CHAPTER ONE

STATE-SPACE MODELS
1.1. MODERN VS CLASSICAL CONTROL SYSTEM

• A control system can be viewed informally as a dynamical object (e.g.

expressed by an ordinary differential equation) containing a parameter that
can be manipulated to influence the behavior of the system to achieve the
desired goal.

• A control system consists of subsystems and processes (or plants) assembled

for the purpose of obtaining a desired output with desired performance,
given a specified input
LIMITATIONS OF CONVENTIONAL APPROACH

• It is insufficient and troublesome to give complete time domain

solutions of higher-order systems.
• It is not very much convenient for the analysis of MIMO systems.
• It is only applicable to LTI systems.
• The classical methods like Root locus, and Bode plot are trial and
error procedures which fail to give the optimal solution required.
• It gives an analysis of the system for some specific types of inputs
like Step, Ramp etc.
MODERN CONTROL SYSTEM

• It is necessary to use a method of analyzing systems that overcome the above

difficulties.

• Modern control uses the concept of the total internal state of the system
considering all initial conditions

• It uses the concept of state called State Variable Analysis or State Space
Analysis
ADVANTAGES OF STATE VARIABLE ANALYSIS
• It takes into account the effect of all initial conditions.
• It can be applied to nonlinear as well as time-varying systems.
• It can be conveniently applied to MIMO systems.
• Any type of input can be considered for designing the system
• It involves matrix algebra and can be conveniently adopted for
digital computers.
• The state variables selected need not be physical quantity.
• The vector-matrix notation simplifies mathematical representation.
1.2. CONCEPTS OF STATE
• The State of a dynamic system is the smallest set of variables (called state
variables) such that knowledge of these variables at 𝑡 = 𝑡0 , together with
knowledge of the input for 𝑡 ≥ 𝑡0 , completely determines the behavior of the
system for any time 𝑡 ≥ 𝑡0 .
• The state variables need not be physically measurable or observable
quantities. Such freedom in choosing state variables is an advantage of the
state-space methods.
• State Space is the n-dimensional space whose coordinate axes consist of the
𝑥1 , 𝑥2 …, 𝑥𝑛 axis, where 𝒙𝟏 , 𝒙𝟐 …, 𝒙𝒏 are state variables. Any state can be
represented by a point in the state space.
1.3. STATE-SPACE REPRESENTATION OF SYSTEMS
• In state-space analysis we are concerned with three types of variables that
are involved in the modeling of dynamic systems: input variables, output
variables, and state variables.
• Consider MIMO, nth order system
𝑢1 𝑡 𝑋1 𝑡
𝑢 𝑡 𝑋 𝑡
𝑢 𝑡 = 2 , X t = 2
⋮ ⋮
𝑢𝑚 𝑡 𝑋𝑛 𝑡
𝑌1 𝑡
𝑌 𝑡
Y 𝑡 = 2
⋮
𝑌𝑝 𝑡
All are column vectors having orders 𝑚 𝑥 1, 𝑛 𝑥 1 and 𝑝 𝑥 1 respectively.
CONT`D
• For such a system, the state variable representation can be arranged
in the form of ‘n’ first order differential equations.
𝑑𝑋1 𝑡
= 𝑥ሶ1 𝑡 = 𝑓1 𝑋1 , 𝑋2 , … 𝑋𝑛 , 𝑢1 , 𝑢2 … 𝑢𝑚
𝑑𝑡
𝑑𝑋2 𝑡
= 𝑥ሶ 2 𝑡 = 𝑓2 𝑋1 , 𝑋2 , … 𝑋𝑛 , 𝑢1 , 𝑢2 … 𝑢𝑚
𝑑𝑡
.
.
.
𝑑𝑋𝑛 𝑡
= 𝑥ሶ 𝑛 𝑡 = 𝑓𝑛 𝑋1 , 𝑋2 , … 𝑋𝑛 , 𝑢1 , 𝑢2 … 𝑢𝑚
𝑑𝑡
CONT`D
• Any ‘n’ dimensional time invariant system has state equations in the functional
form as,
𝑋ሶ 𝑡 = 𝑓 𝑋, 𝑢 … State eqaution
• Functional output equation can be written as,
𝑌 𝑡 = 𝑔 𝑋, 𝑢 … Output equation
• Diagrammatically this can be represented:-
CONT`D

• The functional equations can be expressed in terms of linear

combination of system states and the input as,
CONT`D

• For the LTI systems, the coefficients 𝑎𝑖𝑗 and 𝑏𝑖𝑗 are constants. Thus all the
equations can be written in vector matrix form as,

𝑋ሶ 𝑡 = 𝐴𝑋 𝑡 + 𝐵𝑢 𝑡

𝑋 𝑡 = State vector matrix of order n x 1

𝑈 𝑡 = Input vector matrix of order m x 1
𝐴 = System matrix or Evolution matrix of order n x n
𝐵 = Input matrix or Control matrix of order n x m
CONT`D
• Similarly the output variables at time t can be expressed as the linear
combinations of the input variables at time t as,

• For LTI system, the output equation can be written as

𝑌 𝑡 = 𝐶𝑋 𝑡 + 𝐷𝑢 𝑡
𝑌 𝑡 = Output vector matrix of order p x 1
𝐶 = Output matrix or Observation matrix of order p x n
𝐷 = Direct transmission matrix of order p x m
CONT`D
• The two vector equations together is called the state model of the linear
system.
𝑋ሶ 𝑡 = 𝐴𝑋 𝑡 + 𝐵𝑢 𝑡
𝑌 𝑡 = 𝐶𝑋 𝑡 + 𝐷𝑢 𝑡
EXAMPLE

Obtain the state model of the given electrical system

SOLUTION

• There are two energy-storing elements L and . So the two state variables are
current through inductor i(t) and voltage across capacitor 𝑣0 (𝑡)
𝑋1 𝑡 = 𝑖(𝑡), 𝑋2 𝑡 = 𝑣0 (𝑡) and U 𝑡 = 𝑣𝑖 𝑡 = Input variable
• Applying KVL to the loop
𝑑𝑖 𝑡
𝑣𝑖 𝑡 = 𝑖 𝑡 R + L + 𝑣0 (𝑡)
𝑑𝑡
• Arrange it for 𝑑𝑖𝑑𝑡𝑡
𝑑𝑖 𝑡 1 R 1
= 𝑣𝑖 𝑡 − 𝑖 𝑡 − 𝑣0 (𝑡)
𝑑𝑡 𝐿 𝐿 𝐿
CONT’D
𝑑𝑖 𝑡 R 1 1
= 𝑋ሶ1 𝑡 = − 𝑋1 𝑡 − 𝑋2 𝑡 + U 𝑡
𝑑𝑡 𝐿 𝐿 𝐿
1
While 𝑣0 𝑡 = Voltage across capacitor = 𝐶 ‫𝑡𝑑 𝑡 𝑖 ׬‬
𝑑𝑣0 𝑡 1
= 𝑖 𝑡
𝑑𝑡 𝐶
1
𝑋ሶ 2 𝑡 = 𝑋1 𝑡
𝐶
• The two equations can be written us
R 1
− − 1
𝑋ሶ1 𝑡 𝐿 𝐿 𝑋1 𝑡
= + 𝐿 U 𝑡
𝑋ሶ 2 𝑡 1 𝑋2 𝑡
0 0
𝐶
𝑋ሶ 𝑡 = 𝐴 𝑋 𝑡 + 𝐵 U 𝑡
CONT’D
• While the output variable Y t = 𝑣0 𝑡 = 𝑋2 𝑡
𝑋1 𝑡
𝑌(𝑡) = [0 1]
𝑋2 𝑡
𝑌(𝑡) = 𝐶 𝑋 𝑡
• This is the required state model. As n=2, it is second order system.

• 𝑋1 𝑡 can be 𝑣0 𝑡 and 𝑋2 𝑡 can be 𝑖 𝑡 due to which state model

matrices get changed. Hence state model is not the unique property
of the system.
QUIZ ONE

Obtain the state model of the given electrical network in the standard form.
1.4. TRANSFORMATION FROM STATE-SPACE REPRESENTATION
TO TRANSFER FUNCTION

• Given the state and output equations

𝑋ሶ 𝑡 = 𝐴𝑋 𝑡 + 𝐵𝑢 𝑡
𝑌 𝑡 = 𝐶𝑋 𝑡 + 𝐷𝑢 𝑡
• Take the Laplace transform assuming zero initial conditions:
𝑠𝑋 𝑠 = 𝐴𝑋 𝑠 + 𝐵𝑢 𝑠
𝑌 𝑠 = 𝐶𝑋 𝑠 + 𝐷𝑢 𝑠

• Solving for X(s) yields (𝑠𝐼 − 𝐴)𝑋 𝑠 = 𝐵𝑢 𝑠 or 𝑋 𝑠 = (𝑠𝐼 − 𝐴)−1 𝐵𝑢 𝑠

where I is the identity matrix.
CONT’D

• Solving for Y(s) yields

Y 𝑠 = 𝐶𝑋 𝑠 + 𝐷𝑢 𝑠 = 𝐶[(𝑠𝐼 − 𝐴)−1 𝐵𝑢 𝑠 ] + 𝐷𝑢 𝑠
Y 𝑠 = [𝐶(𝑠𝐼 − 𝐴)−1 𝐵 + 𝐷]𝑢 𝑠

• We call the matrix [𝐶(𝑠𝐼 − 𝐴)−1 𝐵 + 𝐷] the transfer function matrix, since it
relates the output vector, Y 𝑠 , to the input vector, 𝑢 𝑠 . However, if 𝑢 𝑠 and
Y 𝑠 are scalars, we can find the transfer function,

Y 𝑠
𝑇 𝑠 = = 𝐶(𝑠𝐼 − 𝐴)−1 𝐵 + 𝐷
𝑢 𝑠
EXAMPLE

• Find the transfer function from the following state model.

0 1 0 10
𝑋ሶ = 0 0 1 𝑋+ 0 𝑢
−1 −2 −3 0
𝑌 = [1 0 0]𝑋
SOLUTION

Y 𝑠
𝑇 𝑠 = = 𝐶(𝑠𝐼 − 𝐴)−1 𝐵 + 𝐷
𝑢 𝑠

𝑠 0 0 0 1 0 𝑠 −1 0
𝑠𝐼 − 𝐴 = 0 𝑠 0 − 0 0 1 = 0 𝑠 −1
0 0 𝑠 −1 −2 −3 1 2 𝑠+3

𝑠 2 + 3𝑠 + 2 𝑠+3 1
−1 𝑠(𝑠 + 3) 𝑠
adj(sI − A) −𝑠 −(2𝑠 + 1) 𝑠 2
(𝑠𝐼 − 𝐴)−1 = =
det(sI − A) 𝑠 3 + 3𝑠 2 + 2𝑠 + 1
CONT’D

10
𝐶 = [1 0 0] 𝐵= 0 𝐷=0
0

• Substituting (𝑠𝐼 − 𝐴)−1 , 𝐶, 𝐵 and 𝐷 results the following transfer function.

10(𝑠 2 + 3𝑠 + 2)
𝑇(𝑠) = 3
𝑠 + 3𝑠 2 + 2𝑠 + 1
EXERCISE

• Convert the following state and output equations to transfer function

−4 −1.5 2
𝑋ሶ = 𝑋+ 𝑢
4 0 0

𝑌 = [1.5 0.625]𝑋
1.5. STATE DIAGRAM REPRESENTATION

• We can obtain a state space model using phase variables

• Phase variables are state variables obtained by assuming one of the system
variables as a state variable and other state variables are the derivatives of
the selected system variable.
• Consider a linear continuous time system represented by 𝑛𝑡ℎ order differential
equation as
𝑌 𝑛 + 𝑎𝑛−1 𝑌 𝑛−1 + 𝑎𝑛−2 𝑌 𝑛−2 + ⋯ + 𝑎1 𝑌ሶ + 𝑎0 𝑌 𝑡
= 𝑏0 𝑢 + 𝑏1 𝑢ሶ + ⋯ + 𝑏𝑚−1 𝑢𝑚−1 + 𝑏𝑚 𝑢𝑚
CONT’D

𝑑𝑌 𝑛 (𝑡)
• 𝑌𝑛 𝑡 =
𝑑𝑡 𝑛
= 𝑛𝑡ℎ derivative of Y(t)

• Consider the simple case of the system in which derivatives of the control
input 𝒖(𝐭) are absent.
𝑌 𝑛 + 𝑎𝑛−1 𝑌 𝑛−1 + ⋯ + 𝑎1 𝑌ሶ + 𝑎0 𝑌 𝑡 = 𝑏0 𝑢(𝑡)

• The choice of the state variable is generally the output variable Y(t) itself. And
other state variables are derivatives of the selected state variable Y(t)
CONT’D

𝑋1 𝑡 = 𝑌 𝑡
𝑋2 𝑡 = 𝑌ሶ 𝑡 = 𝑋ሶ1 𝑡
𝑋3 𝑡 = 𝑌ሷ 𝑡 = 𝑋ሷ1 𝑡 = 𝑋ሶ 2 𝑡
Thus the various state equations are
𝑋ሶ1 𝑡 = 𝑋2 𝑡
𝑋ሶ 2 𝑡 = 𝑋3 𝑡
𝑋ሶ 𝑛−1 𝑡 = 𝑋𝑛 𝑡
• 𝑋ሶ 𝑛
𝑡 is to be obtained by substituting the selected state variables in 𝑌 𝑛 +
𝑎𝑛−1 𝑌 𝑛−1 + 𝑎𝑛−2 𝑌 𝑛−2 + ⋯ + 𝑎1 𝑌ሶ + 𝑎0 𝑌 𝑡 = 𝑏0 𝑢(𝑡)
CONT’D

• We have
𝑌 𝑡 = 𝑋1 𝑡
𝑌ሶ 𝑡 = 𝑋2 𝑡
𝑌ሷ 𝑡 = 𝑋3 𝑡
𝑌 𝑛−1 (𝑡) = 𝑋𝑛 𝑡
𝑌 𝑛 𝑡 = 𝑋ሶ 𝑛 𝑡
Thus
𝑋ሶ 𝑛 𝑡 + 𝑎𝑛−1 𝑋𝑛 𝑡 + 𝑎𝑛−2 𝑋𝑛−1 𝑡 + ⋯ + 𝑎1 𝑋2 𝑡 + 𝑎0 𝑋1 𝑡 = 𝑏0 𝑢(𝑡)
CONT’D

• The model can be shown in state diagram as

• The output of each integrator is a state variable

EXAMPLE

• Construct the state model using phase variables if the system is described by
the differential equation.
𝑑3 𝑌 (𝑡) 𝑑2 𝑌 (𝑡) 𝑑𝑌 (𝑡)
3
+4 2
+7 + 2𝑌 𝑡 = 5𝑢(𝑡)
𝑑𝑡 𝑑𝑡 𝑑𝑡
• Draw the state diagram
SOLUTION
• Choose the output Y(t) as state variable 𝑋1
𝑡
𝑋1 𝑡 = 𝑌(𝑡)
𝑑𝑌 (𝑡)
= 𝑋ሶ1 𝑡 = 𝑌ሶ 𝑡 =
𝑑𝑡
𝑑2 𝑌 (𝑡)
𝑋3 𝑡 = 𝑋ሶ 2 𝑡 = 𝑌ሷ 𝑡 =
𝑑𝑡 2
• Thus 𝑋ሶ1 𝑡 = 𝑋2 𝑡 and 𝑋ሶ 2 𝑡 = 𝑋3 𝑡
• To obtain 𝑋ሶ 3 𝑡 substitute state variables obtained in the differential equation
𝑑3 𝑌(𝑡) 𝑑 𝑌ሷ 𝑡
ሸ 𝑡 =
=𝑌 = 𝑋ሶ 3 𝑡
𝑑𝑡 3 𝑑𝑡
CONT’D
𝑋ሶ 3 𝑡 + 4𝑋3 𝑡 + 7𝑋2 𝑡 + 2𝑋1 𝑡 = 5𝑢(𝑡)
𝑋ሶ 3 𝑡 = −2𝑋1 𝑡 − 7𝑋2 𝑡 − 4𝑋3 𝑡 + 5𝑢(𝑡)
• The state diagram will be
1.6. NON UNIQUENESS OF STATE VARIABLE MODEL

• Consider a standard state model for a system

𝑋ሶ 𝑡 = 𝐴𝑋 𝑡 + 𝐵𝑢 𝑡
𝑌 𝑡 = 𝐶𝑋 𝑡 + 𝐷𝑢 𝑡
• Let 𝑋 = 𝑀𝑍 where 𝑀 is a non singular constant matrix.
• This means 𝑍 is a new set of state variables which is obtained by the linear
combination of the original state variables
CONT’D

𝑋ሶ 𝑡 = 𝐴𝑋 𝑡 + 𝐵𝑢 𝑡 ⇒ 𝑀𝑍ሶ 𝑡 = 𝐴𝑀𝑍 𝑡 + 𝐵𝑢 𝑡
𝑌 𝑡 = 𝐶𝑋 𝑡 + 𝐷𝑢 𝑡 = 𝐶𝑀𝑍 𝑡 + 𝐷𝑢 𝑡
• Premultiplying by 𝑀−1 yields 𝑍ሶ 𝑡 = 𝑀−1 𝐴𝑀𝑍 𝑡 + 𝑀−1 𝐵𝑢 𝑡
መ 𝑡 + 𝐵𝑢
𝑍ሶ 𝑡 = 𝐴𝑍 ෠ 𝑡
መ 𝑡 + 𝐷𝑢 𝑡
𝑌 𝑡 = 𝐶𝑍
Where 𝐴መ = 𝑀−1 𝐴𝑀, 𝐵෠ = 𝑀−1 𝐵 and 𝐶መ = 𝐶𝑀
• We for a new state model of the system. This shows that the state model is not
a unique property.
1.7. EIGEN VALUES AND EIGENVECTORS

• The roots of the characteristic's equation of 𝑛 𝑥 𝑛

matrix A, which is defined
as det λI − A = 0 are called eigenvalues of the matrix A.

• Any nonzero vectors 𝑣𝑖 which satisfy the matrix equation [λ𝑖 I − A]𝑣𝑖 = 0;
where λ𝑖 are eigenvalues of the matrix are called eigenvector of matrix A
corresponding to eigenvalue λ𝑖 respectively.
EXAMPLE

• Find the eigenvalues and eigenvector of the matrix

2 2
𝐴=
5 −1
SOLUTION

• The eigenvalues are those λ for which det λI − A = 0

2 2 λ 0 2−λ 2
det 𝐴 − λI = det − =
5 −1 0 λ 5 −1 − λ

= 2 − λ −1 − λ − 10
λ2 − λ − 12 = 0
The eigenvalues of A are the solutions of the quadratic equation λ2 − λ − 12 =
0, namely λ1 = −3 and λ2 = 4