Chapter 3 Modeling in the Time Domain

Modeling in the Time Domain

Introduction
The state-space approach (also referred to as the modern, or time-domain, approach) is a unified
method for modeling, analyzing, and designing a wide range of systems.
The state-space approach can be used to represent nonlinear systems that have backlash, saturation,
and dead zone.
It can handle, conveniently, systems with nonzero initial conditions.
Time-varying systems can be represented in state space.
Multiple-input, multiple-output systems can be compactly represented in state space with a model
similar in form and complexity to that used for single-input, single-output systems.
It can be used to represent systems with a digital computer in the loop or to model systems for
digital simulation. With a simulated system, system response can be obtained for changes in system
parameters—an important design tool.
It is also attractive because of the availability of numerous state-space software packages for the
personal computer.
While the state-space approach can be applied to a wide range of systems, it is not as intuitive as the
classical approach.
The designer has to engage in several calculations before the physical interpretation of the model is
apparent, whereas in classical control a few quick calculations or a graphic presentation of data
rapidly yields the physical interpretation.

3.2 Some Observations

We now demonstrate that for a system (as shown in Figure 3.1) with many variables, such as
inductor voltage, vL(t), resistor voltage, vR(t), and capacitor charge, q (t), we need to use differential
equations only to solve for a selected subset of system variables because all other remaining system
variables can be evaluated algebraically from the variables in the subset. Our examples take the
following approach:
1. We select a particular subset of all possible system variables and call the variables in these
subset state variables.
2. or an nth-order system, we write n simultaneous, first-order differential equations in terms
of the state variables. We call this system of simultaneous differential equations state
equations.
3. If we know the initial condition of all of the state variables at t0 as well as the system input
for t  t0, we can solve the simultaneous differential equations for the state variables for t 
t0 .
4. We algebraically combine the state variables with the system’s input and find all of the
other system variables for t  t0. We call this algebraic equation the output equation.
5. We consider the state equations and the output equations a viable representation of the
system. We call this representation of the system a state-space representation.

Observation with First Order System

Let us now follow these steps through an example. Consider the RL network shown in Figure 3.1
with an initial current of i(0).
1. In this system the variables are i(t), vL(t), and vR(t). We select the current, i(t), for which we will
write and solve a differential equation using Laplace transforms.
2. We write the loop equation,

Prepared by: Dr. Mohammad Abdul Mannan Page 1 of 27

 Chapter 3 Modeling in the Time Domain

di (t )
L  Ri(t )  v (t ) (3.1)
dt

Figure 3.1 RL network

3. Taking the Laplace transform, using Table 2.2, Item 7, and including the initial conditions, yields
LsI ( s)  i (0)  RI ( s)  V ( s) (3.2)
Assuming the input, v(t), to be a unit step, u(t), whose Laplace transform is V(s)=1/s, we solve for
I(s) and get

Ls  R I ( s)  1  Li (0)
s
1 1 i(0)
I ( s)  
L  R R
s s   s 
 L L
 

1 1 1  i(0)
I ( s)    (3.3)
Rs R R
 s  s
 L L
from which
1
 
i (t )  1  e  ( R / L)t  i (0)e  ( R / L)t
R
(3.4)

The function i(t) is a subset of all possible network variables [i(t), vL(t), and vR(t)] that we are able
to find from Eq. (3.4) if we know its initial condition, i(0), and the input, v(t). Thus, i(t) is a state
variable, and the differential equation (3.1) is a state equation.

4. We can now solve for all of the other network variables algebraically in terms of i(t) and the
applied voltage, v(t). For example, the voltage across the resistor is
 
v R (t )  Ri(t )  1  e  ( R / L )t  Ri(0)e  ( R / L )t (3.5)
The voltage across the inductor is
v L (t )  v(t )  v R (t )  v(t )  Ri(t ) (3.6)
The derivative of the current is
di 1
 v (t )  Ri(t ) (3.7)
dt L

Thus, knowing the state variable, i(t), and the input, v(t), we can find the value, or state, of any
network variable at any time, t  t0. Hence, the algebraic equations, Eqs. (3.5) through (3.7), are
output equations.

Prepared by: Dr. Mohammad Abdul Mannan Page 2 of 27

 Chapter 3 Modeling in the Time Domain

5. Since the variables of interest are completely described by Eq. (3.1) and Eqs. (3.5) through (3.7),
we say that the combined state equation (3.1) and the output equations (3.5 through 3.7) form a
viable representation of the network, which we call a state-space representation.

Equation (3.1), which describes the dynamics of the network, is not unique. This equation could be
written in terms of any other network variable. For example, substituting i(t) = vR(t)/R into Eq. (3.1)
yields
L dv R (t )
 v R (t )  v (t ) (3.8)
R dt
which can be solved knowing that the initial condition vR(0) = Ri(0) and knowing v(t). In this case,
the state variable is vR(t). Similarly, all other network variables can now be written in terms of the
state variable, vR(t), and the input, v(t).

Observation with Second Order System

Let us now extend our observations to a second-order system, such as that shown in Figure 3.2.

Figure 3.2 RLC network

1. In this system the variables are i(t), vL(t), vC(t), q(t), and vR(t). Since the network is of second
order, two simultaneous, first-order differential equations are needed to solve for two state
variables. We select i(t) and q(t), the charge on the capacitor, as the two state variables.
2. Writing the loop equation yields
di(t ) 1
L
dt
 Ri(t ) 
C 
i (t )  v (t ) (3.9)

Converting to charge, using i(t)=dq(t)/dt, we get

d 2 q (t ) dq (t ) 1
L R  q (t )  v (t ) (3.10)
dt 2 dt C
But an nth-order differential equation can be converted to n simultaneous first-order differential
equations, with each equation of the form
dxi (t )
 ai1 x1 (t )  ai 2 x 2 (t )  .........  ain x n (t )  bi f (t ) (3.11)
dt
where each xi(t) is a state variable, and the aij’s and bi are constants for linear, time-invariant
systems. We say that the right-hand side of Eq. (3.11) is a linear combination of the state variables
and the input, f(t).

Prepared by: Dr. Mohammad Abdul Mannan Page 3 of 27

 Chapter 3 Modeling in the Time Domain

We can convert Eq. (3.10) into two simultaneous, first-order differential equations in terms of i(t)
and q(t). The first equation can be dq(t)/dt=i(t).The second equation can be formed by substituting
i(t)dt=q(t) into Eq. (3.9) and solving for di(t)/dt. Summarizing the two resulting equations, we get
dq (t )
 i (t ) (3.12a)
dt
di(t ) 1 R 1
 q (t )  i (t )  v (t ) (3.12b)
dt LC L L

3. These equations are the state equations and can be solved simultaneously for the state variables,
q(t) and i(t), using the Laplace transform, if we know the initial conditions for q(t) and i(t) and if we
know v(t), the input.

4. From these two state variables, we can solve for all other network variables. For example, the
voltage across the inductor can be written in terms of the solved state variables and the input as
1
v L (t )  v(t )  Ri(t )  q (t ) (3.13b)
C
Equation (3.13) is an output equation; we say that vL(t) is a linear combination of the state variables,
q(t) and i(t), and the input, v(t).

5. The combined state equations (3.12) and the output equation (3.13b) form a viable representation
of the network, which we call a state-space representation.

Another choice of two state variables can be made, for example, vR(t) and vC(t), the resistor and
capacitor voltage, respectively. The resulting set of simultaneous, first-order differential equations
follows:
dv R (t ) R R R
  v R (t )  vC (t )  v(t ) (3.14a)
dt L L L
dvC (t ) 1
 v R (t ) (3.14b)
dt RC

Again, these differential equations can be solved for the state variables if we know the initial
conditions along with v(t). Further, all other network variables can be found as a linear combination
of these state variables.

Is there a restriction on the choice of state variables? Yes! Typically, the minimum number of state
variables required to describe a system equals the order of the differential equation. Thus, a second-
order system requires a minimum of two state variables to describe it.

We can define more state variables than the minimal set; however, within this minimal set the state
variables must be linearly independent. For example, if vR(t) is chosen as a state variable, then i(t)
cannot be chosen, because vR(t) can be written as a linear combination of i(t), namely vR(t)=Ri(t).
Under these circumstances we say that the state variables are linearly dependent. State variables
must be linearly independent; that is, no state variable can be written as a linear combination of the
other state variables, or else we would not have enough information to solve for all other system
variables, and we could even have trouble writing the simultaneous equations themselves.

Prepared by: Dr. Mohammad Abdul Mannan Page 4 of 27

 Chapter 3 Modeling in the Time Domain

The state and output equations can be written in vector-matrix form if the system is linear. Thus,
Eq. (3.12), the state equations, can be written as
dx (t )
 x (t )  Ax (t )  Bu(t ) (3.15)
dt
 dq (t ) 
 q(t ) dx (t )  dt   0 1  0
Where, x (t )    ;  
x (t )   di (t )  ; u (t )  v ( t ) ; A   1 R ; B  1
 i (t )  dt    LC  
L  L 
 dt 
Equation (3.13), the output equation, can be written as
y (t )  Cx (t )  Du(t ) (3.16)
 1 
Where, y(t )  v L (t ) ; C    R ; D  1
 C 
We call the combination of Eqs. (3.15) and (3.16) a state-space representation of the network of
Figure 3.2. A state-space representation, therefore, consists of (1) the simultaneous, first-order
differential equations from which the state variables can be solved and (2) the algebraic output
equation from which all other system variables can be found. A state-space representation is not
unique, since a different choice of state variables leads to a different representation of the same
system.

3.3 The General State-Space Representation

Linear Combination: A linear combination of n variables, xi, for i=1 to n, is given by the following
sum, S:
S  K n x n  K n 1 X n 1  ............K1 x1 (3.17)

Linear Independence: A set of variables is said to be linearly independent if none of the variables
can be written as a linear combination of the others.
For example, given x1, x2, and x3, if x2=5x1+6x3, then the variables are not linearly independent,
since one of them can be written as a linear combination of the other two.
Now, what must be true so that one variable cannot be written as a linear combination of the other
variables?
Consider the example K2x2=K1x1+K3x3. If no xi=0, then any xi can be written as a linear combination
of other variables, unless all Ki=0. Formally, then, variables xi, for i=1 to n, are said to be linearly
independent if their linear combination, S, equals zero only if every Ki=0 and no xi=0 for all t  0.

System Variable: Any variable that responds to an input or initial conditions in a system.

State Variables: The smallest set of linearly independent system variables such that the values of
the members of the set at time t0 along with known forcing functions completely determine the
value of all system variables for all t  t0.

State Vector: A vector whose elements are the state variables.

State Space: The n-dimensional space whose axes are the state variables.

State Equations: A set of n simultaneous, first-order differential equations with n variables, where
the n variables to be solved are the state variables.

Prepared by: Dr. Mohammad Abdul Mannan Page 5 of 27

 Chapter 3 Modeling in the Time Domain

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

Now that the definitions have been formally stated, we define the state-space representation of a
system. A system is represented in state space by the following equations:
dx (t )
 x (t )  Ax (t )  Bu(t ) (3.18)
dt
y (t )  Cx (t )  Du(t ) (3.19)
For t  t0 and initial conditions. x(t0), where
x(t): State vector (xRn i.e. n-dimensional state vector)
y(t): Output vector (yRm i.e. m-dimensional output vector)
u(t): Input or control vector (uRr i.e. r-dimensional input or control vector)
A: System or plant parameter matrix (an nn system or plant parameter matrix)
B: Input or control matrix (an nr input or control matrix)
C: Output matrix (an mn output parameter matrix)
D: Feed forward Matrix (an mr feed forward parameter matrix)
 x1 (t )   u1 (t )   y1 (t ) 
 x (t )  u (t )  y (t )   a11 a12 ....... a1n 
 2   2   2  a 
 21 a 22 ....... a 2 n 
 ...   ...   ... 
x (t )   ; u(t )    y (t )   ; A   ... ... ....... ...  ;
 ...   ...   ...   
 ...   ...   ...   ... ... ....... ... 
       a n1 a n 2 ....... a nn 
 x n (t ) u r (t )   y m (t )
 b11 b12 ....... b1r   c11 c12 ....... c1n   d11 d12 ....... d1r 
b
 21 b22 ....... b2r  c
 21 c 22 ....... c 2n  d
 21 d 22 ....... d 2r 
B   ... ... ....... ...  ; C   ... ... ....... ...  ; D   ... ... ....... ... 
     
 ... ... ....... ...   ... ... ....... ...   ... ... ....... ... 
bn1 bn 2 ....... bnr  c m1 cm2 ....... c mn  d m1 d m2 ....... d mr 

The parameters (elements) aij, bij, cij, and dij of the matrices A, B, C, and D are constant for the
linear-time-invariant (LTI) system. If A is a constant parameter matrix and no input is applied
(u=0), the state equation represents a homogeneous or autonomous linear system. On the other
hand if input is applied (u0), the system is said to be non-homogeneous.

Equation (3.18) is called the state equation, and the vector x, the state vector, contains the state
variables. Equation (3.18) can be solved for the state variables. Equation (3.19) is called the output
equation. This equation is used to calculate any other system variables. This representation of a
system provides complete knowledge of all variables of the system at any t  t0.
Equations (3.18) may be expressed in terms of n first-order differential equations as follows:
x1 (t )  a11 x1 (t )  a12 x1 (t )  ...  a1n x n (t )  b11u1 (t )  b12u1 (t )  ....  b1r u r (t )
x 2 (t )  a 21 x1 (t )  a 22 x1 (t )  ...  a 2n x n (t )  b21u1 (t )  b22 u1 (t )  ....  b2r u r (t )
………………
…………….
x n (t )  a n1 x1 (t )  a n 2 x1 (t )  ...  a nn x n (t )  bn1u1 (t )  bn 2 u1 (t )  ....  bnr u r (t )

The m outputs of the system at time t. may be expressed as linear combination of input and state
variables as follows:
Prepared by: Dr. Mohammad Abdul Mannan Page 6 of 27
 Chapter 3 Modeling in the Time Domain

y1 (t )  c11 x1 (t )  c12 x1 (t )  ...  c1n x n (t )  d11u1 (t )  d12 u1 (t )  ....  d1r u r (t )

y 2 (t )  c 21 x1 (t )  c 22 x1 (t )  ...  c 2n x n (t )  d 21u1 (t )  d 22u1 (t )  ....  d 2r u r (t )
………..
………..
y m (t )  c m1 x1 (t )  c m 2 x1 (t )  ...  c mn xn (t )  d m1u1 (t )  d m 2 u1 (t )  ....  d mr u r (t )

The Du(t) in Eq. (3.19) indicates a direct coupling of the input to the output of the system. Since
this type of direct coupling of the input to the output is rare, the system may described the following
equations:
x (t )  Ax (t )  Bu(t ) (3.18a)
y (t )  Cx (t ) (3.19a)
The system (3.18) and (3.19) represent a multi-input and multi-output (MIMO) system. The
structure of a MIMO system is shown in the following figure.

Figure Structure of MIMO system.

Often the systems have single-input and single-output (SISO). The state variable representation of
SISO system becomes:
x (t )  Ax (t )  bu (t ) (3.18b)
y (t )  cx (t ) (3.19b)
Where, the input u(t) and output y(t) are scalars, and b and c are n-dimensional vectors.

To obtain the internal states, that is the state variables, integrators could be employed in the system
structure. The number of integrators involved in the system would be equal to the number of state
variables. The time domain state variables representation of a MIMO and SISO system is shown in
the following figure.

Why State Variable Approach?

An nth order differential equation is not suitable for solution. It is best to obtain a set of n first order
differential equations using a set of arbitrary variables called state variables. We need to evaluate
the merits of state space approach. The special features of this approach are:
 It can handle models both SISO and MIMO in a unified framework.
 Nonlinear as well as time varying systems can be analyzed conveniently.

Prepared by: Dr. Mohammad Abdul Mannan Page 7 of 27

 Chapter 3 Modeling in the Time Domain

 The initial conditions can be easily incorporated in the state model, whereas transfer
functions are defined for zero initial conditions.
 Unlike transfer functions the system model includes the internal states variables as
well.
 It allows us to design a system with an optimal performance.
 The state model provides a time domain solution, which is of ultimate interest in
many situations.
 The state variable formulation in terms of matrix/vector modeling is very efficient
from a computational standpoint.
 In view of the mathematical elegance also the state variable method has been
adapted for analysis and design.

3.4 Applying the State-Space Representation

The first step in representing a system is to select the state vector, which must be chosen according
to the following considerations:
1. A minimum number of state variables must be selected as components of the state vector.
This minimum number of state variables is sufficient to describe completely the state of the
system.
2. The components of the state vector (that is, this minimum number of state variables) must
be linearly independent.

Example 3.1: Given the electrical network of Figure 3.5, find a state-space representation if the
output is the current through the resistor.

Figure 3.5 Electrical network for representation in state space.

Solution: The following steps will yield a viable representation of the network in state space.

Step 1: Label all of the branch currents in the network. These include iL, iR, and iC, as shown in
Figure 3.5.

Step 2: Select the state variables by writing the derivative equation for all energy-storage elements,
that is, the inductor and the capacitor. Thus,
dv
C C  iC (3.22)
dt
di
L L  vL (3.23)
dt
From Eqs. (3.22) and (3.23), choose the state variables as the quantities that are differentiated,
namely vC and iL. Using Eq. (3.20) as a guide, we see that the state-space representation is complete
if the right-hand sides of Eqs. (3.22) and (3.23) can be written as linear combinations of the state
variables and the input.
Since iC and vL are not state variables, our next step is to express iC and vL as linear combinations of
Prepared by: Dr. Mohammad Abdul Mannan Page 8 of 27
 Chapter 3 Modeling in the Time Domain

the state variables, vC and iL, and the input, v(t).

Step 3: Apply network theory, such as Kirchhoff’s voltage and current laws, to obtain iC and vL in
terms of the state variables, vC and iL. At Node 1,
1
iC  i R  i L   vC  i L (3.24)
R
which yields iC in terms of the state variables, vC and iL.
Around the outer loop,
v L  vC  v(t ) (3.25)
which yields vL in terms of the state variable, vC, and the source, v(t).

Step 4: Substitute the results of Eqs. (3.24) and (3.25) into Eqs. (3.22) and (3.23) to obtain the
following state equations:
dv 1
C C   vC  i L (3.26a)
dt R
di
L L  vC  v (t ) (3.26b)
dt
or
dvC 1 1
 vC  i L (3.27a)
dt RC C
di L 1 1
  vC  v (t ) (3.27b)
dt L L

Step 5: Find the output equation. Since the output is iR(t),

1
i R   vC (3.28)
R
The final result for the state-space representation is found by representing Eqs. (3.27) and (3.28) in
vector-matrix form as follows:
 1 1
v 
v  0
d  C  C
 
 RC C  vC    1  v(t ) (3.29a)
 
dt  i L   iL    1  
0   i L   L 
 L 
1  v 
iR   0  C  (3.29b)
R   iL 
where the dot indicates differentiation with respect to time.

Example 3.2: Find the state and output equations for the electrical network shown in Figure 3.6 if
the output vector is y=[vR2 iR2]T, where T means transpose.

Solution: Immediately notice that this network has a voltage-dependent current source.

Step 1 Label all of the branch currents on the network, as shown in Figure 3.6.

Step 2 Select the state variables by listing the voltage-current relationships for all of the energy-
storage elements:
di
L L  vL (3.30a)
dt
Prepared by: Dr. Mohammad Abdul Mannan Page 9 of 27
 Chapter 3 Modeling in the Time Domain

dvC
C  iC (3.30b)
dt

From Eqs. (3.30) select the state variables to be the differentiated variables. Thus, the state
variables, x1 and x2, are
x1  i L ; x 2  vC (3.31)

Figure 3.6 Electrical network for Example 3.2.

Step 3 Remembering that the form of the state equation is

x  Ax  Bu (3.32)
we see that the remaining task is to transform the right-hand side of Eq. (3.30) into linear
combinations of the state variables and input source current. Using Kirchhoff’s voltage and current
laws, we find vL and iC in terms of the state variables and the input current source.

Around the mesh containing L and C,

v L  vC  v R 2  vC  i R 2 R2 (3.33)

But at Node 2, i R 2  iC  4v L . Substituting this relationship for iR2 into Eq. (3.33) yields
v L  vC  iC  4v L R2 (3.34)

Solving for vL, we get

1
vL  vC  iC R2  (3.35)
1  4 R2

Notice that since vC is a state variable, we only need to find iC in terms of the state variables. We
will then have obtained vL in terms of the state variables.

Thus, at Node 1 we can write the sum of the currents as

vR v
iC  i (t )  i R1  i L  i(t )  1  i L  i (t )  L  i L (3.36)
R1 R1

where vR1=vL. Equations (3.35) and (3.36) are two equations relating vL and iC in terms of the state
variables iL and vC. Rewriting Eqs. (3.35) and (3.36), we obtain two simultaneous equations
yielding vL and iC as linear combinations of the state variables iL and vC:
1  4 R2 v L  R2 iC  vC (3.37a)
1
 v L  iC  i L  i (t ) (3.37b)
R1

Solving Eq. (3.37) simultaneously for vL and iC yields

Prepared by: Dr. Mohammad Abdul Mannan Page 10 of 27
 Chapter 3 Modeling in the Time Domain

1
vL  R2 i L  vC  R2 i(t ) (3.38)

And
1 1 
iC  1  4 R2 i L  vC  1  4 R2 i (t ) (3.39)
 R1 
Where
 R2 
  1  4 R2    (3.40)
 R1 

Substituting Eqs. (3.38) and (3.39) into (3.30), simplifying, and writing the result in vector-matrix
form renders the following state equation:
 R2 1   R2 
 
 iL   L  i
L  L    L i (t )
     1  4 R2       (3.41)
 vC   1  4 R2  
1
vC  
 C R1C   C 

Step 4 Derive the output equation. Since the specified output variables are vR2 and iR2, we note that
around the mesh containing C, L, and R2,
v R 2  v C  v L (3.42a)
i R2  iC  4v L (3.42b)

Substituting Eqs. (3.38) and (3.39) into Eq. (3.42), vR2 and iR2 are obtained as linear combinations of
the state variables, iL and vC. In vector-matrix form, the output equation is
 R2  1   R2 
 1  
v
 R2        i L    
i    1 
1  4 R2  vC   1 
i (t ) (3.43)
 R2    
  R1    

Example 3.3: Find the state equations for the translationalmechanical system shown in Figure 3.7.

Figure 3.7 Translational mechanical system

Solution: First write the differential equations for the network in Figure 3.7, using the methods of
Laplace-transformed equations of motion. Next take the inverse Laplace transform of these
equations, assuming zero initial conditions, and obtain
d 2 x1 dx
M1  D 1  Kx1  Kx2  0 (3.44)
dt 2 dt

Prepared by: Dr. Mohammad Abdul Mannan Page 11 of 27

 Chapter 3 Modeling in the Time Domain

d 2 x2
 Kx1  M 2  Kx 2  f (t ) (3.45)
dt 2
Now let
dx1 dx2 d 2 x1 dv1 d 2 x 2 dv 2
 v1 ;  v2 ;  ; 
dt dt dt 2 dt dt 2 dt
and then select x1, v1, x2, and v2 as state variables. Next form two of the state equations by solving
Eq. (3.44) for dv1/dt and Eq. (3.45) for dv2/dt.
Finally, add dx1/dt= v1 and dx2/dt= v2 to complete the set of state equations. Hence,
dx1
 v1 (3.46a)
dt
dv1 K D K
 x1  v1  x2 (3.46b)
dt M1 M1 M1
dx 2
 v2 (3.46c)
dt
dv2 K K 1
 x1  x2  f (t ) (3.46d)
dt M2 M2 M2

In vector-matrix form,
 0 1 0 0  0 
 x1   K D K   x1  
 v    0  
v  0 
1 M M M 1
  1 1 1    f (t ) (3.47)
 x 2   0 0 0 1  x 2   0 
   K K    1 
 v2   M 0  0   v2   
 2 M 2 
 M2 
where the dot indicates differentiation with respect to time.
What is the output equation if the output are x1(t) and x2(t)?
 x1 
 
 x1  1 0 0 0  v1 
 x   0 0 1 0  x  (3.47a)
 2   2
 
v2 

4.10 Laplace Transform Solution of State Equations

Taking the Laplace transformation in both-sides Eq. (3.18.1) can be written as follows:
sX ( s )  x (0)  AX ( s )  BU ( s ) (3.18.1)
Solving Eq. (3.18.7) for X(s), we get
X (s)  sI  A-1 x (0)  sI  A-1 BU ( s) (4.96a)
X (s)  sI  A-1  x (0)  BU ( s) (4.96b)
adjsI  A
X ( s)  x(0)  BU (s) (4.96c)
det sI  A 
x(t) can be obtained by taking inverse Laplace transform of Eq. (3.18.8).

Prepared by: Dr. Mohammad Abdul Mannan Page 12 of 27

 Chapter 3 Modeling in the Time Domain

Example 4.11: Given the system represented in state space by Eqs. (4.99),
 0 1 0 0

x   0 0 1  x  0 e  t
 (4.99a)
  24  26  9 1 

y  1 1 0x (4.99b)

1 
x 0   0 (4.99c)
2
do the following:
(a) Solve the preceding state equation and obtain the output for the given exponential input.
(b) Find the eigenvalues and the system poles.

Solution:
(a) We will solve the problem by finding the component parts of Eq. (4.96), followed by
substitution into Eq. (4.97). First obtain A and B by comparing Eq. (4.99a) to Eq. (4.92). Since
 s 1 0 
sI  A   0 s  1 
 (4.101)
 24 26 s  9
and
 s 2  9 s  26 s9 1
1  
sI  A1  3   24 s 2  9s s (4.102)
s  9s 2  26 s  24 
 24 s  26s  24  s 2 
 

Since U(s) (the Laplace transform for e-t) is 1=(s + 1); X(s) can be calculated. From Eq. (4.96b) as
 X 1 (s )   s 2  9 s  26 s9 1   1  0 
  1  2      1 
X (s )   X 2 ( s )    24 s  9s s   0  0 (4.103)
s 3
 9 s 2
 26 s  24  2  s  1
 X 3 ( s)    24s  26s  24 s   2 1 

s 3  10s 2  37s  29
X 1 ( s)  (4.104a)
( s  1)(s  2)(s  3)(s  4)
2 s 2  21s  24
X 2 (s)  (4.104b)
( s  1)(s  2)(s  3)(s  4)
s(2 s 2  21s  24)
X 3 ( s)  (4.104c)
(s  1)(s  2)(s  3)(s  4)

The output equation is found from Eq. (4.99b). Performing the indicated addition yields

Prepared by: Dr. Mohammad Abdul Mannan Page 13 of 27

 Chapter 3 Modeling in the Time Domain

 X 1 (s) 
Y (s )  1 1 0 X 2 ( s)  X 1 (s)  X 2 ( s) (4.105)
 X 3 ( s) 
or
( s 3  12s 2  16s  5)  6.5 19 11.5
Y (s)     (4.106)
( s  1)(s  2)(s  3)(s  4) s  2 s  3 s  4
where the pole at -1 canceled a zero at -1. Taking the inverse Laplace transform,
y (t )  6.5e 2t  19e 3t  11.5e 4t (4.107)

(b) The denominator of Eq. (4.102), which is det(sI-A), is also the denominator of the system’s
transfer function. Thus, det(sI-A)= 0 furnishes both the poles of the system and the eigenvalues:
-2, -3, and -4.

4.11 Time Domain Solution of State Equations

To solve the differential equation (3.18), we have
x (t )  Ax (t )  Bu(t ) (3.18.3)
We now multiply both sides of Eq. (3.18.1) by e-At:
e  At x (t )  e  At Ax (t )  e  At Bu(t ) (3.18.4)
We know,

d e  At x 
 e  At x (t )  e  At Ax (t ) (3.18.5)
dt
From Eqs. (3.18.2) and (3.18.3) we get:

d e  At x 
 e  At Bu(t ) (3.18.6)
dt
Multiplying both sides by dt and integrating between the limits t0 and t, we obtain
t
e  At x (t )  e  At 0 x (t 0 )   A
t0 e Bu( )d (3.18.7)

Where the dummy variable  has been used in the integral to avoid confusion with t.

Solving Eq. (3.18.5) for x(t), we get

t
x (t )  e  At  t 0  x (t 0 )   At  
t0 e Bu( )d
(4.109)
t
  (t ) x (t 0 )  t 0 (t   )Bu( )d
Where, (t )  e  At  if t0=0, by definition, and which is called the state-transition matrix.
The first term (is called zero input response) on the right-hand-side of Eq. (3.18.6) is seen to be the
complementary function or natural response of the system, while the second term (is called
convolution integral), which is the convolution integral, is the particular solution or forced
response or zero state response.

Comparing Eq. (3.18.8) with Eq. (3.18.2), we recognize that e-At is inverse Laplace transform of (sI-
A)-1. Similarly, the inverse Laplace transform of the second terms on the right-hand-side of Eq.
(3.18.2) is recognized as the convolution integral in Eq. (3.18.8).
Since
Prepared by: Dr. Mohammad Abdul Mannan Page 14 of 27
 Chapter 3 Modeling in the Time Domain

 adjsI  A 
 
L 1 sI  A-1  L 1     (t )
 det sI  A 
(4.111)

each term of (t ) would be the sum of exponentials generated by the system’s poles.

The initial values of state-transition matrix and its derivative are:

 ( 0)  I (4.111.1)
 (0)  Α (4.111.2)

Example 4.12: For the state equation and initial state vector shown in Eqs. (4.112), where u(t) is a
unit step, find the state-transition matrix and then solve for x(t).

0 1 0
x    x   u (t ) (4.112a)
 8  6 1

1
x 0    (4.112b)
0

s 0  0 1  s  1 
sI  A      
0 s   8  6 8 s  6

det sI  A  s 2  6 s  8  ( s  2)(s  4)  0 from which the poles or eigenvalues are -2 and -4.
Since each term of the state-transition matrix is the sum of responses generated by the poles or
eigenvalues, we assume a state-transition matrix of the form:
 K e  2t  K 2 e  4t K 3 e  2t  K 4 e  4t 
 (t )   1  2t  (4.114)
 K 5 e  K 6 e  4t K 7 e  2t  K 8 e  4t 
The derivative of state-transition matrix is
 2t

 (t )    2 K1e  4 K 2 e  4t  2 K 3e  2t  4 K 4 e  4t 
  2t  (4.114.1)
 2 K 5 e  4 K 6 e  4t  2 K 7 e  2t  4 K 8 e  4t 
We know,
 K  K2 K3  K4  1 0
 (0)   1  I   (4.115)
 K 5  K 6 K 7  K 8   0 1 
And

 (0)    2 K1  4 K 2  2 K 3  4 K 4   A   0 1
(4.117)
 2 K  4 K   
 5 6  2K 7  4 K8   8  6 

Comparing both-sides of Eq. (4.115) we have

K1  K 2  1 (4.116a)
K3  K4  0 (4.116b)
K5  K6  0 (4.116c)
K 7  K8  1 (4.116d)

Comparing both-sides of Eq. (4.117) we have

Prepared by: Dr. Mohammad Abdul Mannan Page 15 of 27

 Chapter 3 Modeling in the Time Domain

 2 K1  4 K 2  0 (4.118a)
 2K 3  4 K 4  1 (4.118b)
 2 K 5  4 K 6  8 (4.118c)
 2 K 7  4 K 8  6 (4.118d)

Solving (4.116a) and (4.118a), we get K1  2; K 2  1

1 1
Solving (4.116b) and (4.118b), we get K 3  ; K 4  
2 2
Solving (4.116c) and (4.118c), we get K 5  4; K 6  4
Solving (4.116d) and (4.118d), we get K 7  1; K 8  2

Now, we can write

 1  2t 1  4t 
2e  2t  e  4t e  e 
 (t )   2 2 (4.119)
  2t 
 4e  4e  4t e  2t
 2e  4t 

 1  2(t  ) 1  4(t  ) 
 2e  2(t  )  e  4(t  ) e  e  0
 (t   ) B  2 2  
  2(t  )  4 ( t  )  2 ( t  )  4(t  )  1
  4 e  4e  e  2 e 
 1  2(t  ) 1  4(t  ) 
e  e
 (t   ) B   2 2  (4.120)
  2 ( t  )  4(t  ) 
e  2e 

 1  2t 1  4t 
 2e  2t  e  4t e  e  1
 (t ) x 0   2 2  
  2t  4t  2t  4t  0
 4e  4e e  2e 
 2e  2 t  e  4 t 
 (t ) x 0     2t  (4.121)
 4e  4e  4t 
 1  2t t 2 1 t 4 
t 2 e  e d  e  4 t  e d 
0 2 0
0
 (t   ) Bu( )d  
  e  2t
t 2
e d   2e  4 t
t 4 
e d 
  0 0 
 1 1  2t 1  4t 
t   e  e 
 (t   ) Bu( )d   8 4 8
0

1 1 
e  2t  e  4t 
(4.122)

 2 2 
 1 1  2t 1  4t 
t  2e  2t  e  4t   8  4 e  e 
8

x (t )   (t ) x (t 0 ) 
t0
(t   ) Bu( ) d  
 4e
 2t
 4e  

 4t   1  2t 1  4t 
e  e 
 2 2 
 1 7  2t 7  4t 
  e  e 
x (t )   8 4 8 (4.123)
7  2t 7  4t 
  e  e 
 2 2 
Prepared by: Dr. Mohammad Abdul Mannan Page 16 of 27
 Chapter 3 Modeling in the Time Domain

Example 4.13: Find the state-transition matrix of Example 4.12, using (sI-A)-1.
s 0 0 1  s  1 
sI  A       
0 s   8  6 8 s  6
from which
 s6 1 
1  s  6 1   s  2 s  4  s  2 s  4 
sI  A1  2    
8 s
(4.125)
s  6s  8   8 s   
 s  2 s  4  s  2 s  4  

Expanding each term in the matrix on the right by partial fractions yields
 2 1 1/ 2 1/ 2 
   
sI  A1   s42  s 4 4  s12 s 2 4  (4.126)
   
 s  2  s  4  s  2  s  4 

Finally, taking the inverse Laplace transform of each term, we obtain

 1  2t 1  4t 
 2e  2t  e  4t e  e 
 (t )  2 2 (4.127)
  2t  4t  2t 
 4e  4e e  2e  4t 

3.5 Converting a Transfer Function to State Space

Phase Variable Representation
It is often convenient to consider the output of the system as one of the state variable and remaining
state variable as derivatives of this state variable. The state variables thus obtained from one of the
system variables and its (n-1) derivatives, are known as n-dimensional phase variables.

Let, the transfer function is given by

C s  Y s  b s m  bm 1s m 1  ....  b0
G (s )    m (3.48.1)
Rs  U s  s n  a n 1s n 1  ....  a1s  a0

(a)

(b)

Figure 3.11.1 Decomposing a transfer function.

Prepared by: Dr. Mohammad Abdul Mannan Page 17 of 27
 Chapter 3 Modeling in the Time Domain

We can write Eq. (3.48.1) in the form as

G ( s)  G1 ( s)G2 ( s ) (3.48.2)
Where,
X s  X s  1
G1 ( s )  1  1  (3.48.3)
R s  U s  s  an 1s
n n 1
 ....  a1s  a0
C s  Y s 
G2 ( s)    bm s m  bm 1s m 1  ....  b0 (3.48.4)
X 1 s  X 1 s 
Eq. (3.48.3) can be written as follows:
s n

 an 1s n 1  ....  a1s  a0 X 1 s   U s  (3.48.5)
Taking the inverse Laplace of (3.48.5), we have
 dn d n 1 d 
 n  a n 1 n 1  ....  a1  a0  x1 t   u t 
 dt dt dt 
d n x1 t  d n 1 x1 t  dx1 t 
 an 1  ....  a1  a0 x1 t   ut  (3.48.6)
n n 1 dt
dt dt

Now, let
dx1 t 
x1 t    x2 (3.48.7a)
dt
dx 2 t  d 2 x1 t 
x 2 t     x3 (3.48.7b)
dt dt 2
dx3 t  d 3 x1 t 
x 3 t    x4 (3.48.7c)
dt dt 3
…………………………….
…………………………….

…………………………….
dxn  2 t  d n  2 x1 t 
x n  2 t     xn 1 (3.48.7d)
dt dt n  2
dxn 1 t  d n 1 x1 t 
x n 1 t     xn (3.48.7e)
dt dt n 1
dxn t  d n x1 t 
x n t    (3.48.7e)
dt dt n

Using Eq. (3.48.7), the Eq. (3.48.6) can be written as follows:

d n 1 x1 t  dx1 t 
x n t   an 1  ....  a1  a0 x1 t   u t 
dt n 1 dt
x n t    a n 1 x n t    a n  2 x n 1 t   .......  a1 x 2 t   a0 x1 t   u t 
x n t    a0 x1 t   a1 x 2 t   .......  a n  2 x n 1 t   a n 1 x n t   u t  (3.48.9)

In vector-matrix form, Eq. (3.48.7) and (3.48.9) become

Prepared by: Dr. Mohammad Abdul Mannan Page 18 of 27

 Chapter 3 Modeling in the Time Domain

 x1   0 1 0 .... 0 0   x1   0 
 x   0 0 1 .... 0 0   x2   0 
 2    
 x3   0 0 0 .... .... ....   x3   0 
       u (t ) (3.48.10)
 ....   .... .... .... .... .... ....   ....  ....
 x n 1   0 0 0 .... 0 1   xn 1   0 
      
 x n  a 0 a1 a2 .... an  2 an 1   xn   1 

Eq. (3.48.4) can be written as follows:


Y s   bm s m  bm 1s m 1  ....  b2 s 2  b1s  b0 X 1 s   (3.48.11)
Taking the inverse Laplace of (3.48.10), we have
 dm d m 1 d2 d 
y t   bm  bm 1  ....  b2  b1  b0  x1 t 
m
 dt dt m 1 dt 2 dt 
d m x1 t  d m 1 x1 t  d 2 x1 t  dx1 t 
yt   bm  bm 1  ....  b2  b1  b0 x1 t  (3.48.12)
m m 1 2 dt
dt dt dt
Using Eq. (3.48.7) we can write Eq. (3.48.11) as
y t   bm x m 1 t   bm 1 x m t   ....  b2 x3 t   b1 x 2 t   b0 x1 t  (3.48.13)
In vector-matrix form, Eq. (3.48.12) becomes (where, mn)
 x1 (t ) 
 x (t ) 
 2 
 ..... 
 
y (t )  b0 b1 ... bm 1 bm ... 0 xm (t )  (3.48.14)
 xm 1 (t ) 
 
 .... 
 x (t ) 
 n 

If the transfer function is given by the following type

C s  Y s  b0
G (s )    (3.48.15)
Rs  U s  s  a n 1s
n n 1
 ....  a1s  a0

(a)

(b)

Prepared by: Dr. Mohammad Abdul Mannan Page 19 of 27

 Chapter 3 Modeling in the Time Domain

Figure 3.11.2 Decomposing a transfer function.

Figure 3.11.2 shows the decomposing of transfer function (3.48.15). According to Figure 3.11.2 and
the expressions of Eqs. (3.48.10) and (3.48.14) we have
 x1   0 1 0 .... 0 0   x1   0 
 x   0 0 1 .... 0 0   x2   0 
 2  
 x 3   0 0 0 .... .... ....   x3   0 
       u (t ) (3.48.16)
 ....   .... .... .... .... .... ....  ....   ....
 x n 1   0 0 0 .... 0 1   xn 1   0 
      
 x n  a0 a1 a2 .... an  2 an 1   xn  b0 
 x1 (t ) 
 x (t ) 
 2 
 ..... 
 
y (t )  1 0 ... 0 0 ... 0 xm (t )  (3.48.17)
 xm 1 (t )
 
 .... 
 x (t ) 
 n 

Example 3.4: Find the state-space representation in phase-variable form for the transfer function
shown in Figure 3.10(a).

Solution:
Step 1 Find the associated differential equation. Since
C s  Y s  24
G ( s)    (3.54)
Rs  U s  s 3  9s 2  26s  24

cross-multiplying yields
s 3  9s 2  26s  24C(s)  24Rs (3.55)

The corresponding differential equation is found by taking the inverse Laplace transform, assuming
zero initial conditions:
d 3 c(t ) d 2 c(t )
dc(t )
9  24c(t )  24r t 
 26 (3.56)
3 2 dt
dt dt
Step 2 Select the state variables.
Choosing the state variables as successive derivatives, we get
x1 (t )  c(t ) (3.57a)
dc(t )
x2 (t )  (3.57b)
dt
d 2 c(t )
x3 (t )  (3.57c)
dt 2

Prepared by: Dr. Mohammad Abdul Mannan Page 20 of 27

 Chapter 3 Modeling in the Time Domain

dx3 (t ) d 3 c(t )
 (3.57d)
dt dt 3
Differentiating both sides and making use of Eq. (3.57) to find dx1/dt and dx2/dt, and Eq. (3.56) to
find d3c/dt3=dx3/dt, we obtain the state equations. Since the output is c=x1, the combined state and
output equations are
dx1 (t )
 x1 (t )  x2 (t ) (3.58a)
dt
dx2 (t )
 x 2 (t )  x3 (t ) (3.58b)
dt
dx3 (t )
 x3 (t )  24x1 (t )  26x2 (t )  9 x3 (t )  24r (t ) (3.58c)
dt
y  c  x1 (t ) (3.58d)

Figure 3.10 (a) Transfer function; (b) equivalent block diagram showing phase variables where
y(t)=c(t).
In vector-matrix form,
 x1 (t )   0 1 0   x1 (t )   0 
 x (t )   0 0 1   x2 (t )   0  r (t )
 2       (3.59a)
 x3 (t )   24  26  9  x3 (t )  24
 x1 (t ) 
y (t )  1 0 0 x2 (t ) (3.59b)
 
 x3 (t ) 
At this point, we can create an equivalent block diagram of the system of Figure 3.10(a) to help
visualize the state variables. We draw three integral blocks as shown in Figure 3.10(b) and label
each output as one of the state variables, xi(t), as shown. Since the input to each integrator is
xi(t),use Eqs.(3.58a),(3.58b),and(3.58c)to determine the combination of input signals to each
integrator. Form and label each input.
Finally, use Eq. (3.58d) to form and label the output, y(t)=c(t). The final result of Figure 3.10(b) is a
Prepared by: Dr. Mohammad Abdul Mannan Page 21 of 27
 Chapter 3 Modeling in the Time Domain

system equivalent to Figure 3.10(a) that explicitly shows the state variables and gives a vivid
picture of the state-space representation.

Example 3.5: Find the state-space representation of the transfer function shown in Figure 3.12(a).

Figure 3.12 (a) Transfer function; (b) decomposed transfer function; (c) equivalent block diagram.
Where, y(t)=c(t).

Solution:
Step 1 Separate the system into two cascaded blocks, as shown in Figure 3.12(b). The first block
contains the denominator and the second block contains the numerator.

Step 2 Find the state equations for the block containing the denominator. We notice that the first
block’s numerator is 1/24 that of Example 3.4. Thus, the state equations are the same except that
this system’s input matrix is 1/24 that of Example 3.4. Hence, the state equation is

Prepared by: Dr. Mohammad Abdul Mannan Page 22 of 27

 Chapter 3 Modeling in the Time Domain

 x1 (t )   0 1 0   x1 (t )  0
 x (t )   0 0 1   x2 (t )  0 r (t )
 2       (3.63)
 x3 (t )   24  26  9  x3 (t )  1

Step 3 Introduce the effect of the block with the numerator. The second block of Figure 3.12(b),
where b2=1; b1=7, and b0=2, states that
 
C s   s 2  7s  2 X 1 s  (3.64)

Taking the inverse Laplace transform with zero initial conditions, we get
ct   x1 t   7 x1 t   2 x1 t  (3.65)
But
x1 (t )  x2 (t )
x1 (t )  x 2 (t )  x3 (t )
Hence
y t   ct   x3 t   7 x2 t   2 x1 t  (3.66)
Thus, the last box of Figure 3.11(b) ‘‘collects’’ the states and generates the output
equation. From Eq. (3.66),
 x1 (t ) 
y t   2 7 1 x2 (t ) (3.67)
 
 x3 (t ) 

Although the second block of Figure 3.12(b) shows differentiation, this block was implemented
without differentiation because of the partitioning that was applied to the transfer function. The last
block simply collected derivatives that were already formed by the first block.
Once again we can produce an equivalent block diagram that vividly represents our state-space
model. The first block of Figure 3.12(b) is the same as Figure 3.10(a) except for the different
constant in the numerator. Thus, in Figure 3.12(c) we reproduce Figure 3.10(b) except for the
change in the numerator constant which appears as a change in the input multiplying factor. The
second block of Figure 3.12(b) is represented using Eq. (3.66), which forms the output from a linear
combination of the state variables, as shown in Figure 3.12(c).

3.6 Converting from State Space to a Transfer Function

[Transfer Function from State Equations]
The transfer function can be obtained uniquely from the state equations. Taking the Laplace
transform of both sides of Eqs. (3.18) and (3.19) with x(0) set equal to zero we get
sX ( s )  AX ( s )  BU ( s ) (3.18.9)
Y ( s )  CX ( s )  DU ( s ) (3.19.1)
By solving (3.18.1) we get
X ( s)  sI - A1 BU ( s) (3.18.2)
Substitute Eq. (3.18.2) in the Eq. (3.19.1), we have

Y ( s)  C sI - A1 B  D U ( s) (3.19.2)
The transfer function is now obtained as
Y ( s)
G ( s)   C sI - A 1 B  D (3.19.3)
U ( s)

Prepared by: Dr. Mohammad Abdul Mannan Page 23 of 27

 Chapter 3 Modeling in the Time Domain

Eigenvalues and Transfer Function Poles

The expression of (3.19.3) can be written as follows:
C adjsI - A B CadjsI - A B  D det sI - A  P ( s )
G ( s)  D  (3.19.4)
det sI - A  det sI - A  Q (s )
The denominator polynomial Q(s) is:
Q ( s )  det sI - A  (3.19.5)
The polynomial Q(s) is of degree n if A is an nn matrix.
Since the poles of the transfer function are the roots of Q(s) and the roots of det(sI-A) are the
eigenvalues of A, it is evident that the poles of the transfer function are identical to the eigenvalues
of A.
The numerator of the transfer function is given by
P ( s )  C adjsI - A B  DQ ( s ) (3.19.6)
The degree of the first term in equation (3.19.6) is less than or equal to n-1, while that of the second
term will be n, since D is scalar. Hence, the transfer function will always be proper if D is nonzero
and will be strictly proper if D=0.

Example 3.6: Given the system defined by Eq. (3.74), find the transfer function, T(s) = Y(s)/U(s),
where U(s) is the input and Y(s) is the output.
0 1 0 10

x  0 0 1 x   0 u (t )

    (3.74a)
 1  2  3  0 
yt   1 0 0x (3.74b)

Solution:
1 0 0   0 1 0  s 1 0 
    
sI - A  s 0 1 0   0 0 1   0 s  1  (3.75)
0 0 1    1  2  3 1 2 s  3

The denominator polynomial is obtained as

Q ( s )  det sI - A   s 3  3s 2  2 s  1

The adjoint of (sI-A) is obtained as:

 s 2  3s  2 s 3 0
 
adjsI - A    1 s ( s  3) s
 s  ( 2 s  1) s 2 
 
The numerator polynomial is obtained as
P ( s )  C adjsI - A B
 s 2  3s  2 s3 0  10 
 
P ( s )  1 0 0  1 s ( s  3) 
s   0   10 s 2  3s  2 
 s  ( 2 s  1) s 2   0 
 
The transfer function may now be written as

Prepared by: Dr. Mohammad Abdul Mannan Page 24 of 27

 Chapter 3 Modeling in the Time Domain

T ( s) 
Y ( s)
 G ( s) 
P (s)


10 s 2  3s  2  (3.77)
U ( s) Q ( s ) s 3  3s 2  2 s  1

Example: Determine the transfer function for the following case:

 2 0 1 1 
A   1  2 0  ; B  0  ; C  2 1 1 ; D  0
 
 1 1  1 1 
Solution:
1 0 0    2 0 1 s  2 0 1 
  
sI - A  s 0 1 0   1  2 0    1 s2 0 

0 0 1   1 1  1  1 1 s  1

The denominator polynomial is obtained as

Q ( s )  det sI - A   s 3  5 s 2  7 s  1

The adjoint of (sI-A) is obtained as:

 s 2  3s  2 1 s2 
 2 
adjsI - A    s  1 s  3s  1 1 
 s3 s2 s  4 s  4 
2
 

The numerator polynomial is obtained as

P ( s )  C adjsI - A B
 s 2  3s  2 1 s2  1 
  
P ( s )  2 1  1 s  1 s 2  3s  1 1 2
 0   s  4 s  3
 s3 s2 s 2  4 s  4  1 
 

The transfer function may now be written as

P(s) s 2  4s  3
G ( s)  
Q ( s) s 3  5s 2  7 s  1

Linearization
Example 3.7: First represent the simple pendulum shown in Figure 3.14(a) (which could be a
simple model for the leg of the robot shown in Figure 3.13) in state space: Mg is the weight, T is an
applied torque in the  direction, and L is the length of the pendulum. Assume the mass is evenly
distributed, with the center of mass at L/2. Then linearize the state equations about the pendulum’s
equilibrium point—the vertical position with zero angular velocity.

Solution:
First draw a free-body diagram as shown in Figure 3.14(c). Summing the torques, we get
d 2 MgL
J  sin   T (3.79)
dt 2 2

Prepared by: Dr. Mohammad Abdul Mannan Page 25 of 27

 Chapter 3 Modeling in the Time Domain

Where, J is the momment of innertia of the pendulum around the joint of rotation. Select the state
variables x1 and x2 as phase variables. Letting
x1  
And
d dx1
x2    x1
dt dt
We write the state equation as
x1  x 2 (3.80a)
d 2 MgL T
x 2   sin x1  (3.80b)
dt 2 2J J

Thus, we have represented a nonlinear system in state space. It is interesting to note that the
nonlinear Eq. (3.80) represent a valid and complete model of the pendulum in state space even
under nonzero initial conditions and even if parameters are time varying. However, if we want to
apply classical techniques and convert these state equations to a transfer function, we must linearize
them.

Figure 3.14 (a) Simple pendulum; (b) force components of Mg; (c) free-body diagram

Let us proceed now to linearize the equation about the equilibrium point, x1=0; x2=0, that is, =0
and d/dt=0. Let x1 and x2 be perturbed about the equilibrium point, or
x1  0  x1 (3.81a)
x 2  0  x 2 (3.81a)
Using Eq. (2.182), we have
d (sin x1 )
sin x1  sin 0  x1  x1 (3.82)
dx1 x  0
1
From which
sinx1  x1 (3.83)

Substituting Eqs. (3.81) and (3.83) into Eq. (3.80) yields the following state
equations:
x1  x 2 (3.84a)

Prepared by: Dr. Mohammad Abdul Mannan Page 26 of 27

 Chapter 3 Modeling in the Time Domain

MgL T
x 2   x1  (3.84b)
2J J
which are linear and a good approximation to Eq. (3.80) for small excursions away from the
equilibrium point. What is the output equation?

References
[1] Norman S. Nise, “Control Systems Engineering,” Sixth Edition, John Wiley and Sons, Inc,
2011.

Prepared by: Dr. Mohammad Abdul Mannan Page 27 of 27