Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
11
,
No.
3
,
June
2021,
pp. 2
045~
2053
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ije
ce
.
v11
i
3
.
pp2045
-
20
53
2045
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om
Model r
eduction
of un
stabl
e s
yst
em
s base
d on bal
anced
trunc
ation al
gorith
m
Ngoc Ki
en
Vu
,
and
Hong
Q
ua
n
g N
guyen
Tha
i
Ngu
y
en
Un
ive
rsit
y
of Tec
h
nolog
y
,
Thai
N
gu
ye
n
Ci
ty
, V
i
et
nam
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
pr
4,
2020
Re
vised
Sep
5
,
2020
Accepte
d
Oct
11, 202
0
Model
red
uction
of
a
s
y
stem
is
an
appr
oximati
on
of
a
highe
r
-
orde
r
s
y
stem
to
a
lower
-
ord
er
s
y
stem
whil
e
th
e
d
y
n
amic
b
eha
v
ior
of
th
e
s
y
s
tem
is
al
m
ost
uncha
nged
.
In
t
his
pape
r,
we
will
discuss
m
o
del
orde
r
red
u
ction
(MO
R)
strat
eg
i
es
for
unstabl
e
s
y
s
te
m
s,
i
n
which
the
m
ethod
base
d
on
the
bal
an
ce
d
trunc
a
ti
on
al
gori
thm
will
be
foc
used
on.
Sin
ce
e
ac
h
MO
R
a
lgorithm
has
it
s
strengt
hs
and
w
ea
kness,
pr
ac
t
ical
app
li
c
at
ions
s
hould
be
suitabl
e
for
each
spec
ific
req
uirem
ent
.
Sim
ula
ti
o
n
result
s
will
demons
tra
te
the
c
orre
ct
n
ess
of
the
al
gori
thms
.
Ke
yw
or
d
s
:
Ba
la
nced
t
runc
at
ion
alg
ori
thm
High
order co
nt
ro
ll
er
Mod
el
orde
r
re
du
ct
io
n
Un
sta
ble syst
e
m
This
is an
open
acc
ess arti
cl
e
un
der
the
CC
B
Y
-
SA
l
ic
ense
.
Corres
pond
in
g
Aut
h
or
:
Hon
g Qu
a
ng
N
gu
ye
n
Thai
Nguyen
Un
i
ver
sit
y o
f Te
ch
no
l
og
y
No.
666, 3/
2
St
reet, T
hai Ng
uy
en
Ci
ty
, V
ie
tn
a
m
Em
a
il
: qu
an
g.n
gu
ye
nhong@t
nu
t.e
du.
vn
1.
INTROD
U
CTION
Mod
el
or
der
r
edu
ct
io
n
(MO
R)
is
a
m
et
ho
d
that
sim
pl
ifi
es
higher
-
orde
r
com
plex
m
od
el
s.
T
his
is
a
top
ic
receive
d
a
lot
of
at
te
ntion.
H
ow
e
ve
r,
t
her
e
sti
ll
ha
ve
m
any
issu
es
to
be
ad
dr
essed.
T
he
bal
anced
tru
ncati
on
(BT
)
al
gorithm
is
on
e
of
the
m
os
t
popula
r
m
eth
ods
of
orde
r
reducti
on
f
or
l
inear
syst
em
s.
This
m
et
ho
d
was
i
nt
rodu
ce
d
i
n
[
1]
.
Lat
er,
the
a
bi
li
ty
to
pr
ese
rv
e
sta
bili
ty
du
ri
ng
the
proc
ess
of
or
der
re
du
ct
i
on
is
pro
ved
i
n
[
2],
and
a
f
or
m
ula f
or cal
culat
in
g t
he
erro
r
li
m
it
a
ti
on
is d
et
e
rm
i
ned
i
n
[
3,
4]. In
or
der
to pe
rfor
m
the
BT
al
gorithm
,
it
is
necessary
to
perf
or
m
the
diag
on
al
iz
at
io
n
proce
ss
G
ra
m
ia
n’
s
co
ntr
ol
m
at
rix
and
Gr
a
m
ia
n’
s
ob
s
er
ver
m
at
rices
of
the
s
yst
e
m
s
i
m
ultan
eo
us
ly
.
Th
e
equ
ivale
nce
of
tw
o
dia
gonal
m
at
rices
al
lows
trans
form
ing
the
or
i
gin
al
m
od
el
re
presente
d
in
a
ny
ba
se
syst
e
m
to
an
e
qu
i
valent
base
syst
e
m
rep
rese
nted
i
n
the
co
ordinate
syst
e
m
of
inter
nal
bala
nced
s
pace.
Ba
se
d
on
Mo
or
e'
s
BT
m
et
ho
d
[1
]
,
ot
her
al
gorithm
s
hav
e
been
pr
opos
e
d
su
c
h
a
s
the
ra
ndom
balancin
g
m
et
ho
d
[
5],
po
sit
ive
real
ba
la
nce
m
et
ho
d
[
6],
Hankel
st
and
a
r
d
appr
ox
im
at
ion
m
et
ho
d
[7
]
,
et
c.
Re
cent
st
ud
i
es
on
the
m
et
ho
d
of
BT
[8
-
13]
f
ocu
s
on
de
velo
ping
or
a
djust
in
g
the
BT
al
go
rithm
cor
res
pond
ing
t
o
eac
h
s
pe
ci
fic
ap
plica
ti
on
.
T
he
B
T
al
gorithm
and
ot
he
r
al
gorithm
s
[
14
-
16
]
are
m
ai
nly
app
li
ed
to
sta
ble
l
inear
syst
em
s
because
the
ori
gin
al
co
nce
pts
of
t
he
BT
m
e
thod
(c
ontr
ollabil
it
y
Gr
am
ia
n
and
obser
va
bili
ty
Gr
a
m
ia
n)
are
al
w
ay
s
accom
pan
ie
d
by
the
requi
rem
ent
that
the
syst
e
m
is
s
ta
ble,
ie
the
syst
e
m
has
al
l
the
po
le
s
on
the
le
ft
of
th
e
i
m
aginar
y
ax
is.
In
pr
act
ic
al
app
li
cat
io
ns
,
howe
ve
r,
hi
gh
e
r
-
o
rd
e
r
li
near
m
od
el
s
(h
i
gh
e
r
-
orde
r
obj
ect
m
od
el
s,
higher
-
ord
er
c
on
t
ro
ll
ers
[17
-
19
]
)
m
ay
be
un
sta
ble.
T
he
refor
e
,
to
m
eet
the
req
uir
e
m
ents
of
the p
r
oble
m
of
or
de
r
re
du
ct
io
n,
al
gorithm
s
need
to
be
ca
pab
le
t
o
re
duce
the
or
der
of
bo
t
h
sta
ble and
unstable sy
ste
m
s.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
o
m
p
En
g,
V
ol.
11
, No
.
3
,
J
une
2021 :
20
45
-
2053
2046
To
s
olv
e
the
pr
ob
le
m
o
f orde
r
r
e
du
ct
io
n for
un
sta
ble syst
e
m
s,
there ar
e t
wo strate
gies:
Strate
gy
1
(in
direct
or
der
re
du
ct
io
n
al
gorithm
):
Firstl
y,
t
he
unsta
ble
or
iginal
syst
em
is
separ
at
e
d
in
to
sta
ble
an
d
uns
ta
ble
pa
rts.
Aft
er
that,
a
n
ord
er
re
duct
io
n
al
gorithm
is
ap
pl
ie
d
to
t
he
sta
ble
pa
rt
[17,
18
]
,
[
20
-
23]
. F
i
nally
, w
e a
dd the
s
ta
ble r
e
du
ce
d
-
order pa
rt
with
the
unsta
ble
pa
rt.
Strate
gy
2
(
di
rect
order
reducti
on
al
gorith
m
):
In
this
str
at
egy,
orde
r
r
edu
ct
io
n
al
gor
it
h
m
s
fo
r
sta
bl
e
syst
e
m
s
are
m
od
i
fied
a
nd
a
djust
ed
s
o
that
t
hese
al
go
rithm
s
can
perform
order
re
du
ct
io
n
irre
sp
ect
ive
of
wh
et
her the
or
i
gin
al
sys
te
m
is
stable
or
unsta
ble [2
4
-
28
].
Each
of
the
tw
o
ab
ov
e
order
r
edu
ct
io
n
al
gori
thm
s
has
i
ts
app
r
oac
h
an
d
nee
ds
to
be
e
valua
te
d
for
us
e
in
sp
eci
fic
a
pp
li
cat
ion
s.
T
o
pro
vid
e
a
s
pecif
ic
evaluati
on
of
pro
pose
d
MOR
al
gorithm
s
for
unsta
ble
syst
e
m
s
,
the
auth
or
s
focus
on
intr
oducin
g
an
d
eva
luati
ng
al
gorit
hm
s
fo
r
unsta
ble
syst
e
m
s
based
on
Strate
gy
2.
Sp
eci
fical
ly
,
w
e
com
par
e
thr
ee
exten
de
d
B
T
al
gorithm
s
fo
r
unsta
ble
sy
stem
s:
LQG
ba
sed
al
go
rithm
[24],
Zh
ou’s
BT
alg
or
it
hm
s [
25]
, a
nd Zil
oc
hian
B
T alg
or
it
hm
[
26]
.
2.
BT
BASED
O
RD
E
R RED
U
CTIO
N ALG
ORI
TH
MS F
OR U
NS
TAB
LE
SY
STE
M
S
2
.
1.
Pr
ob
le
m
of o
r
der red
u
ction
Give
n
a
li
nea
r,
co
ntinuo
us
,
ti
m
e
-
inv
aria
nt
s
yst
e
m
with
m
ulti
ple
inp
u
ts
,
m
ul
ti
ple
ou
tp
ut
s,
desc
r
ibe
d
in the st
at
e sp
a
ce as s
how
n
i
n
(
1
)
:
x
x
u
yx
AB
C
(1)
in which
, x
R
n
, u
R
p
, y
R
q
, A
R
nxn
, B
R
nxp
, C
R
qxn
.
The
obj
ect
iv
e
of
the
pro
blem
of
order
re
duc
ti
on
for
t
he
m
od
el
descr
i
bed
by
(
1)
is
t
o
fi
nd
th
e
m
od
el
des
cri
bed b
y
(
2
)
:
r
r
r
r
r
r
r
x
x
u
yx
AB
C
(2)
in
w
hic
h,
x
r
R
r
,
u
R
p
,
y
r
R
q
,
A
r
R
rxr
, B
r
R
rxp
,
C
r
R
qxr
,
with
r
n;
so
that
the m
od
el
descr
i
be
d
by
(
2)
can
rep
la
ce t
he
m
od
el
d
escri
be
d by (
1), a
pp
li
ed
to
an
al
yz
e,
desig
n, an
d
c
ontr
ol the s
yst
e
m
.
2
.
2.
Z
ho
u’
s
B
T alg
orith
m
As
m
entioned
in
Sect
io
n
1,
the
di
ff
ic
ulty
of
app
ly
in
g
the
BT
al
gorithm
fo
r
unsta
ble
sys
tem
s
is
that
the
determ
inati
on
of
G
ram
ian
s
al
ways
c
om
es
with
the
requirem
ent
that
the
ori
gin
al
syst
e
m
is
asym
pto
ti
c
sta
ble.
To
dete
rm
ine
the
Gr
a
m
ia
n
s
of
un
sta
ble
syst
e
m
s,
Zhou
[
25]
pro
ve
d
that
sp
eci
al
functi
ons
X
a
nd
Y,
roots
of the t
w
o
Ly
a
punov eq
uations, ca
n be
u
se
d:
'
'
0
'
'
0
X
A
A
X
X
B
B
X
A
Y
Y
A
Y
C
C
Y
(3)
Fr
om
the
two
sp
eci
al
functi
ons
X
a
nd
Y
by
set
ti
ng
F=
-
B'
X
a
nd
L=
-
YC'
,
we
can
determ
ine
the
con
t
ro
ll
abi
li
ty
Gr
am
ia
n
and
ob
s
er
vab
il
it
y
Gr
am
ia
n
of
th
e
unsta
ble
syst
e
m
by
two
Ly
apun
ov
e
quat
ion
s
as
sh
ow
n
in
(
4
)
:
'0
'
'
0
A
B
F
P
P
A
B
F
B
B
Q
A
L
C
A
A
L
C
C
C
(4)
Af
te
r
dete
rm
i
ning
th
e
c
on
t
ro
ll
abili
ty
Gr
a
m
ia
n
an
d
ob
s
erv
a
bili
ty
Gr
a
m
ia
n,
we
f
ollow
ste
ps
of
Moore'
s BT algorit
hm
to
obta
in th
e
re
du
ce
d
-
orde
r
syst
em
o
f
the
unsta
ble
or
i
gin
al
syst
em
.
Detai
l of Zh
ou'
s BT alg
or
it
hm
[25
]
is as
foll
ows:
Inp
ut
:
Syst
e
m
(A
, B
, C)
desc
ribe
d
i
n (1)
(unstable
syst
e
m
).
Step
1
:
Ca
lc
ulate
the s
pecial
fun
ct
io
ns
X
a
nd
Y
, a
cc
ordin
g
t
o
(
3).
Step
2
:
Set
F=
-
B'
X
an
d
L=
-
Y
C'
Step
3
:
Ca
lc
ulate
the c
on
t
ro
ll
abili
ty
G
ram
ia
n
P
an
d t
he obse
va
bili
ty
G
ram
ia
n
Q
, a
ccordin
g
t
o
(
4).
Step
4
:
An
al
yz
e the
fo
ll
ow
in
g
m
at
ric
es:
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Mo
del reduct
io
n of unst
ab
le
s
yst
ems ba
s
ed
on
…
(
Ngoc
Ki
en
V
u
)
2047
An
al
yz
e the
C
ho
le
s
ky m
at
rix:
P=R
R
T
, wit
h
R
is a
n u
pp
e
r
t
rian
gle m
at
rix.
An
al
yz
e S
V
D m
at
rix:
TT
R
Q
R
U
ΛV
.
Step
5
:
Ca
lc
ulate
m
at
ri
x:
1
/
2
L
V
Ca
lc
ulate
the s
ingular m
at
rix:
1
-
1
/
2
L
T
T
R
U
Step
6
:
Ca
lc
ulate
11
,
,
,
,
A
B
C
T
A
T
T
B
C
T
.
Step
7
:
Sele
ct
the
orde
r
that
nee
ds
t
o be
reduce
d
r
so t
hat
r
<
n
.
Pr
ese
nt
,
,
A
B
C
in a ma
trix fo
rm
as f
ollows:
1
1
1
2
1
12
2
1
2
2
2
,
,
,
A
A
B
A
B
C
C
C
A
A
B
In whic
h
x
x
x
1
1
1
1
,
,
r
r
r
p
q
r
R
R
R
A
B
C
.
Ou
t
pu
t
: R
e
du
c
ed
-
orde
r
syst
e
m
1
1
1
1
,
,
A
B
C
.
2
.
3.
Z
il
ochia
n
’s
e
xt
e
nded
BT
a
lg
orith
m
To
overc
om
e
t
he
dif
ficult
y
of
identify
ing
G
r
a
m
ia
ns
of
unsta
ble
syst
e
m
s,
Zil
och
ia
n
[
26]
pro
po
se
d
an
idea
of
co
nver
ti
ng
the
syst
e
m
fr
om
an
un
s
ta
ble
form
to
a
sta
ble
form
th
rou
gh
c
oor
dina
te
axis
disp
la
ce
m
ent
(or
m
app
ing).
Wh
e
n
the
syst
em
is in a
sta
ble f
orm
, w
e can
r
edu
ce t
he or
de
r
of the syst
em
accordi
ng to
t
he
B
T
al
gorithm
.
Finall
y,
the
al
gorithm
per
f
or
m
s
a
reverse
pro
j
ec
ti
on
(
re
ver
se
di
sp
la
cem
ent
of
coor
din
at
e
or
ig
in
)
t
o
conve
rt the sta
ble r
e
duced
-
or
der syst
em
into
an u
ns
ta
ble
form
li
ke
the original sy
ste
m
.
To
car
ry
ou
t
the
idea,
the
fi
rst
and
c
ru
ci
al
ste
p
is
determ
ining
the
va
lue
of
the
c
oord
i
nate
axis
disp
la
cem
ent
so
that
the
re
duced
-
order
res
ult
is
an
op
ti
m
al
so
luti
on
.
Zil
och
ia
n
[26]
dem
on
strat
ed
that
the
coor
din
at
e
a
xis
dis
placem
ent
base
d
on
the
la
rg
est
-
real
-
par
t
un
sta
ble
po
le
ha
s
an
opti
m
a
l
red
uce
d
-
orde
r
r
esult.
The
Zil
oc
hia
n’s
BT al
gorithm
[26
]
is stat
ed
a
s b
el
ow
:
Inp
ut: Syst
e
m
(A
,
B,
C)
de
scr
ibed
i
n (
1)
(uns
ta
ble syst
em
)
with a t
ran
s
fer
functi
on:
1
(
)
:
ss
G
C
I
A
B
Step
1
:
Determ
ine
the
un
sta
ble
po
le
of
(
1)
t
hat
has
the
la
rg
est
re
al
par
t.
Set
r
e
a
l
(
)
,
in
w
hic
h
R
ar
bitrary sm
al
l
and
0
.
Convert t
he
sy
stem
(A
, B
, C)
into a
n u
ns
ta
bl
e syst
e
m
()
s
G
accor
ding to
b
el
ow
:
A
A
I
BB
CC
Step
2
:
Ca
lc
ulate
the
obser
va
bili
ty
Gr
am
ia
n
Q
and
co
ntr
ollabil
it
y
Gr
am
ian
P
of
t
he
s
yst
e
m
,
,
A
B
C
by s
olv
i
ng tw
o fo
ll
owin
g
Li
apu
l
ov equati
ons:
TT
TT
,
.
A
P
P
A
B
B
A
Q
Q
A
C
C
Step
3
:
An
al
yz
e the
fo
ll
ow
in
g
m
at
ric
es:
An
al
yz
e the
C
ho
le
s
ky m
at
rix:
T
pp
P
R
R
, w
it
h
p
R
is an
uppe
r
tria
ng
le
m
at
rix.
An
al
yz
e the
C
ho
le
s
ky m
at
rix:
T
oo
Q
R
R
, w
it
h
o
R
is
an u
pp
e
r
tria
ng
le
m
at
rix.
An
al
yz
e S
V
D m
at
rix:
TT
op
R
R
U
ΛV
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
o
m
p
En
g,
V
ol.
11
, No
.
3
,
J
une
2021 :
20
45
-
2053
2048
Step
4
:
Ca
lc
ulate
the s
ingular m
at
rix
T
1
1
/
2
p
T
R
V
Λ
Step
5
:
Ca
lc
ulate
:
11
ˆˆ
ˆ
,
,
,
,
A
B
C
T
A
T
T
B
C
T
Step
6
:
Sele
ct
the
orde
r
that
nee
ds
t
o be
reduce
d
r
so t
hat
r
<
n
.
Pr
ese
nt
ˆˆ
ˆ
,
,
A
B
C
in a
m
a
trix fo
rm
as f
ol
lows
:
1
1
1
2
1
12
2
2
1
2
2
ˆˆ
ˆ
ˆ
ˆ
ˆ
ˆ
ˆ
,
,
,
ˆ
ˆ
ˆ
A
A
B
A
B
C
C
C
B
AA
in which
x
x
x
1
1
1
1
ˆˆ
ˆ
,
,
r
r
r
p
q
r
R
R
R
A
B
C
.
We
ob
ta
in
a st
able re
duced
-
orde
r
syst
em
1
1
1
1
ˆˆ
ˆ
,
,
A
B
C
.
Step
7
:
Convert
sta
ble
syst
e
m
1
1
1
1
ˆˆ
ˆ
,
,
A
B
C
into
-
sta
ble
syst
em
1
1
1
1
ˆˆ
ˆ
,
,
A
B
C
acco
rd
i
ng
to
the
fo
ll
owin
g
e
qua
ti
on
s:
1
1
1
1
11
11
ˆˆ
,
ˆˆ
,
ˆˆ
.
A
A
I
BB
CC
Ou
t
pu
t:
Re
du
c
ed
-
orde
red syst
e
m
1
1
1
1
ˆˆ
ˆ
,
,
A
B
C
2
.
4.
LQG
b
ase
d BT
The
i
dea
of
t
he
LQ
G
base
d
BT
al
go
rith
m
[2
4]
is
that
instea
d
of
ca
lc
ulati
ng
the
con
t
ro
ll
abili
ty
Gr
am
ia
n
and
t
he
obse
rv
a
bili
ty
Gr
am
ia
n
us
ing
Ly
ap
unov
equ
at
io
ns,
the
pro
po
se
d
al
go
r
it
h
m
cal
culat
es
the
m
us
in
g
t
he
e
xten
ded Ri
ccat
i eq
uation as
fo
ll
ows:
'
'
'
'
0
'
'
'
0
A
P
P
A
P
C
C
P
B
B
A
Q
Q
A
Q
B
B
Q
C
C
Af
te
r
determ
ining
the
co
ntr
ollabil
it
y
Gr
a
m
i
an
an
d
th
e
ob
serv
a
bili
ty
Gr
a
m
ia
n,
we
f
ollo
w
ste
ps
of
Moore'
s BT algorit
hm
to
obta
in the re
du
ce
d
-
orde
r
syst
em
o
f
the
unsta
ble
or
i
gin
al
syst
em
.
Detai
l of the
L
QG b
ase
d
B
T [24
]
is a
s foll
owin
g:
Inp
ut: Syst
e
m
,
,
A
B
C
descr
i
bed in
(
1) (u
ns
ta
ble syst
e
m
)
Step
1
:
Ca
lc
ulate
the c
on
t
ro
ll
abili
ty
G
ram
ia
n
P
an
d
t
he obse
r
vab
il
it
y Gr
am
ia
n
Q
arc
ordin
g
t
o
(
11)
.
Step
2
:
An
al
yz
e the
fo
ll
ow
in
g
m
at
rixes:
An
al
yz
e the
C
ho
le
s
ky m
at
rix:
T
P
R
R
, w
it
h
R
is an
uppe
r
tria
ng
le
m
at
rix.
An
al
yz
e S
V
D m
at
rix:
TT
R
Q
R
U
ΛV
.
Step
3
:
Ca
lc
ulate
m
at
ri
x:
1
/
2
L
V
Step
4
:
Ca
lc
ulate
the s
ingular m
at
ri
x:
1
-
1
/
2
L
T
T
R
U
Step
5
:
Ca
lc
ulte:
11
,
,
,
,
A
B
C
T
A
T
T
B
C
T
.
Step
6
:
Sele
ct
the
orde
r need
to be
re
du
ce
d
r
s
o
that
r
<
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Mo
del reduct
io
n of unst
ab
le
s
yst
ems ba
s
ed
on
…
(
Ngoc
Ki
en
V
u
)
2049
Pr
ese
nt
,
,
A
B
C
in a
m
a
trix fo
rm
as f
ol
lows
:
1
1
1
2
1
12
2
1
2
2
2
,
,
,
A
A
B
A
B
C
C
C
A
A
B
In whic
h
x
x
x
1
1
1
1
,
,
r
r
r
p
q
r
R
R
R
A
B
C
.
Ou
t
pu
t:
Re
du
c
ed
-
orde
r
syst
e
m
1
1
1
1
,
,
A
B
C
.
3.
SIMULATI
O
N AND
DISC
US
SI
ON
3.1.
Order
re
d
ucti
on
of hig
h o
r
der c
ontr
oller
Con
sis
der a
28
th
orde
r
c
on
tr
oll
er as f
ollow
s:
()
()
()
s
s
s
N
R
D
In whic
h
2
8
2
7
2
6
2
5
4
2
4
6
2
3
7
2
2
8
2
1
1
0
2
0
1
1
1
9
1
2
1
8
1
3
1
7
1
4
1
6
1
5
1
5
1
5
1
4
1
(
)
0
.
0
0
4
8
6
7
0
.
7
5
1
9
5
8
.
8
2
5
2
6
8
.
3
5
.
1
0
2
.
1
2
8
.
1
0
4
.
3
8
3
.
1
0
7
.
5
4
2
.
1
0
1
.
1
0
8
.
1
0
1
.
4
1
1
.
1
0
s
1
.
5
2
7
.
1
0
1
.
5
4
4
.
1
0
1
.
3
4
1
.
1
0
1
.
0
3
2
7
.
0
2
1
.
1
0
4
.
2
1
1
.
1
0
s
s
s
s
s
s
s
s
s
s
s
s
s
e
s
s
N
6
1
3
1
7
1
2
1
8
1
1
1
8
1
0
1
9
9
1
9
8
1
9
7
2
0
6
2
0
5
2
0
4
1
9
3
1
9
2
1
6
2
.
2
1
3
.
1
0
1
.
0
1
.
1
0
3
.
9
5
4
.
1
0
1
.
3
0
6
.
1
0
3
.
5
6
4
.
1
0
7
.
8
4
5
.
1
0
1
.
3
4
8
.
1
0
1
.
7
2
3
.
1
0
1
.
5
2
.
1
0
8
.
1
6
2
.
1
0
1
.
9
8
4
.
1
0
3
.
8
9
.
1
0
1
2
5
.
2
s
s
s
s
s
s
s
s
s
s
s
s
s
a
nd
5
2
8
2
7
2
6
2
5
2
4
4
2
3
6
2
2
7
2
1
8
2
0
9
1
9
1
0
1
8
1
2
1
7
1
2
1
6
1
3
1
5
1
4
1
4
1
5
1
3
(
)
5
.
2
5
0
.
0
0
9
7
8
6
0
.
8
6
7
5
4
8
.
8
1
9
6
5
6
.
0
5
6
.
1
0
1
.
4
9
.
1
0
3
.
0
1
8
.
1
0
5
.
1
4
.
1
0
7
.
4
8
3
.
1
0
9
.
4
2
5
.
1
0
1
.
0
3
5
.
1
0
9
.
9
6
8
.
1
0
+
8
.
4
3
2
.
1
0
6
.
2
6
6
.
1
0
4
.
0
7
9
.
1
0
s
e
s
s
s
s
s
s
s
s
s
s
s
s
s
s
s
s
D
1
6
1
2
1
7
1
1
1
7
1
0
1
8
9
1
8
8
1
9
7
1
9
6
1
9
5
1
9
4
1
9
3
1
8
2
4
1
1
2
.
3
1
4
.
1
0
1
.
1
3
4
.
1
0
4
.
7
4
.
1
0
1
.
6
6
.
1
0
4
.
7
6
2
.
1
0
1
.
0
8
5
.
1
0
1
.
8
9
1
.
1
0
2
.
3
9
9
.
1
0
2
.
0
6
2
.
1
0
1
.
0
6
5
.
1
0
2
.
4
7
9
.
1
0
1
.
5
9
.
1
0
2
.
9
4
5
.
1
0
s
s
s
s
s
s
s
s
s
s
s
s
In
pract
ic
e,
a
28
th
or
der
co
nt
ro
ll
er
ha
s
m
any
disadv
a
ntage
s
wh
e
n
it
co
m
es
to
real
-
tim
e
con
tr
ol,
s
o
order
re
du
ct
io
n
of
this
c
on
tr
oller
is
a
cruci
al
prob
le
m
.
The
28
th
order
c
ontr
oller
is
a
sta
ble
li
near
m
odel
bu
t
it
has
tw
o
po
le
s
t
hat
ap
pro
xim
a
t
el
y
equ
al
to
ze
ro.
The
28
th
or
der
c
ontrolle
r
i
s
us
e
d
as
a
n
obj
ect
to
eval
ua
te
the
eff
ic
ie
ncy
of
order
reducti
on
al
gorithm
s
i
ntr
oduce
d
in
s
ect
ion
2.
Re
s
ults
are
obta
in
ed
in
the
fo
ll
ow
i
ng
Tables
1
-
3
:
Table
1.
Res
ults o
f order
r
e
du
ct
io
n for a
hi
gh
-
orde
r
c
ontroll
er a
cc
ordin
g
t
o Zh
ou'
s BT alg
or
it
hm
Order
()
r
s
R
5
5
4
3
4
2
4
4
5
4
3
2
2
9
2
.
8
9
.
1
0
.
9
2
5
.
9
6
8
5
1
4
.
9
3
2
1
.
6
3
5
1
.
9
9
9
4
2
.
2
4
7
0
6
.
8
6
9
5
5
1
4
3
2
6
3
9
1
0
.
1
0
s
s
s
s
s
s
s
s
s
s
4
4
3
2
4
4
4
3
2
9
1
2
.
8
6
7
7
2
4
.
8
0
1
1
.
1
0
6
4
2
.
1
7
0
1
.
4
6
8
6
9
1
3
1
2
9
2
.
8
9
.
1
0
.
1
0
s
s
s
s
s
s
s
s
3
32
2
4
3
9
0
6
.
5
4
.
7
6
3
.
1
0
.
0
3
6
9
6
.
4
6
7
5
7
9
2
.
8
9
6
5
1
8
42
s
s
s
s
s
s
2
2
2
2
5
2
1
.
2
3
7
7
2
6
.
6
6
1
7
8
.
7
9
2
.
8
9
4
ss
ss
No
te
:
W
e
will call
th
e r
ed
u
ced
-
o
rder
co
n
troller (
o
rder
of
r
)
is an
r
-
o
rder c
o
n
troller
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
o
m
p
En
g,
V
ol.
11
, No
.
3
,
J
une
2021 :
20
45
-
2053
2050
Table
2.
Res
ults o
f
the
or
der
r
edu
ct
io
n o
f
a
hi
gh
-
orde
r
c
ontr
oller acc
ordi
ng
to
Zil
oc
hian
'
s BT alg
or
it
hm
Order
()
r
s
R
5
5
4
3
4
2
4
4
5
4
3
2
9
2
.
8
9
4
3
8
.
1
7
5
7
0
2
.
6
0
3
.
1
0
3
.
7
5
9
.
1
0
1
.
2
6
.
1
0
3
6
.
8
5
5
5
7
.
6
4
7
9
9
4
4
2
8
1
6
5
3
s
s
s
s
s
s
s
s
s
s
4
4
3
2
4
4
4
3
2
9
2
.
8
9
4
2
4
7
5
3
5
2
.
4
8
3
.
1
0
3
.
5
1
3
.
1
0
3
6
.
7
5
5
2
.
5
4
7
2
0
3
9
2
3
s
s
s
s
s
s
s
s
3
2
2
34
3
4
0
7
.
5
6
8
5
3
2
.
3
8
6
6
9
2
.
8
9
5
3
8
.
8
4
.
1
0
7
3
6
.
5
2
44
s
s
s
s
s
s
2
2
2
2
2
2
0
.
1
4
5
7
6
2
9
.
1
1
3
3
6
.
2
9
2
.
8
9
ss
ss
Table
3.
Res
ult o
f
the
ord
e
r re
du
ct
io
n o
f
a
hi
gh
-
order co
ntr
oller acc
ordin
g t
o
L
Q
G based
BT
Order
()
r
s
R
5
5
4
3
4
2
4
5
4
3
2
9
2
.
8
9
.
1
0
.
1
0
3
6
.
3
1
5
8
7
.
8
5
5
9
9
2
8
4
5
5
6
0
.
5
7
5
8
8
3
.
1
9
7
2
.
6
3
2
3
7
.
4
8
0
.
0
0
6
4
4
1
s
s
s
s
s
s
s
s
s
s
4
4
3
2
4
4
3
2
3
.
1
3
9
6
5
.
4
5
0
.
0
1
9
1
2
9
2
.
8
9
5
4
7
7
0
3
0
.
1
0
3
6
.
1
7
5
7
7
.
1
5
3
6
0
s
s
s
s
s
s
s
s
3
32
3
2
+
2
7
8
.
7
3
.
7
8
1
2
9
.
2
8
0
.
0
0
3
8
0
9
9
2
.
8
9
3
6
1
3
2
5
4
.
1
s
s
s
s
s
s
2
2
2
2
8
9
1
.
7
0
.
1
3
2
8
2
2
.
6
8
-
0
.
0
0
0
4
0
3
1
9
2
.
8
9
ss
ss
3.2.
Simul
at
io
n results
an
d
discussi
on
To
e
valuate
a
nd
ide
ntify
an
a
ppr
opriat
e
re
duced
-
order
m
odel
,
ste
p
res
po
ns
e
a
nd
f
reque
ncy
res
pons
e
of
the
ori
gin
al
co
ntr
oller
a
nd
the
re
duced
-
order
co
ntr
o
ll
er
will
be
us
e
d.
Sim
ulati
on
r
esults
are
s
hown
in
Figure
s
1
-
3.
(a)
(b)
Figure
1.
Tra
nsi
ent r
es
pons
e
a
nd freq
ue
ncy r
esp
on
se
of t
he ori
gin
al
c
on
t
rol
le
r
an
d
the
4
th
order co
ntr
oller,
(a)
t
ran
sie
nt r
e
sp
onse
, a
nd (b) fr
e
quency
res
pons
e
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Mo
del reduct
io
n of unst
ab
le
s
yst
ems ba
s
ed
on
…
(
Ngoc
Ki
en
V
u
)
2051
(a)
(b)
Figure
2.
Tra
nsi
ent r
es
pons
e
a
nd freq
ue
ncy r
esp
on
se
of t
he ori
gin
al
c
on
t
rol
le
r
an
d
the
3
rd
order co
ntr
oller,
(a)
transient
res
pons
e a
nd (b
)
f
re
qu
e
ncy
respo
nse
(a)
(b)
Figure
3
.
Tra
nsi
ent r
es
pons
e
a
nd freq
ue
ncy r
esp
on
se
of t
he ori
gin
al
c
on
t
rol
le
r
an
d
the
2
nd
orde
r
c
on
t
ro
ll
er, (a)
transient
res
pons
e,
and
(b)
fr
e
qu
e
ncy
respo
nse
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
o
m
p
En
g,
V
ol.
11
, No
.
3
,
J
une
2021 :
20
45
-
2053
2052
Fr
om
si
m
ulati
on
resu
lt
s,
w
e
se
e that:
a)
W
it
h
tra
ns
ie
nt
respo
ns
e:
The
tra
ns
ie
nt
r
esp
on
se
of
the
4th
orde
r
c
on
t
ro
ll
er
base
d
on
Zil
och
ia
n’
s
M
OR
al
gorithm
is
cl
os
est
t
o
the
tra
ns
ie
nt
re
sp
onse
of t
he o
rigin
al
syst
em
a
m
on
g t
hree c
on
si
der
e
d
al
gorithm
s.
The
tra
ns
ie
nt
r
esp
on
se
of
the
3rd
order
co
nt
ro
ll
er
based
on
Z
hou’s
MO
R
al
gorithm
is
cl
os
est
to
th
e
transient
res
pons
e
of the
ori
gi
nal syst
em
a
m
ong
t
hr
ee
consi
der
e
d
al
gorith
m
s.
Th
e tra
ns
ie
nt resp
on
ses t
he
se
cond
-
orde
r
c
ontrolle
r
ba
sed o
n
al
l con
si
der
e
d
MOR al
gorit
hm
s ar
e
m
uch
diff
e
re
nt
fr
om
the
transient
re
sp
onse
of
the
ori
gin
al
28th
or
der
co
ntr
oller.
The
seco
nd
-
order
co
ntr
olle
r
base
d on Zil
oc
hian
’s
al
go
rith
m
h
as the sm
allest
error
am
ong
th
ree c
onside
red al
gorithm
s.
b)
W
it
h fr
e
qu
e
nc
y respo
ns
e:
The
am
plit
ud
e
respo
ns
es
of
the
4th
,
3rd
,
an
d
2n
d
order
c
ontr
ollers
acco
r
ding
to
al
l
th
re
e
co
ns
ide
re
d
al
gorithm
s
are
m
uch
diff
e
re
nt
from
the
am
pl
it
ud
e
res
po
ns
e
of
t
he
ori
gin
al
c
on
tr
olle
r,
in
w
hich
t
he
error
of
the
m
et
hod
based
on
LQR
is
sm
al
lest.
The
sim
ulatio
n
res
ults
of
the
th
ree
m
et
ho
ds
al
s
o
sho
w
that t
he
l
ow
e
r
t
he or
der o
f
t
he c
on
t
ro
ll
er,
the
higher
the e
rro
r.
c)
Thro
ugh
a
naly
sis of tra
ns
ie
nt
and f
reque
ncy
respo
ns
es,
w
e
see that
:
W
it
h
transie
nt
re
spo
ns
e:
redu
ced
-
orde
r
co
ntro
ll
ers
based
on
Zil
och
ia
n’
s
BT
al
go
rithm
s
giv
e
the
best
resu
lt
s
(closest
to
the
tra
ns
ie
nt
s on
the
origin
al
co
nt
ro
ll
er
).
W
it
h
fr
e
quenc
y
res
pons
e:
All
re
du
ce
d
-
orde
r
co
ntr
oller
give
not
good
res
ults.
(ch
a
racter
ist
ic
s
are
m
uch
diff
e
r
e
nt fro
m
the origi
nal contr
oller sett
ings).
d)
Fr
om
h
ere
w
e
see that
:
If
the
re
quire
m
ent
of
the
or
der
re
duct
io
n
pro
blem
is
m
a
inly
con
cer
ne
d
with
transie
nt
response
error,
we
s
houl
d
c
hoos
e t
he
Zil
oc
hi
an’
s
MOR al
gorithm
.
If
the
re
qu
i
re
m
ents
of
t
he
pro
blem
of
or
de
r
re
du
ct
i
on
ta
ke
into
acco
un
t
bo
th
tran
sie
nt
response
an
d
fr
e
qu
e
ncy
respon
s
e,
we
s
houl
d
c
hoos
e
the L
QG b
ase
d al
go
rithm
.
Th
us
,
al
l
three
direct
MOR
a
lgorit
hm
s
are
capab
le
of
re
duci
ng
or
der
for
unsta
ble
syst
e
m
s
with
thei
r
adv
a
ntage
s
a
nd
disad
van
ta
ge
s.
T
o
c
ho
os
e
a
n
a
pprop
riat
e
MOR
m
et
ho
d
for
un
sta
ble
sy
stem
s,
we
ne
e
d
to
base
on
the
requirem
ents
s
et
out
f
or
the
pro
blem
of
or
der
re
duct
ion
f
or
unsta
ble
sys
tem
s
to
sel
ect
the app
ropr
ia
te
algorit
hm
.
4.
CONCL
US
I
O
N
The
pa
per
has
introdu
ce
d
thr
ee
MOR
al
go
r
it
h
m
s
fo
r
un
st
able
syst
e
m
s
b
ased
on
the
BT
al
go
rithm
.
By
analy
zi
ng
these
MOR
al
gorithm
s
through
il
lustrati
ve
exam
ples,
we
see
that
each
al
gorithm
has
it
s
adv
a
ntage
s
an
d
disa
dvanta
ge
s.
From
these
assessm
ents,
we
ca
n
ch
oose
the
M
OR
m
et
ho
d
unde
r
the
requirem
ents
of
the
p
r
oble
m
of
orde
r
re
duct
ion
for
unsta
bl
e
syst
e
m
s.
Our
on
-
go
i
ng
rese
arch
directi
on
is
to
evaluate
a
nd
c
om
par
e
oth
e
r
i
nd
i
rect
MOR
a
lgorit
hm
s
with
the
three
m
entione
d
al
gorithm
s
an
d
use
the
r
esults
in s
pecific a
pp
l
ic
at
ion
s to
h
a
ve
a m
or
e accu
r
at
e q
ualit
y eval
ua
ti
on of
reduc
ed
-
orde
r
syst
e
m
s.
ACKN
OWLE
DGE
MENTS
This
resea
rch
was
f
unde
d
by
Thai
Nguyen
U
niv
e
rsity
of
Tech
no
l
ogy,
No
.
666,
3/2
Street
,
Thai
Nguyen
Ci
ty
, V
ie
t Nam
.
REFERE
NCE
S
[1]
Moore
B.
C.
,
“
Princi
p
al
compon
ent
an
aly
sis
in
l
i
nea
r
s
y
s
te
m
s:
Control
la
b
il
i
t
y
,
ob
serva
bilit
y
,
and
m
odel
red
uction,
”
IEE
E
Tr
ans.
Auto.
Contr
.,
vo
l.
2
6,
no
.
1
,
pp
.
17
-
32,
1981
,
doi
:
10
.
1109/T
AC.1981
.
1102568
.
[2]
Pere
nbo
I.,
and
Silve
rm
an
L.
M.
,
“
Model
red
uc
tion
via
balan
c
ed
stat
e
spa
ce
rep
r
e
-
sentation,”
IE
EE
Tr
ans.
Aut
o
,
cont
r.
,
vol
.
27
,
n
o.
2
,
pp
.
328
-
38
7,
1982
,
doi
:
10
.
1109/T
AC.1982.1102945
.
[3]
Glover
K.
,
“
All
opti
m
al
Hanke
l
norm
appr
oxim
at
ion
of
li
n
ea
r
m
ult
iva
ri
abl
e
s
ystem
and
the
ir
L2
err
or
bounds,”
IEE
E
Tr
ans.
Auto.
Contr.
,
vo
l. 2
9
,
no
.
6
,
pp
.
110
5
-
1113,
1984
,
d
oi:
10
.
1080/002
07178408933239
.
[4]
D.
Enns,
“
Model
r
educ
t
ion
with
ba
la
nc
ed
reali
z
ations:
A
n
err
or
bou
nd
and
a
fr
e
quency
we
ighted
gene
ra
li
z
at
ion
,
”
Proc.
23rd
IEEE
Conf. Decision
and
Contro
l,
198
4,
doi
:
10
.
1109/
CDC.1984.
2722
86
.
[5]
Desai
U.
B
.
,
an
d
Pal
D.
,
“
A
Tr
ansform
at
ion
A
pproa
ch
to
Stoc
hasti
c
Mode
l
R
educ
t
ion,
”
IE
EE
Tr
ansacti
ons
on
Aut
omatic Cont
r
ol,
vo
l. 29, no. 1
2,
pp
.
1097
-
110
0,
1984
,
doi
:
10.
1
109/T
AC.1984.
1
103438
.
[6]
Gree
n
M.
,
“
A
Rel
ative
Err
or
Bound
for
Bal
anc
ed
Stocha
sti
c
Trunc
at
ion
,
”
I
EE
E
Tr
ans.
Aut
o.
Contr.
,
vol.
33,
no.
10
,
pp
.
961
-
965,
1988
,
doi
:
1
0.
1109/9.
7255
.
[7]
Antoula
s
A.
C.
,
Sorensen
D.
C.
,
Gugerc
in
S.
,
“
A
Surve
y
of
Model
Redu
ct
ion
M
e
thods
for
La
r
ge
-
sca
le
S
y
stems
,
”
Struct
ured
Ma
tri
ce
s in
Math
emat
ic
s,
Comput
er
Sc
ie
nc
e, and
En
gin
ee
ring
,
AMS
200
1,
pp
.
193
-
219
,
2001.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Mo
del reduct
io
n of unst
ab
le
s
yst
ems ba
s
ed
on
…
(
Ngoc
Ki
en
V
u
)
2053
[8]
H.
Sandber
g,
“
Model
red
uc
ti
on
of
li
ne
ar
s
y
st
e
m
s
using
ext
ende
d
balanc
ed
tru
nca
t
ion,
”
IEEE
Ame
rican
Contr
ol
Confe
renc
e
,
200
8
,
pp
.
4
654
-
465
9,
doi
:
10
.
1109/
ACC.2008.
4587
229
.
[9]
Heink
ensc
hloss,
M.,
Sorensen,
D.
C.
,
and
Sun,
K.,
“
Bal
an
ced
Trunc
ation
Model
Reduc
t
io
n
for
a
Cla
ss
of
Descri
ptor
S
y
st
ems
with
Applic
ation
to
the
Os
ee
n
Eq
uat
ions
,
”
SIAM
J.
Scie
nti
fic
Computin
g,
vol.
30,
no.
2,
pp.
1038
-
1063
,
2008,
doi
:
10.1137/0
7068
1910
.
[10]
St
y
ke
l,
T
.
,
“
Balance
d
truncat
io
n
m
odel
red
u
ct
i
on
for
sem
idi
sc
ret
i
ze
d
Stok
es
e
quat
ion
,
”
Lin
ea
r
Al
gebra
Ap
p
l.
,
vol.
415
,
no
.
2
-
3
,
pp
.
262
-
289
,
2
006,
doi
:
10
.
101
6/j
.
l
aa.2004.
01
.
0
15
.
[11]
Duff,
I.
P.
,
Go
yal
,
P.K.,
and
Benne
r,
P.,
“
Bal
an
ce
d
Trunca
t
ion
for
a
Spec
ia
l
C
la
ss
of
Bi
li
ne
ar
Descri
p
to
r
S
y
stems
,
”
IE
EE Cont
rol S
yste
ms
Letters
,
vol
.
3
,
p
p.
535
-
540
,
201
9,
doi
:
10.
1109
/LCS
YS
.
2019.
2911904
.
[12]
M.
G.
Safonov
and
R.
Y.
Chiang,
“
A
Sc
hur
Method
for
Ba
l
anc
ed
Mod
el
R
educ
t
ion,
”
1988
Ame
rican
Contr
ol
Confe
renc
e
,
198
8,
doi
:
10
.
23919
/ACC.1988.
478
9873
.
[13]
Benne
r
P.
,
Hos
sain
MS
.
,
St
y
kel T
.
,
“
Model
Redu
ct
ion
o
f
Period
ic
Descri
ptor
S
y
st
ems
Us
ing
Bal
a
nce
d
Truncat
ion
,
”
In:
Be
nn
er
P
.
,
Hinze
M.
,
t
er
M
ate
n
E
.
(
eds)
Mode
l
R
eduction
f
or
Circui
t
S
imul
ati
on.
Lect
ure
N
ote
s
in
Elec
tri
ca
l
Engi
ne
ering
,
Springer, Dordre
c
h
t,
vo
l. 74,
2011.
[14]
Hade
el
N
.
Abdulla
h
,
“
A
h
y
bri
d
bac
t
erial
fora
ging
and
m
odified
par
t
ic
l
e
sw
ar
m
opti
m
iz
at
ion
for
m
odel
orde
r
red
uction
,”
In
te
r
nati
onal
Journal
of
El
e
ct
rica
l
and
Computer
Engi
nee
ring
(
IJECE)
,
vol
9,
no
2,
pp.
1100
-
110
9,
2019,
doi
:
10.
115
91/i
j
e
ce.v9i
2
.
pp.
1100
-
1109.
[15]
I.
Kasire
dd
y
,
A.
W
.
Nasir,
and
A.
K.
Singh,
“
N
on
-
int
eg
er
IMC
Based
PID
Desi
gn
for
Loa
d
Freque
nc
y
Con
trol
of
Pow
er
Sy
st
em
through
Reduc
ed
Model
Order
,”
Inte
rnational
J
ournal
of
El
e
ct
rical
and
Com
pute
r
Engi
n
ee
r
ing
(IJECE)
,
vo
l. 8,
no.
2
,
pp
.
837
-
8
44,
2018
,
doi
:
1
0.
11591/i
j
ece.
v8
i2.
pp837
-
844
[16]
D.
Sriniva
sa
Ra
o,
M.
Siva
Kumar
,
M.
Ramali
ng
a
Raj
u,
“
Design
of
Robust
Cont
roll
er
for
Highe
r
Order
Inte
rva
l
S
y
stem
Us
ing
Diffe
ren
t
ia
l
Evo
lu
ti
onar
y
Algor
i
thm
,”
Inte
rnat
io
nal
Journal
of
Robot
ic
s
and
A
utomati
on
(
IJR
A
),
vol.
7
,
no
.
4
,
pp
.
232
-
250,
2019
,
doi:
10
.
11591/ijr
a.
v7i4
.
pp232
-
25
0.
[17]
Cong
Huu
Ngu
y
en,
Kien
Ngoc
Vu,
and
Hai
Tr
ung
Do,
“
Model
red
uction
base
d
on
tri
ang
le
r
e
alization
with
po
l
e
ret
en
ti
on,
”
Ap
pli
ed
Ma
the
m
ati
cal
Scienc
es,
vol
.
9,
2
015,
no.
44
,
pp.
2187
-
21
96,
2015,
do
i
:
10.
12988/a
m
s.2
015.
5290.
[18]
Kiki
Mus
ta
qim,
Didik
Khus
nul
Arif,
E
rna
April
ia
ni
and
Diek
y
Adzki
y
a,
“
Model
red
u
ct
ion
of
unstabl
e
s
y
s
te
m
s
using
bal
anced
t
runc
ation
m
et
hod
and
i
ts
appl
ica
ti
on
to
shall
ow
wate
r
equations,”
Journal
of
Phys
ic
s:
Confe
renc
e
Serie
s,
vo
l. 855,
no.
1
,
2017
.
[19]
Rodrígue
z
,
D.
F.
,
Cuellar
,
B.
D
.
,
Sename,
O.
,
and
Vill
a
,
M.
V.
,
“
On t
he
stab
il
i
zatio
n
of
high
orde
r
sy
stems
with
tw
o
unstabl
e
pol
es
p
lus
ti
m
e
dela
y
,”
20th
Me
diterr
anean
Confe
ren
c
e
on
Control
&
Aut
omation
(
MED)
,
2012,
doi:
10.
1109/MED.
2
012.
6265607
.
[20]
C.
S.
Hs
u
and
D
.
Hou,
“
Reduc
in
g
unstabl
e
li
n
ear
cont
rol
s
y
st
ems
via
rea
l
schur
tra
nsform
at
ion,”
El
ectron.
Lett.,
pp.
984
-
986
,
19
91,
doi
:
10
.
1049
/e
l:
19910614
.
[21]
S.
K.
Nag
ar
and
S.
K.
Singh,
“
An
al
gori
thmic
appr
oac
h
for
s
y
s
te
m
dedc
om
posi
ti
on
and
ba
la
n
ced
reali
z
ed
m
ode
l
red
uction,
”
J
.
o
f Frankl
in
Inst.
,
v
ol.
341
,
pp
.
615
-
630,
2004
,
doi
:
10.
1016/j.jf
r
anklin.
2004.
07
.
005
[22]
Varga
,
“
Model
r
educ
t
ion
softwar
e
in
the
sl
ic
o
t
l
i
bra
r
y
,
”
App
li
ed
and
Computati
o
nal
Control,
Sig
nals
and
Circuit
s,
vol.
629
,
pp
.
239
-
282,
2001
.
[23]
Fatmawat
i,
Sar
agi
h,
R.
,
Bambang,
R
.
T
.
,
an
d
Soehar
y
ad
i,
Y.
,
“
Balance
d
trunc
a
ti
on
for
unstabl
e
infi
ni
t
e
dimensional
s
y
s
te
m
s
via
r
ec
ip
ro
ca
l
tra
nsform
ati
on,
”
In
te
rnat
ion
al
Journal
of
C
ontrol,
Aut
omati
on
and
Syst
ems,
vol.
9
,
pp
.
249
-
2
57,
2011
.
[24]
Jonckhe
ere
E.
A.,
Silve
rm
an
L
.
M.,
“
A
New
Set
of
Inva
r
ia
n
ts
for
Li
ne
ar
S
y
st
e
m
-
Applic
at
ion
t
o
Reduc
ed
Ord
e
r
Com
pensa
tor
De
sign,”
IE
EE Tr
ansacti
ons on Aut
omatic
Con
trol,
AC
-
28,
n
o
.
10
,
pp.
953
-
964
,
19
83.
[25]
Zhou
K.,
Salo
m
on
G.,
W
u
E.
,
“
Balanc
ed
rea
l
iz
a
ti
on
and
m
odel
red
uct
io
n
m
et
hod
for
unstabl
e
s
y
s
te
m
s,”
Inte
rnational
Jo
urnal
of Robust and Nonli
n
ear
C
ontrol,
vo
l. 9, no
.
3
,
pp
.
183
-
198
,
1999.
[26]
Zi
loc
h
ia
n
A.
,
“
Bal
anc
ed
Stru
ct
ure
s
and
Mo
del
Redu
ct
ion
of
Uns
ta
ble
S
y
stems
,”
IE
E
E
Proceedi
ngs
of
Southe
astcon
91
,
vol
2,
pp.
1198
-
1201,
1991
.
[27]
Boess,
C.
,
Ni
ch
ols,
N.,
Bunse
-
Gerstne
r,
A.
,
“
Model
red
uc
ti
o
n
for
discrete
unstabl
e
cont
rol
s
y
stems
using
a
bal
an
ce
d
truncat
ion
appr
o
ac
h
,
”
P
reprint
MPS
,
Un
iv
ersit
y
of
R
eading
,
2010
.
[28]
Boess,
C.
,
L
awl
ess,
A.,
Nichol
s
,
N.,
Bunse
-
Gerst
ner
,
A.
,
“
Stat
e
es
ti
m
at
ion
using
m
odel
orde
r
r
eduction
for
unstable
s
y
stems
,
”
Comp
ut.
Fl
uids
,
vol
.
3
6,
no
.
1
,
pp
.
155
-
160,
2011
,
doi
:
10.
1016/j.c
om
pflui
d.
2010
.
11.
033
.
Evaluation Warning : The document was created with Spire.PDF for Python.