1131- Signals and Systems 講義 國立陽明交通大學 電機系 陳永平 編

12. System in State-Space Description

It has been introduced that an LTI system can be expressed by an ordinary differential
equation to describe the relation between input and output. Usually, we say that the
system is in input-output description. However, the input-output description is
unavailable to express the whole features of a system. If a system is known all in
detail, not just the relation between input and output, we often write it in state-space
description.

For simplicity, we only present a 2nd order system in state-space description here and
all the properties discussed below are also available for higher order systems.

a11

.
x1(t) x1(t)
b1 +
 + c1

u(t) a21 y(t)

d +
a12

x.2(t) x2(t)
b2 +
 + c2

a22

In general, an LTI system in state space description is expressed by two 1st order
differential equations, called the state equations, given as below:

(1) x1 ( t ) = a11 x1 ( t ) + a12 x2 ( t ) + b1u ( t ) , x1 ( 0 ) = x10

(2) x 2 (t ) = a 21 x1 (t ) + a 22 x 2 (t ) + b2 u (t ) , x2 ( 0 ) = x20

and its output eqation is

(3) y (t ) = c1 x1 (t ) + c2 x2 (t ) + du (t )

The block diagram is shown in the figure.

1/5
1131- Signals and Systems 講義 國立陽明交通大學 電機系 陳永平 編

For convenience, rearrange the 2nd order LTI system into a vector form, where the
state equation is

 x1 (t )  a11 a12   x1 (t )  b1   x1 (0 )  x10 

 x (t ) = a  + u (t ) ,  x (0 ) =  x 
a 22   x 2 (t ) b2 
(4)
 2  
   
21 2
  20 

x  t  A x  t  b  
x 0 x0

and the output equation is

 x (t )
(5) y (t ) = c1 c2   1  + du (t )
   x2 (t )
c

Both are rewritten as below:

(6) x  t  = A x  t  + b u (t ) , x  0  = x 0

(7) y(t ) = c  x  t  + du (t )

Below shows the block diagram correspondingly.

u(t) .
x(t) x(t) y(t)
b +
 c

It is known that the system state x(t) can be obtained as

x  t  = e A t x 0 +  e A(t − ) b u ( )d

t
(8)
0

Then the output is expressed as

y  t  = cx(t ) + du (t ) = ce At x 0 +  ce A(t − )bu ( )d + du (t )

t
(9)
0

However, it is not an easy work to determine x ( t ) by directly calculating (8) and (9).

Instead, we often apply numerical tools to find the numerical results, such as the
software MATLAB.

2/5
1131- Signals and Systems 講義 國立陽明交通大學 電機系 陳永平 編

Let’s take the following LTI system as an example

(10) x1 (t ) = x2 (t )

(11) x 2 (t ) = −3x1 (t ) − 4 x 2 (t ) + u (t )

(12) y (t ) = 2 x1 (t ) − x 2 (t )

with initial conditions x1(0)=1 and x2(0)=0. If u(t)=sin(cos(5t)), then the system
response y(t) can be obained by the instruction ode45 in MATLAB, or by Simulink.

==========================================
Create m-file: second.m
function dx=second(t,x)
dx=zeros(2,1);
dx(1)=x(2);
dx(2)=-3*x(1)-4*x(2)+sin(cos(5*t));
>> % key in the following instructions
>> [t,x]=ode45(@second,[0:0.01:6],[1 0])
>> y=2*x(:,1)-x(:,2);
>> plot(t,y); xlabel(‘t’); ylabel(‘y(t)’)

2.5

1.5
y(t)

0.5

-0.5
0 1 2 3 4 5 6
t

==========================================

Simulink:

3/5
1131- Signals and Systems 講義 國立陽明交通大學 電機系 陳永平 編

If an LTI system is given in state space description, then we can change it into
input-output description.

Now, let’s present the procedure to derive the input-output description from the
state space description.

First, find the characteristic equation of system matrix A, which is defined as

 − a11 − a12
(13) p( ) =  I − A = =0
− a 21  − a 22

i.e., p( ) = 2 − (a11 + a 22 ) + a11 a 22 − a12 a 21 = 0 , where p (  ) is the characteristic

polynomial. Let a1 = −(a11 + a 22 ) and a 0 = a11 a 22 − a12 a 21 , then

(14) p( ) = 2 + a1 + a0

According to the Cayley-Hamilton theoreom, we have

(15) p( A) = A 2 + a1 A+ a0 I = 0

Then, calculate the following equation

(17) y ( t ) + a1 y ( t ) + a0 y ( t )

= c ( x ( t ) + a1 x ( t ) + a0 x ( t ) ) + du ( t ) + a1du ( t ) + a0 du ( t )
 
= c  A2 x ( t ) + Abu ( t ) + bu ( t )+ a1 ( Ax ( t ) + bu ( t ) ) + a0 x ( t ) 
 
+ du ( t ) + a1du ( t ) + a0 du ( t )

= c ( A2 + a1 A+ a0 I ) x ( t ) + du ( t )+ ( cb + a1d ) u ( t )
=0

+ ( cAb + a1cb + a0 d ) u ( t )

4/5
1131- Signals and Systems 講義 國立陽明交通大學 電機系 陳永平 編

Which results in the following input-output description

(18) y(t ) + a1 y (t ) + a0 y (t ) = b2 u(t ) + b1u (t ) + b0 u (t )

where b2 = d , b1 = cb and b0 = cAb + a1cb + a0 d .

Problems

1 Consider the following LTI system:

x1 (t ) = x 2 (t ) , x1(0)=1
x 2 (t ) = −5 x1 (t ) − 2 x 2 (t ) , x2(0)=1
y (t ) = x1 (t ) + 3x 2 (t )
where the input is u(t)=sin(cos(3t)). Solve the output y(t) by MATLAB.

5/5