AX S = B U S
S is an operator and A is a matrix of the order nxn. Hence to match the
orders, multiply S by an identity matrix of the order nxn.
SI X 𝑆 −AX S = B U S
[SI−A] X 𝑆 =B U S
X 𝑆 =[SI − A]−1 B U S ………………..(3)

Substituting (3) in (2)

Y(𝑆)= C[SI − A]−1 B U S +D U(S)

𝑌(𝑆)
∴ Transfer function = C [SI − A]−1 B + D
𝑈(𝑆)
−1 𝐴𝑑𝑗[SI−A]
Where, [SI − A] =
SI−A
Equation obtained by equating the denominator to 0 is called characteristic
equation. Roots of this equation are called closed loop poles.
∴ Characteristic equation of the system is given by
SI − A =0
In matrix algebra, roots of SI − A =0 are called Eigen values of matrix A and
denoted by λ
Example 1
Consider a system having the state model
𝑥1ሶ −2 −3 𝑥1 3 𝑥1
= 𝑥 + 𝑢 and y= [1 1] 𝑥 with D=0 . Obtain the
𝑥2ሶ 4 2 2 5 2
transfer function
Solution
𝑆 0 −2 −3 𝑆+2 3
[SI−A]= − =
0 𝑆 4 2 −4 𝑆 − 2

𝑆 − 2 −3
Adj [SI−A]= and SI − A = 𝑆 2 + 8
4 𝑆+2
𝑆−2 −3
4 𝑆+2
[SI − A]−1 =
𝑆 2 +8
𝑆−2 −3 3 3𝑆−21
[1 1] [1 1] 8𝑆+1
4 𝑆+2 5 5𝑆+22
Transfer function= C [SI − A]−1 B = = =
𝑆 2 +8 𝑆 2 +8 𝑆 2 +8
Example 2
Consider a system having the state model
𝑥1ሶ −2 1 0 𝑥1 0 𝑥1 (𝑡)
𝑥2ሶ = 0 −3 1 𝑥2 + 0 𝑢(t) and y= [0 1 0] 𝑥2 (𝑡) with D=0 .
𝑥3ሶ −3 −4 −5 𝑥3 1 𝑥3 (𝑡)

Obtain the transfer function

Solution
𝑆 0 0 −2 1 0 𝑆 + 2 −1 0
[SI−A]= 0 𝑆 0 − 0 −3 1 = 0 𝑆+3 −1
0 0 𝑆 −3 −4 −5 3 4 𝑆+5
𝑆 2 + 8𝑆 + 19 𝑆+5 1
Adj [SI−A]= −3 𝑆 2 + 7𝑆 + 10 𝑆+2
−3𝑆 − 9 −4𝑆 − 11 𝑆 2 + 5𝑆 + 6

SI − A = 𝑆 3 + 10𝑆 2 + 35𝑆 + 41

Adj [SI−A]
[SI − A]−1 =
[SI−A]
 𝑆 2 +8𝑆+19 𝑆+5 1 0
0 1 0 −3 𝑆 2 +7𝑆+10 𝑆+2 0
−3𝑆−9 −4𝑆−11 𝑆 2 +5𝑆+6 1
Transfer function= C [SI − A]−1 B =
𝑆 3 +10𝑆 2 +35𝑆+41

𝑆+2
= 3
𝑆 +10𝑆 2 +35𝑆+41
Solution of State Equation

Output response depends on state variables and their initial values. Hence it
is necessary to obtain the state vector x t which satisfies the equation
ሶ
X(𝑡)= AX t +B U(t) at any time t.
This is called solution of the state equation which helps to obtain the output
response of a system
State equation can be of two types:
1.Homogenous ; input control forces are zero
ሶ
X(𝑡)= AX t
In such a system, driving force is provided by the initial conditions of the
system.
2.Non-Homogenous:
ሶ
X(𝑡)= AX t +B U(t)
A is a constant matrix and u(t) is a non zero vector which means input
control forces are applied to the system.
A
∴X t =𝑒 𝐴𝑡 X 0
This is the required solution of homogeneous equation in the scalar form.
Thus if the homogeneous state equation X(𝑡)=ሶ AX t is considered,
then is solution can be written as
X t =𝑒 𝐴𝑡 X 0 ; 𝑒 𝐴𝑡 is not a scalar but a matrix of order nxn as that of A
It can be observed that without input, initial state X 0 drives the state X t
at any time t. Thus there is a transition of the initial state X 0 from initial
time t=0 to any time t through the matrix 𝑒 𝐴𝑡

∴ 𝑒 𝐴𝑡 is called State Transition Matrix denoted by ϕ t

Solution of State Equation by Laplace Transform Method
ϕ S =[SI − A]−1 is called Resolvant Matrix of A.
All the elements of this matrix are rational functions S
Example 3
Determine the state transition matrix of system with system matrix
0 −1
A=
2 −3

Solution
(S + 3) −1
Adj SI − A =
2 S

(𝑆+3) −1
−1 (S+1)(S+2) (S+1)(S+2)
Resolvant Matrix ϕ(S)= [SI − A] = 2 𝑆
(S+1)(S+2) (S+1)(S+2)
State Transition Matrix 𝑒 𝐴𝑡 =ϕ t =ℒ −1 [SI − A]−1
Example 4
Determine the state transition matrix of system with system matrix
0 1
A=
−3 −4
Controllability and Observability
**Is it possible to transfer a system from any initial state to other desired
state in finite time by applying suitable control force?
**Knowing the output vector for a finite duration of time, is it possible to
determine the initial state of the system?
System is said to be State Controllable if it is possible to transfer the system
state from x(𝑡0 ) to any desired state x(t) in a finite time by a control vector
u(t)
System is said to be Output Controllable if it is possible to construct an
unconditional control vector u(t) that will transfer initial output y(𝑡0 ) to y(t)

System is said to be Observable if every state x(t) can be completely

identified by measuring the output y(t) over a finite time interval
Controllability and Observability test using Kalman’s method
For a system described by the state model,
ሶ
X(𝑡)= AX t +B U(t) ; where x(t) is of the order n*1
Y(𝑡)= CX t +D U(t)
to be completely state controllable, it is necessary and sufficient that the
following n*n controllability matrix has a rank ‘n’
𝑆𝑆𝐶 = 𝐵 𝐴𝐵 𝐴2 𝐵 . . 𝐴𝑛−1 𝐵
For the same system to be output controllable, it is necessary and sufficient
that the following m*(n+1) controllability matrix has a rank ‘m’ where m is
the number of outputs
𝑆𝑂𝐶 = 𝐶𝐵 𝐶𝐴𝐵 𝐶𝐴2 𝐵 . . 𝐶𝐴𝑛−1 𝐵 𝐷
System is observable if the observability matrix has the rank ‘n’
𝑆𝑜 = [𝐶 𝑇 𝐴𝑇 𝐶 𝑇 . . . 𝐴𝑇 𝑛−1 𝐶 𝑇 ]
Example 5
Test whether the system represented by the state model
𝑥1ሶ 0 0 1 𝑥1 0 𝑥1
𝑥2ሶ = −2 −3 0 𝑥2 + 2 𝑢 and y= [1 0 0] 𝑥2 is
𝑥3ሶ 0 2 −3 𝑥3 0 𝑥3
controllable and observable.

Solution:
n*n Controllability matrix
0 0 4
𝑆𝑆𝐶 OR 𝑄𝐶 = 𝐵 𝐴𝐵 𝐴2 𝐵 = 2 −6 18 Rank=3
0 4 −24
∴ System is completely controllable
Observability matrix
1 0 0
𝑆𝑜 or 𝑄𝑜 = [𝐶 𝑇 𝐴𝑇 𝐶 𝑇 𝐴𝑇 2 𝐶 𝑇 ]= 0 0 2 ; Rank=3
0 1 −3

∴ System is completely observable.

Example 6
Test whether the system represented by the system matrix
𝑥1ሶ 1 2 1 𝑥1 0 𝑥1
𝑥2ሶ = 0 1 0 𝑥2 + 0 𝑢 and y= [1 1 0] 𝑥2 is
𝑥3ሶ 1 −4 3 𝑥3 1 𝑥3
Controllable

Solution:
n*n controllability matrix
0 1 4
𝑆𝑆𝐶 OR 𝑄𝐶 = 𝐵 𝐴𝐵 𝐴2 𝐵 = 0 0 0 Rank=1
1 3 10
∴ System is not completely controllable
(It means only 1 out of 3 states are controllable)
Example 7
Test whether the system represented by the system matrix
𝑥1ሶ 1 1 𝑥1 0 𝑥1
= 𝑥 + 𝑢 and y= 1 0 𝑥 is
𝑥2ሶ −3 −2 2 1 2
Observable

Solution:
Observability matrix
1 1
𝑆𝑜 or 𝑄𝑜 = [𝐶 𝑇 𝑇 𝑇
𝐴 𝐶 ]= ; Rank=2 (complete rank)
0 1

∴ System is completely observable.

Example 8
Test whether the system represented by the state model
𝑥1ሶ 0 1 0 𝑥1 0 𝑥1
𝑥2ሶ = 0 0 1 𝑥2 + 0 𝑢 and y= [1 1 0] 𝑥2 is
𝑥3ሶ −6 −11 −6 𝑥3 1 𝑥3
observable.
Solution:
Observability matrix
1 0 −6
𝑆𝑜 or 𝑄𝑜 = [𝐶 𝑇 𝐴𝑇 𝐶 𝑇 𝐴𝑇 2 𝐶 𝑇 ]= 1 1 −11 ; Rank≠3
0 1 −5
∴ System is not completely observable
Example 9
Examine whether the system characterized by the differential equation
ሸ + 7𝑌ሷ + 12𝑌ሶ + 6𝑌 + 𝑈 = 0 is controllable and observable using
𝑌
Kalman’s method.

Solution:
Corresponding State Model can be written as
𝑥1ሶ 0 1 0 𝑥1 0 𝑥1
𝑥2ሶ = 0 0 1 𝑥2 + 0 𝑢 and y= [1 0 0] 𝑥2
𝑥3ሶ −6 −12 −7 𝑥3 −1 𝑥3
n*n controllability matrix
0 0 −1
𝑆𝑆𝐶 OR 𝑄𝐶 = 𝐵 𝐴𝐵 𝐴2 𝐵 = 0 −1 7 Rank=3
−1 7 −37
∴ System is completely controllable

Observability matrix
1 0 0
𝑆𝑜 or 𝑄𝑜 = [𝐶 𝑇 𝐴𝑇 𝐶 𝑇 𝐴𝑇 2 𝐶 𝑇 ]= 0 1 0 ; Rank=3
0 0 1
∴ System is completely observable.