ECE Department, Florida Int¶l University
A state space representation or model is a
Most practical engineering problems are not mathematical abstraction of the behavior of
physical or dynamic system as a set of input,
easy to solve analytically. In such cases, we may
output and state variables related by first-
require numerical methods to solve such order differential equationsp [3]. The state space
problems. These solutions are usually iterative model provides a convenient and compact way
and may need a lot of intermediate steps to get to model and analyze systems with multiple
to a solution. A very useful characterization of inputs and outputs. With inputs and outputs,
linear circuits is the use of transfer functions, we would otherwise have to write down
which is the frequency domain response Laplace transforms to encode all the
information about a system. Unlike the
equivalent of the circuit¶s time domain input-
frequency domain approach, the use of the state
output relations. The limitation of this method is space model is not limited to systems with linear
that it does not use information of the internal components and zero initial conditions. "State
structure of the system thereby, giving limited space" refers to the space whose axes are the
control of the closed loop behavior in a system state variables. The state of the system can be
with feedback. represented as a vector within that space [2].
"%
"%
The standard form for the mathematical The p of a system is any variable of the
description of a system is written as a set of p system that is of great concern. A set of state
coupled first-order ordinary differential variables will not necessarily include all the
equations for continuous time systems and a set variables of direct engineering interest. An
of n coupled first order difference equations for arbitrary output variable
in a system of order
discrete time systems. This standard form of pwith ppinputs may be generally written:
expression is known as the p . In
the continuous time system, the time derivative X
X X X
of each state variable is expressed in terms of the
state variables ()
p p p p
p () and the For continuous time systems.
system inputs 1()
pppp
p(). The general
case for the continuous time system with n state X
X X X
equation is
For discrete time systems.
] V1 (
p
p)
] V2 (
p
p)
: : &'
( )
] Vp(
p
p) (
Where ]
,and and
are vectors. The system model for a linear time invariant
continuous system is made up of a set of n state
equations described in terms of matrices A and
In the discrete time system, the general case is
B, and a set of output equations described in
given as
terms of matrix C and D. Modeling a linear time
] X
X X X
invariant system involves deriving the elements
] X
X X X
that make up these matrices A, B, C and D.
The system model can then be written in this
: :
] X
X X X
form
(t) c(t) +
(t)
Again, and
are vectors
;X(t) +
(t)
And making necessary substitution gives
pppp
- X X
X X Equation (8) is a second order LTI equation
X X
X
----p
p
è è )
p
X] ) ]]
- --
]
]
p
+ Block Diagram of State-Space Modeled
Circuit ] !
è ]
è
p
From the block diagram representation of the ! ] - !
system, a state-space equation of the system in
]
] ) p
]
]
controllable form can be written as
]
] ] p
The poles of the system are atp] p
] ] ]
p
&
; &
& ¥¥¥ (14) (] ] )
p
ø
ø p
pAgain, the system being stable would dependponpp
ø
ø p
p
k ø ø
ø
ø p
+&
,
.
k k ø ø
For this system the controllability matrixppp
Finally, closed loop system is given
Ë ) p
)
k
1
! !
Ë p
ø ø ø
ø
!
The rank of thep$%&pis 2. Therefore, the system
1 p
is controllable.
Since the system is controllable, a controller can
be designed to improve its stability. -#'c"'*"
!"#
Suppose the system is to be redesigned with
poles atp ø -andp ø p such that it¶s closed loop For this simulation, we investigate the effect of
characteristic polynomial is given asp varying a1 and a2 in the state-space equation
using SIMULINK®. Note that a1and a2 are
'()ip*
p!p*(ø
p*(ø
p
dependent on the values of the circuit elements
By finding the inverse ofp $%&and the chosen for the double differentiator.
transformation matrix +p
p
0 !
% ! ppppppppppp+! ! p
! 0) !
!pi p!pi+ p
pLetpk
p