1

State-space behaviours 7
discrete systems
J A Rossiter

Slides by Anthony Rossiter

 2

Introduction
The previous videos focussed on the solutions for
continuous time systems.
x  Ax  Bu  x(t )  (t ) x(0)  H (t )u

(t )  e ; H (t )  A (t )  I B
At 1

t
e  We V
At

This video considers how these observations change

for discrete systems. The observations are given
without detailed derivation as mostly analogous.
Slides by Anthony Rossiter
 3

Discrete state space model

The basic equations we seek take the following
form.
xk 1  Axk  Buk  xk  n   (n) xk  H (n)uk
This video considers the derivation of the transition
matrix Φ(n) and step response matrix H(n).

It is assumed that for a step response:

uk  uk 1  uk  2  
Slides by Anthony Rossiter
 4

State transition matrix

Equivalent to continuous time we can use
transforms and hence:
xk 1  Axk  ( zI  A) x( z )  x0
 ( z )  ( zI  A)  1
  I  Az 1 2 2
 A z 

x( z )   ( z ) x0  xk  A x0 k

This is also obvious from a

(k )  A k
recursion of the formula
xk 1  Axk
Slides by Anthony Rossiter
 5

Remark
In discrete time the state transition matrix has an
easier form that in continuous time.

( k )  A k

Nevertheless, computation of this for several

values of k is not an insignificant task.

Slides by Anthony Rossiter

 6

Eigenvalue/vector decomposition
The decomposition (distinct eigenvalues)is:

1  0  v 
T
1
   
A  WV  w1  wn        
 T
 0  n  vn 

v 
T

 
1
Awi  i wi
w1  wn     I  VW
vnT 
Slides by Anthony Rossiter
 
 7

Discrete system with eigenvalues

For a discrete system, the eigenvalue/vector
decomposition can be deployed, as in continuous
time, to investigate the impact of different modes
on the overall behaviour.

xk  A x0  W Vx0
k k

Focus here is on the case of distinct eigenvalues only.

Derivation done quickly as equivalent to that in video 5.

Slides by Anthony Rossiter
 8

Projection onto eigenvectors

The initial condition can be projected onto the
eigenvectors.

x0  1w1     n wn
The relevant coefficients can be determined as
follows: T
 i  vi x0

Slides by Anthony Rossiter

 9

Superposition
Hence analyse each component in turn.
x0  1w1   2 w2     n wn  i  v x0
T
i

v w T
i j  0, i  j  v w  1
T
i i

xk  W Vx0  w  1  w   2  
k k
1 1
k
2 2

• The solution has n distinct modes linked directly to the

eigenvalues.
• The contribution, or decay, along each eigenvector direction is
linked directly to the corresponding eigenvalue.
Slides by Anthony Rossiter
 10

Step response
The derivation of the step response matrix can be
done quickly using the existing results.
xk 1  Axk  Buk 
  xk  n   (n) xk  H (n)uk
uk  uk 1  uk  2  
1
( zI  A) x( z )  Bu ( z )  x( z )  ( zI  A) Bu ( z )
1 1 2 3
( zI  A) B  Bz  ABz  A Bz   2

n 1
xk  n  Buk  n 1  ABuk  n  2    A Buk
n 1
uk  uk 1    H (n)   A B i

i 0
Slides by Anthony Rossiter
 11

NUMERICAL EXAMPLES

Slides by Anthony Rossiter

 12
x(k) tends towards the
EXAMPLE 1 eigenmode with
A
 0.89 0.044  2 w2 slowest convergence.
 2

 0 . 022 0 . 81 
 1  0 .5  1.5

W  
0.25 1 
1

 0 .9 0 
 
 0 0 . 8 
0.5

1.5
x ( 0)   
0

2
-0.5
-0.5 0 0.5 1 1.5 1w12

Some concept as observed for continuous time so no

more
Slides by Anthonyexamples
Rossiter are presented (see video 5).
 13

Remark
If just one of the eigenvalues corresponds to a
divergent mode then clearly the trajectory will
approach this asymptotically.

xk  w11 1    wn n  n
k k

i  1  lim k  xk  wi i  i
k

Slides by Anthony Rossiter

 14

Summary
The behaviours of discrete models have close
analogies to those of the continuous state space
models.
Free response:
xk  A x0  W Vx0  w  v x0  w  v x0  
k k k T
1 1 1
k T
2 2 2

Step response.
xk 1  Axk  Buk 
  xk  n   (n) xk  H (n)uk
uk  uk 1  uk  2  
n 1
 ( n)  A ;
Slides by Anthony Rossiter
n
H (n)  B  AB    A B
 Anthony Rossiter
Department of Automatic Control and
Systems Engineering
University of Sheffield
www.shef.ac.uk/acse

© 2016 University of Sheffield

This work is licensed under the Creative Commons Attribution 2.0 UK: England & Wales Licence. To view a copy of this licence, visit
http://creativecommons.org/licenses/by/2.0/uk/ or send a letter to: Creative Commons, 171 Second Street, Suite 300, San Francisco, California 94105, USA.

It should be noted that some of the materials contained within this resource are subject to third party rights and any copyright notices must remain with these materials
in the event of reuse or repurposing.

If there are third party images within the resource please do not remove or alter any of the copyright notices or website details shown below the image.

(Please list details of the third party rights contained within this work.

If you include your institutions logo on the cover please include reference to the fact that it is a trade mark and all copyright in that image is reserved.)