Example 5-1
Consider the following system:
YQ) ztt
U(z) 27 +132 +04
‘The state-space representations in the controllable canonical form, observable canon-
ical form, and diagonal canonical form become as follows:
CONTROLLABLE CANONICAL FORM:
Tnk+y]_f 0 1 Txt), fo
[tk + 2] - [2 -1 23] * [S}.a
= alk)
vk) = [1 fe]
OBSERVABLE CANONICAL FORM:
x(k +1] fo -04] xa], [1]
[nt + | - [° “1 s[28)] * [ca
ye =10 1120]
DIAGONAL CANONICAL FORM:
‘The given pulse transfer function ¥(zV/U(z) ean be expanded as follows:
Ye) _ 53-23
Ua) 2405 2+08
Hence,
(k+H]_[-05 0 Vfxue], fa
[tt + a]-[ 0-0 8] * [tee
=[i —2] eo
neo =[3 [x2]
Example 5-2
Obtain the state transition matrix of the following discrete-time system:
x(k + 1) = Gx(k) + Hu(k)
y(k) = Cx(k)
where
o-[8 J} aL] eno
Then obtain the state x(k) and the output y(k) when the input u(k) = 1 for k = 0,1,
2, Assume that the initial state is given by
=) OP) 1
r0- 9)From Equations (5~35) and (5-43) the state trensition matrix (Kk) is
Wk) = GE = 2~[(21 - Gz]
Therefore, we first obtain (21 ~ G)':
aafz ay"
(t- [vie a
zt 1
_|@ FONG r08) ]FODE TOS)
~0.16 z
EFIAE FO CHOHEFOH,
4 4
208 2402
3H
240
~)_=
yH02 2408 240:
‘The state transition matrix ‘W(E) is now obtained as follows:
W(k) = G* (zl - G)*z]
4_2_ dig
3z+02 32+08
08 oz Bo
37402 32408
$(-0.2)* - 4(-0.8)* 43(-0.2)' — $(-0
92-02) + BR(-08)" 4-02)" + $(-08)8
Equation (5-45) gives the state transition matrix.
Next, compute x(k). The z transform of x(k) is given by
Z [x(k)] = X(z) = (21 = G)"'2x(0) + (1 - G) HUG)
= (21 - G)"fzx(0) + HU(2)]
Since
U(z)
we obtain
zx(0) + HU(2) = [-
Hence
X(z) = (21 ~ G)'[ex(0) + HU(e)]
(+22 ]
GF0DE +0 -H
(2 + 1.842)
@+0NE+0NE-D“Bz Bz
_|z+0a*z+08
Mz, =e
wz
z+02 2+08
Thus, the state vector x(k) is given by
= 2-09) + B09 + #
HEED = | sao ny ya(-08) + i
Finally, the output y(k) is obtained as follows
~B(-0 2) + B(-0.8) + %
J(k) = Cx(k) = (1 0] 24(-0.2)* — 38(-0.8)' + 4
= 8b. + B08) +
Example 5-3
Determine the inverse of the matrix (21 ~ G), where
01 01 0
=|03 -01 -02
0 0 -03.
Also, obtain G*
From Equation (5-46), we have
adj (zt - G)
\aI-G
Although the determinant [21 ~ G{ can be expanded easily, here for demonstration
purposes let us use Equation (5-50) to compute a, az, and a First, notice that
(1-6
o 0 ~03
Then, from Equation (5-49), we obtain
oL o1 0 030 0
=G+al=/03 -01 -02/+/0 03 0
01 O1 0
a= -wG=-t/03 -01 -02/=03
0 0 -03} [o o 03
o4 O1 0
=|03 02 -02
a)
Hence 4
[fox 01 0 fos 01 0
a, = -{tGH,=-tr}/03 -01 -02/]03 02 -02
[lo 0 -o3f0 0 0
007 003 -O
=-ttr]009 001 0
0
o 0By substituting the matrix H, and the value of a; just obtained into Equation (549),
we get
003 003 -002
GH, +a:1=|009 -0.03 0.02
0 0-004
and
001220 0
a= ~-}tGH.=—-}tr/0 0012 0
0 0 oi
Notice that
Hy = GH, ~ 0.0121 = 0
The adjoint of (21 ~ G) can now be given by Equation (5~48), or
adj(zI - G) = Iz? + Hyz + Hy
100 04 01 0 003 0.03 -0.02
=]0 1 0f%+/03 02 -02/z+|0.09 -0.03 0.02
oo1 en) 0 0 ~0.04
2+ 0.42 + 0.03 0.12 + 0.03 0.02
= | 0.32 + 0.09 2 +02z-003 —0.22 + 0.02
° 0 004
Also,
fel G]= 274 a2? + az + a = 2 + 0.32" ~ 0.042 - 0.012
= (z+ 0.22 -0.2)(z + 0.3)
Hence,
~ gy = adiG@I- 6)
(Al 6)" = ay
z+01 ot
@¥0Iz=02) E002) (+0E-0DE FO
0.3 z-01 =0.2(z ~ 0.1)
(2+0.2)(2-02) (2 +0.2)(2-02) (2 022-02) +03)
0 0 1
z+03
This last equation gives the inverse of (21 - G)
‘Next, we shall obtain G* From Equation (5-43), we have
Gi = Ze ~ GY"z]
0.252 , 0.752 __0.252
z+02 z+02
=o] 0.752, 0.752
- 2402 2402
0 00.25(—0 2) + 0.75(0 2)" -0.25(-0.2)' + 0.25(0.2)*
0.75(-0.2)' + 0750.2)" — 0.75(-0.2)* + 0.25(0.2)*
0 0
0.5(-0.2)* ~ 0100.2)! - 0.4(-0.3"
=1,5(-0.2)* 0.10.2)" + 1 6(-0.3)*
(-0.3)*
Example 5-4
Consider the continuous-time system given by
Yo) od
SO) G6) “Sa
Obtain the continuous-time state-space representation of the system. Then discretize
the state equation and output equation and obtain the discrete-time state-space repre~
sentation of the system. Also, obtain the pulse transfer function for the system by using
Equation (5-60).
‘The continuous-time state-space representation of the system is simply
tena tu
yur
Now we discretize the state equation and the output equation. Referring to Equations
(5-73) and (5-74), we have
Gn =e"
H(T) = [lena =
Hence, the discretized version of the system equations is
1
x(k +1)
¥(k) = x(k)
Referring to Equation (5~60), the pulse transfer function for this system is
F(z) = Cl - GH
=@ee
Gne a aa z=)
This result, of course, agrees with the z transform of G(s) where it is preceded
by a sampler and zero-order hold [that is, where the signal u(t) is sampled and fed to
a zero-order hold before being applied to G(s)]:Example 5-5
Obtain the discrete-time state and output equations and the pulse transfer function
(when the sampling period T = 1) of the following continuous-time system:
~%6) 4
CO) = TG) = Gea
which may be represented in state space by the equations
[Ch
The desired discrete-time state equation will have the form
x((k + 1)T) = G(T)x(KT) + HT) (AT)
where matrices G(T) and H(7) are obtained from Equations (5-73) and (5-74) as
follows:
ain=er=[} 40 6]
e
wn (Levan {f]s 22a] -
‘Thus, r
fuck + 07)] “yx 1
[aa enn} Lo .(4T)] * |?
‘The ouput equation becomes :
(kT) * [1 ost?
‘When the sampling period is 1 sec, or J = 1, the discrete-time state equation and the
output equation become, respectively,
i(k +1)] [1 0.4323][x¢k)] , [02838
[: k + 3] * [; 0 B28) * [3 zo
woo
and‘The pulse-transfer-function representation of this system can be obtained from Equa-
tion (5-60), as follows:
F(z) = C(zl ~ GH + D
ott of? 3 0.4323, | [3333] +0
2-0 1353 0.4323
a 043
2-1 @~ HG - C1355} [0 2838
“pg 0 1 0.4323 |
701353
8382 + 0.1485
= Iz = 0.1353)
= 0.283827! + 0.14852
= eq = 0135327)
Note that the same pulse transfer function can be obtained by taking the z
transform of G(s) when it is preceded by a sampler and zero-order hold Assuming
T = 1, we obtain
c@ |
_ 0.283827! + 0.1485:
“= 2 )(1 = 0.13:Example 5-6
Consider the system discussed in Example 5-5. Obtain the discrete-time state equation
and the output equation at = kT + AT’ Also, obtain the specific expressions for the
state equation and output equation when T = 1 sec and AT = 0.5 sec
In Example 5-5, the matrices G(T) and H(T) were obtained as follows:
To obtain the state equation and the output equation at ¢ = kT + AT, where
0< AP Fe) = 227 coswh +2
Note that the pulse transfer function obiained in this way is the same as that obtained
by taking the z transform of the system that is preceded by a zero-order hold. That is,
¥@) fine? ow |g yel}
Ue) [ 3 7, (ne nel!
= -riita
(L= coswT)(1 + 2")2
“Y= 22 cos wT + 2
‘Thus, we get the same expression for the pulse transfer function. The reason for this
is that discretization in the state space yields the zero-order hold equivalent of the
continuous-time system
Problem A-5-16
Even though the double-integrator system is dynamically simple, it represents an
important class of systems. An example of double-integrator systems is a satellite
attitude control system, which can be described by
Jésurty
where J is the moment of inertia, is the attitude angle, u is the control torque, and
vis the disturbance torque.
Consider the double-integrator system in the absence of disturbance input.
Define J@ = y. Then the system equation becomes
peu
Obtain a continuous-time state-space representation of the system. Then obtain a
discrete-time equivalent. Also obtain the pulse transfer function for the discrete-time
systemSolution Define
uy
may
Then the continuous-time state equation and output equation become
H) 2) 0 Tix 0
[2)-[ ole}
- a
yet of
The discrete-time equivalent of this system can be given by
x(k + 1)T) = Gx(kT) + Hu(kT)
¥(KT) = Cx(kT)
Matrices G and H are obtained from Equations (5-73) and (5-74). Noting that
he
“Rd
r-(Coabo((C abl
Hence, the discrete-time state equation and output sean become
ak +00] [1 TI nen 7
[Gt pp =lo iffnan|*|- "4?
yQkT) = [1 af |
The pulse transfer function of the discrete-time system is obtained from Equation
(5-60) as follows:
we have
and
¥@) _
UG)
= F(z) = C@l-G)"H+DProblem A-5-17
Show that the following quadratic form is positive definite:
V(x) = 10st + dx} tx} + eres — Deets ~ des
Solution The quadratic form V(x) can be written as follows:
101 -2]fx
VQ) =x7Px = [xy ot el) Lo 4 -1]lae
-2 -10 1x5
Applying Sylvester's criterion, we obtain
| wo 1 2
10>0, ” i>o. 1 4 -1)>0
2 1 1
Since all the successive principal minors of the matrix P are positive, V(x) is positive
definite.
Problem A-S-22
Consider the system defined by the equations
alk +1) = Qni(k) + 0.Sxx(k) - 5
alk +1) = 08r(k) +2
Determine the stability of the equilibrium state,
Solution Define the equilibrium state as
2K) = they (Kk) = te
‘Then such an equilibrium state can be determined from the following two simultaneous
‘equations:
Bie = Dy + O.5tae = 5
Ja = O8ty +2
or
The equilibrium state is thus (0, 10)
Now let us consider a new coordinate system with the origin at the equilibrium.
state Define
4k) = i(k)
24K) = x(k) - 10
‘Then the system equations become
Ak +1) = 28K) + 0 S[EA{k) + 10] - 5
x(k + 1) +10 = 0 8L8(K) + 10] +2
(ees ]-[6 S38]
orTo determine the stability of the crigin of the system in the new coordinate system, let
us apply the Liapunov stability equation given by Equation (5-90):
2.0 lpu pelf2 05]_[pu pe] __[9 0
0s oslo pal 0 08 Pir Px. 0 0.35
where we choose Q to be a positive definite matrix having elements that simplify the
computation involved Solving this last equation for matrix P, we obtain
P= [er Pel . [73 5]
Px pn} | 5 10]
By applying the Sylvester criterion for positive definiteness, we find that matrix Pis not
positive definite. Therefore, the origin (equilibrium state) is not stable.
The instability of the equilibrium state can, of course, be determined by the z
transform approach. Let us first eliminate 22 from the state equation. Then we have
i(k + 2) ~ 28E(k + 1) + 16(k) = 0
‘The characteristic equation for the system in the z plane is
F-282+1650
or
(2 ~ 2z ~ 0.8) = 0
Hence,
2=08
Since pole z = 2is located outside the unit circle in the z plane, the origin (equilibrium
state) is unstable