Int
ern
at
i
onal
Journ
al of
P
ower E
le
ctr
on
i
cs a
n
d
Drive
S
ystem
(I
J
PE
D
S
)
Vo
l.
11
,
No.
3
,
Septem
be
r
2020
, pp.
1333
~
1434
IS
S
N:
20
88
-
8694
,
DOI:
10
.11
591/
ij
peds
.
v
1
1
.i
3
.
pp
1333
-
1
34
3
1333
Journ
al h
om
e
page
:
http:
//
ij
pe
ds
.i
aescore.c
om
Model
refer
ence
self
-
t
uni
ng fracti
onal o
rder
PID
contr
ol b
ased
on for a
powe
r system
stabiliz
er
M.
A. A
bdel
Ghany
1
,
Moh
amed
A
.
Sha
msel
din
2
1
Depa
rtment of
El
e
ct
ri
ca
l
Eng
in
ee
ring
,
Fa
cul
ty
o
f
Engi
n
ee
ring
O
ct
ober
6
Univ
ersit
y,
Egypt.
2
Facul
ty
of Engi
nee
rin
g
,
Futur
e Universit
y
in
Eg
ypt,
Egypt
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
J
ul
6
,
201
9
Re
vised
Feb 2
,
20
20
Accepte
d
Apr
8
, 2
0
20
Thi
s
pape
r
pr
ese
nts
a
novel
appr
oac
h
of
self
-
tuning
for
a
Modified
Frac
ti
on
al
Order
PID
(MF
OP
ID)
depe
nds
on
the
Model
R
efe
ren
ce
Adapt
i
ve
Sys
te
m
(MR
AS
).
The
p
roposed
self
-
tun
ing
cont
ro
ll
er
is
appl
i
ed
to
Po
wer
Sys
te
m
Stabi
lizer
(PS
S).
T
aka
j
i
-
Sugeno
(TS)
fuz
zy
lo
gic
t
ec
hniqu
e
i
s
used
to
construc
t
the
M
FO
PID
cont
roll
e
r.
Th
e
ob
jecti
v
e
of
MRA
S
is
to
update
the
five
p
ara
m
eters
of
Ta
k
aj
i
-
Sugen
o
Modifie
d
FO
PI
D
(TSMF
OP
I
D
)
cont
ro
ll
er
onli
ne
.
For
diff
e
ren
t
op
era
t
ing
p
oint
s
of
PS
S,
MRA
S
is
appl
ie
d
t
o
inve
stig
ate
the
eff
ec
t
ive
n
e
ss
of
proposed
cont
rol
le
rs.
The
har
mony
opti
mization
te
chn
ique
used
t
o
obt
ai
n
the
op
t
im
al
p
ara
m
eters
of
TSMF
OP
ID
con
trol
l
ers
and
MRA
S
par
am
e
te
rs.
For
diffe
ren
t
oper
ating
point
s
wi
t
h
diffe
r
ent
disturba
nc
e
und
er
pa
ramet
ers
v
ari
a
ti
ons
th
e
si
mul
ation
r
esult
s
are
obt
ai
n
ed.
Thi
s
is
to
show
tha
t
Se
lf
-
Tuni
n
g
of
TSMF
OP
I
D
base
d
on
(MRA
S)
have
bet
t
er
pe
rform
an
ce
tha
n
the fi
x
ed
par
a
me
t
ers
TS
MO
FOPID c
ontr
oll
er
.
Ke
yw
or
d
s
:
Fr
act
io
nal or
de
r
P
ID
Harmo
ny r
e
sea
rch (
HS
)
M
odel
ref
e
ren
c
e ada
ptive
con
t
ro
l
(
M
RA
C)
Power syste
m
sta
bili
zer (
P
SS
)
Takaji
-
s
ug
e
no
fu
zz
y
This
is an
open
acc
ess arti
cl
e
un
der
the
CC
BY
-
SA
l
ic
ense
.
Corres
pond
in
g
Aut
h
or
:
M
. A.
Ab
del Ghan
y,
Dep
a
rteme
nt
of Elec
tric
al
E
nginee
rin
g,
Faculty
of E
ngineerin
g Octo
be
r 6
Un
i
ver
sit
y,
Emai
l:
mghany
1988@
hotmai
l.com
1.
INTROD
U
CTION
Gen
e
rato
r
e
xc
it
at
ion
co
ntr
ol
sy
ste
ms
co
nt
ai
n
A
utomat
ic
Vo
lt
a
ge
Re
gula
tors
(
AV
R
)
for
vo
lt
age
regulat
ion
a
nd
co
nventi
on
al
Power
S
ys
te
m
Stabil
iz
ers
(C
PSS)
for
da
m
ping
mec
han
ic
al
mode
os
ci
ll
at
ion
s.
The
c
ha
nges
i
n
operati
ng
c
onditi
ons
of
PS
S
is
c
halle
nge
to
update
t
he
c
on
t
ro
ll
er
p
a
ra
mete
rs
[
1]
.
T
he
refor
e
,
the
ne
w
st
ud
ie
s
seek
t
o
de
sign
ad
va
nced
c
on
t
ro
l
te
ch
niques,
w
hich
c
ontr
ollers
ada
pt
with
the
co
nt
inuous
chang
e
in
ope
r
at
ing
points
[
2
-
4].
T
he
c
onve
ntion
al
P
ID
c
ontr
oller
is
co
m
mon
us
e
in
se
ver
al
of
e
ngin
eerin
g
app
li
cat
io
ns
.
D
ue
t
o
the
str
uc
ture
s
impli
ci
ty
an
d
ea
sy
pa
ra
mete
r
t
un
i
ng,
i
t
is
su
it
able
f
or
a
certai
n
ope
rati
ng
po
i
nt.
In
a
dd
it
ion
,
it
s
perfor
mance
is
good
f
or
li
near
an
d
simple
syst
ems
[5
,
6].
Sti
ll
,
the
be
ha
vior
of
PID
con
t
ro
l i
s li
nea
r
an
d
ca
nnot deal
w
it
h
the
hig
h dist
urb
a
nce
and
high
nonlinearit
y
i
n
co
m
plica
te
d
sy
ste
ms [5
,
7
,
8].
The
c
urre
nt
resea
rch
di
rected
to
us
e
the
Fr
act
io
nal
Order
PID
(
FO
P
ID)
c
on
tr
ol
w
her
e
it
presents
the
no
nlinear
face
of
P
ID
c
on
t
ro
l
[9
-
11
].
In
FOPI
D
c
ontrolle
r,
t
wo
a
ddit
ion
al
par
a
m
et
ers
(the
f
rac
ti
on
al
int
egr
al
a
nd
de
rivati
ve
gain
s
)
will
be
s
uppl
ementar
y
to
i
ncr
ease
the
fle
xib
il
it
y
a
nd
re
li
abili
ty
of
co
ntr
oller
[12
-
14]
.
T
herefo
re,
t
he
dyna
mic
pe
r
forma
nce
of
FO
P
ID
co
ntr
oller
is
e
nhanced
co
m
pa
red
t
o
the con
ven
ti
onal
PID c
on
tr
oller [1
5
-
17].
At
dif
fe
re
nt
op
erati
ng
points
for
a
certai
n
s
yst
em,
ada
ptati
on
onli
ne
was
use
d
sel
f
-
tu
ning
us
i
ng
for
the
s
ys
te
m.
I
n
this
case,
t
he
f
uzzy
lo
gic
cal
c
ulati
on
s
nee
d
a
long
ti
me
a
nd
ad
diti
on
e
ffo
rts
by
tr
y
a
nd
e
r
ror
is
performe
d
to
ob
ta
in
normali
zi
ng
gains
sel
ect
ion
[
18].
S
o,
this
stu
dy
r
esor
t
to
the
M
RA
S
t
o
sel
f
-
tu
ning
the
T
SMFO
PID
onli
ne
w
here
it
has
simpl
e
str
uctu
re,
ea
sy
to
impleme
nt
a
nd
fast
ca
lc
ulati
on
s
[
19
,
20].
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
Dr
i
S
ys
t
,
V
ol
.
1
1
, N
o.
3
,
Se
ptembe
r
2020
:
13
3
3
–
1
343
1334
The
m
odel
-
refe
ren
ce
ad
a
ptiv
e sy
ste
m
(MR
AS
)
prese
nts one
of
t
he best a
dap
ti
ve
contr
ol
techn
i
q
ue
s. It
may be
reg
a
rd
e
d
as
a
n
ada
ptive
ser
vo
s
ys
te
m
in
w
hich
the
desi
re
d
perf
or
m
ance
is
e
xpresse
d
i
n
te
r
ms
of
a
re
fer
e
nc
e
model,
w
hich
giv
es
the
desir
ed
res
pons
e
t
o
a
comma
nd
s
ign
al
[
20].
It
f
or
ces
t
he
over
al
l
sy
ste
m
to
f
ollow
the
beh
a
vior
of
pr
esel
ect
e
d
model
r
efe
rence
.
T
he
presel
ect
ed
m
odel
c
an
be
first
or
seco
nd
or
der
s
ys
te
m
accor
ding t
o
th
e point
of v
ie
w
the
desig
ner a
nd compli
cat
ed
d
e
gr
ee
of the
s
ys
te
m
[19
].
In
this
Stu
dy,
Takaji
-
S
ug
e
no
F
uzzy
T
ype
3
(T
S
-
F
uzz
y
T
yp
e
3)
is
desig
ned
by
21
r
ule
tria
ngula
r
membe
rs
hip
s
.
In
t
he
TS
MFO
PI
D
desi
gn,
th
e
powe
r
of
S
operat
or
for
F
O
PI
D
is
li
ed
between
zer
o
<
λ
>
2
f
o
r
integral
f
racti
on
orde
r
an
d
zer
o
<
μ
>
2
f
or
de
rivati
ve.
Let
the
FOPI
D
ope
r
at
ed
via
t
he
Ni
ntege
r
To
olbo
x
with
internall
y
unkn
own
fi
ve
par
a
mete
rs
(
kp,
ki,
kd,
λ
a
nd
μ)
be
na
med
as
(
TB
FO
P
ID).
Wh
il
e,
a
m
od
i
fied
F
OP
I
D
wh
ic
h
ha
s
ext
ern
al
ly
unkn
own
five
pa
ram
et
ers
(
kp,
ki,
kd,
λ
a
nd
μ
)
c
on
st
ru
ct
e
d
de
s
ign
e
d
by
T
S
f
uzzy
is
name
d
a
s
T
S
M
F
OPID
[
21]
.
T
he
perf
or
ma
nce
of
t
he
T
S
M
F
OPID
base
d
on
model
refe
ren
ce
a
dap
ti
ve
s
ys
te
m
con
t
ro
ll
er
ca
n
be
imp
r
ov
e
d
usi
ng
dif
fer
e
nt
typ
e
s
of
opti
mi
zat
ion
te
ch
nique
Ha
rm
ony
S
earch
(HS)
[22
,
23]
.
This
pa
pe
r
pr
esents,
a
new
com
bin
at
io
n
betwee
n
m
od
i
fied
F
OPID
c
on
t
ro
ll
er
base
d
on
TS
te
c
hniq
ue
(TSMF
OPID
) a
nd
M
odel
r
e
fe
ren
ce
as a t
une
r
to
d
esi
gn a
n
e
w
a
dap
ti
vel
y o
utput fee
dback
contr
oller for
PSS.
2.
POWER
S
YST
EM ST
ABILIZ
ER MOD
EL
A
sin
gle
mac
hi
ne
-
in
finite
bu
s
sy
ste
m
w
ho
s
e
li
near
iz
ed
i
nc
reme
ntal
mod
el
con
ta
ini
ng
t
he
volt
age
regulat
or
a
nd
excit
er
ca
n
be
de
monstrate
d
by
the
bl
oc
k
di
agr
am
as
sho
w
n
in
Fig
u
re
1.
The
par
a
me
te
rs
of
the
s
ys
te
m
a
re
gi
ven
in
Ta
ble
1
[1
,
24
].
A
num
ber
of
ca
ses
are
done
t
hat
c
ov
e
r
dif
fe
ren
t
operati
ng
po
i
nts
(normal
,
hea
vy
an
d
li
ght
loa
d
co
ndit
ion
s
)
an
d
pa
rameters
va
riat
ion
in
the
pr
ese
nce
of
a
sever
e
disturba
nce.
These
cases
a
r
e
a
ppli
ed
t
o
t
he
s
ys
te
m
wit
h
op
ti
mal
FOPID
a
nd
sel
f
-
tun
i
ng
by
M
od
el
Re
fer
e
nce
c
on
t
ro
l.
The
c
onsta
nts
K
=
[K1,
…,
K6]
in
the
nor
mal,
hea
vy
a
nd
li
ght
loa
ds
a
r
e
giv
e
n
i
n
Ta
bl
e
2.
The
pa
ram
et
ers
of
the s
ys
te
m a
re
giv
e
n
in
Ta
ble
2
[2
,
25].
Figure.
1. Line
ari
zed i
ncr
e
me
ntal mo
del
of s
yn
c
hro
nous ma
chine wit
h
a
n e
xcite
r
a
nd stabi
li
zer.
Table
1.
T
he
paramet
ers
of t
he
sy
ste
m
Para
m
eter
Valu
e
Para
m
eter
Valu
e
Ka
400
D
0
Ta
0
.05
Ef
d
m
ax
7
.3
Tdo
5
.9
Ef
d
m
in
-
7
.3
M
4
.74
u
m
ax
0
.12
Kf
0
.02
5
u
m
in
-
0
.12
Tf
1
0
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
P
ow Elec
& Dri S
ys
t
IS
S
N:
20
88
-
8
694
Mo
del refe
renc
e self
-
tun
i
ng fr
actional
order
PID co
ntr
ol ba
sed o
n
…
(
M.
A. A
bd
el
G
hany
)
1335
Tab
le
2.
O
per
a
ti
ng
c
onditi
ons
for
K
_1 to K
_6
K
OP1
=[1.0
,
1
.0]
No
rm
al
load
OP2
=[1.3
,
0
.9]
Heavy
load
OP3
=[0.8
,
1
.2]
Ligh
t load
K1
1
.07
5
3
0
.62
3
4
1
.40
7
6
K2
1
.25
8
1
1
.28
1
3
1
.19
8
4
K3
0
.30
7
1
0
.30
7
1
0
.30
7
1
K4
1
.71
3
1
1
.71
2
3
1
.64
6
1
K5
-
0
.04
7
6
-
0
.20
9
1
0
.07
4
2
K6
0.
4972
0
.45
6
5
0
.54
8
8
3.
CONTR
OL T
ECHNIQ
UES
This
sect
io
n
s
hows
the
pro
pose
d
c
on
tr
ol
t
echn
i
qu
e
s.
Th
e
first
te
c
hn
i
que
is
t
he
Modi
fied
F
OPID
con
t
ro
ll
er
base
d
on
Harmo
ny
search
.
T
he
s
econd
te
ch
ni
que
is
the
sel
f
-
t
un
i
ng
of
m
od
i
fied
F
OPID
c
ontr
oller
base
d on opti
m
al
M
RA
S.
3.1.
The FOPI
D C
on
t
rol
The
To
olbo
x
FO
P
ID
(TBF
OP
I
D
)
is
usu
al
ly
use
d
to
s
imulat
e
the
F
OP
I
D
co
ntr
ol.
It
has
fi
ve
par
a
mete
rs
int
ern
al
ly
sel
ect
e
d
by
de
sig
ner
in
on
e
cl
os
e
d
bl
ock
as
s
how
n
in
Fi
gur
e.2
.
T
o
ma
ke
adap
t
the
F
OPID
c
ontrol
onli
ne
a
T
akaji
-
Suge
no
F
uzz
y
(ty
pe
-
3
fu
zz
y)
te
ch
nique
is
de
vel
op
e
d
FOPI
D
has
e
xter
nal
five
te
r
minals
par
a
mete
rs
as
s
how
n
in
Fig
ure
3 [
21
,
26].
Figure
2. The
blo
c
k diag
ram wit
h
inte
rn
al
fi
ve
unknow
n para
mete
rs
kp, ki
,
kd, λ
an
d μ
Figure
3. The
blo
c
k diag
ram wit
h
the
ex
te
rnal
five
par
a
mete
rs
t
o be s
uitable
fo
r mo
del r
e
fer
e
nc
e self
-
tun
in
g
The
desig
n
ste
ps
of
T
SMFO
I
and
TS
M
F
O
D
can
be
s
ummari
zed
a
s
f
ollow
s
[
21]
.
T
he
first
ste
p,
le
t
the
in
pu
t
T
S
f
uzzy
mem
bers
hip
functi
ons
f
or
t
he
f
racti
on
al
order
s
of
th
e
integral
a
nd
der
i
vative
(λ
a
nd
μ
)
va
lues
are
sel
e
ct
to
be
21
t
riangular
me
mbe
r
s
hip
s
functi
on
s
.
The
un
i
ve
rse
of
disco
urs
e
val
ues
a
re
e
qu
al
ly
distrib
uted
ove
r
the
range
[0,
2]
a
nd
ha
ve
t
he
ir
mid
dle
ve
rtic
es
placed
at
t
he
points
{
0,
0.1,
0.2,
0.3
,
0.4
,
0.5,
0.6,
0.7,
0.8
,
0.9,
1,
1.1,
1.2,
1.3
,
1.4,
1.5
,
1.6,
1.7,
1.8,
1.9,
2}
.T
he
me
mb
e
rsh
i
p
sel
ec
te
d
by
21
tria
ngula
r
membe
r
s
hip
s
as
show
n
i
n
Fi
g
ure
4.
T
he
bl
ock
of
TS
MFO
D
or
T
SMFO
I
and
TS
M
F
OPI
D
21
ru
le
s
re
presents
in Figu
re. 5.
Figure
4. I
nput
mem
ber
s
h
ip
of the
v
a
riables
of
λ a
nd μ
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
Dr
i
S
ys
t
,
V
ol
.
1
1
, N
o.
3
,
Se
ptembe
r
2020
:
13
3
3
–
1
343
1336
Figure
5. TS
MFOP
ID
co
ntr
ol
le
r
with
21 me
mb
e
rsh
i
ps
The
sec
ond
ste
p,
rec
ognize
th
e
TS
-
F
uzz
y
for
mu
la
f
or
the
frac
ti
on
al
orde
rs
of
integ
ral
a
nd
de
rivati
ve
par
a
mete
rs
(λ
and
μ
)
as
s
ho
wn
i
n
Fi
gure.
6.
If
t
he
in
put
is
λ
the
blo
c
k
diag
ram
repres
ents
TS
M
F
OI
wh
il
e
if
the in
pu
t i
s
μ the
blo
c
k diag
r
am r
e
pr
e
sents
TSMFO
D.
Figure
6. Bl
oc
k
diag
ram
t
he f
racti
on
al
orde
r
s of the
integ
ra
l and de
rivati
ve
.
The
final
outp
uts
of
t
he
fu
z
zy
s
ys
te
ms
th
at
inferre
d
for
the
T
SMFO
D
or
TS
MFO
I
I
mp
le
me
nted
us
in
g
Ce
ntr
oi
d f
or
t
he def
uzzi
ficat
ion
meth
od
[21]
:
=
∑
.
∑
;
=
∑
.
∑
;
(
1
)
Wh
e
re:
λi
,µi
ϵ
{0
,
0.1,
0.2,
0.3,
0.4,
0.5,
0.6
,
0.7,
0.8
,
0.9
,
1,
1.1
,
1.2,
1.3,1.
4,
1.5,
1.6,
1.7,
1.8,
1.
9,
2}for 2
1 r
ules
Wλi is the
w
ei
g
ht
of λi,
Wµi is t
he
w
ei
gh
t
of µi,
Fλi is t
he o
utput o
f
TB
FOPI
whose λ
value i
s λi
Fµi i
s the
outp
ut of TB
FOPD
who
se
µ
value
is µi
[12].
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
P
ow Elec
& Dri S
ys
t
IS
S
N:
20
88
-
8
694
Mo
del refe
renc
e self
-
tun
i
ng fr
actional
order
PID co
ntr
ol ba
sed o
n
…
(
M.
A. A
bd
el
G
hany
)
1337
3.2.
Ha
rm
on
y se
ar
ch optimi
za
tio
n t
ec
hnique
The
c
halle
ng
e
po
i
nt
in
the
PID
an
d
FO
P
I
D
con
t
ro
ll
ers
a
re
sel
ect
ing
the
a
ppr
opria
te
par
a
mete
rs
f
or
a
certai
n
c
on
t
ro
l
le
d
pla
nt.
T
he
r
e
are
seve
ral
methods
to
fin
d
the
par
amet
e
rs
of
F
OP
I
D
c
on
t
ro
ll
er
f
or
e
xam
ple,
try
a
nd
er
ror
a
nd
Zie
gler
-
Nicho
ls
meth
od
but,
m
os
t
of
the
se
te
ch
niques
are
r
ough
road
s.
I
n
t
his
pa
pe
r,
th
e
harmo
ny
sea
rc
h
opti
mi
zat
ion
te
ch
nique
wi
ll
be
us
e
d
t
o
ob
ta
in
the
op
t
imal
value
s
of
F
OPI
D
c
on
t
ro
ll
er
par
a
mete
rs
acc
ordin
g
t
o
the
object
ive fu
nction as s
how
n
i
n (
2
)
[
22]
.
=
1
(
1
−
−
)
(
+
)
+
−
(
−
)
(
2
)
Wh
e
re
is
the
ste
ady
sta
te
e
r
ror,
is
the
ove
rsho
ot
of
s
ys
te
m
res
pons
e
,
is
the
set
tl
ing
ti
me
and
is
the
rise
ti
me.
Also
,
th
is
ob
je
ct
ive
functi
on
is
a
ble
t
o
co
mpr
om
ise
the
desi
gn
e
r
re
qu
i
reme
nts
us
i
ng
the
wei
gh
ti
ng
par
a
mete
r
valu
e
(β).
T
he
pa
ra
mete
r
is
set
la
r
ger
t
han
0.7
t
o
reduce
ov
e
r
s
hoot
a
nd
ste
a
dy
sta
te
error.
If
t
his
pa
rameter
is
adj
ust
ing
small
er
t
han
0.7
the
rise
ti
me
and
set
tl
ing
ti
me
will
be
re
du
ce
d.
Ha
rm
ony
search
(
HS)
w
as
sug
gested
by
Z
ong
W
oo
Geem
i
n
2001
[27
]
.
It
is
wel
l
known
that
HS
is
a
phe
nomen
on
-
mimi
ckin
g
al
gorith
m
in
sp
ire
d
by
the
imp
r
ov
isa
ti
on
proc
ess
of
mu
sic
ia
ns
[28
]
.
T
he
init
ia
l
popula
ti
on
of
Harmo
ny
M
e
mory
(
H
M
)
is
ch
os
e
n
rand
oml
y.
H
M
c
ons
ist
s
of
Ha
rm
ony
M
em
ory
S
olu
ti
on
(HMS)
vect
or
s
.
Table
3
s
hows
the
obta
ined
par
a
mete
rs
of
TSMFO
PID
c
on
t
ro
ll
er
base
d
on
har
m
ony
searc
h
opti
miza
ti
on
te
chn
iq
ue.
Table
3.
T
S
M
F
O
PID
pa
ramet
ers.
TSM
F
OPID
p
ara
m
ete
rs
Kp
Kd
vd
ki
vi
Para
m
eters
valu
es
9
.56
0
3
5
.35
0
6
0
.23
7
1
4
2
.59
2
6
0
.92
2
Both
of
c
onve
ntion
al
to
olbo
x
of
FOPI
D
an
d
the
Ta
kaji
-
Sugeno
(T
S)
m
od
ifie
d
FOPI
D
(
TSMFO
PID)
hav
e
the
same
res
pons
e
th
rough
t
he
sim
ul
at
ion
res
ults
a
t
differe
nt
op
e
rati
ng
co
ndit
ion
s.
I
n
a
ddit
ion,
it
is
pro
vid
e
d
in
[
21]
.
=
[
(
1
,
1
)
(
1
,
2
)
(
1
,
3
)
(
1
,
4
)
(
1
,
5
)
(
2
,
1
)
(
22
)
(
2
,
3
)
(
2
,
4
)
(
2
,
5
)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
(
,
1
)
(
,
2
)
(
,
3
)
(
,
4
)
(
,
5
)
]
(
3
)
3.3.
The
sel
f
-
tuni
n
g
TS
MFO
PID
ba
se
d
on
mo
d
el
reference
te
ch
nique
In
this
pa
per
,
the
m
odifie
d
FO
P
ID
c
on
tr
ol
par
a
mete
rs
will
be
a
dju
st
ed
on
-
li
ne
us
i
ng
the
m
od
el
ref
e
ren
ce
te
ch
nique.
The
Mo
del
Re
fe
ren
ce
Ad
a
ptive
C
ontrol
(
M
RAC)
c
on
si
ders
high
-
eff
ect
ive
ness
a
dap
ti
ve
con
t
ro
ll
er
[19].
I
t wor
ks
as a
n adaptive
ser
vo
sy
ste
m
in
w
h
ic
h
the
wan
te
d
pe
rformance is
descr
i
bed
i
n
f
orm
of
a
ref
e
ren
ce
m
od
el
.
Fig
ur
e
.
7
dem
onstrat
es
the
main
c
onst
ru
ct
io
n
of
sel
f
-
tu
ning
m
od
i
fied
F
OPID
bas
ed
on
model re
fer
e
nc
e tec
hn
i
que. T
he deta
il
s o
f
Mod
el
Re
fer
e
nce
Adap
ti
ve
S
ys
t
em d
e
rivati
on
are
giv
e
n
i
n [
29]
.
Figure
7. The
ov
e
rall
syst
em
with self
-
tu
ning T
SMFO
PID
bas
e
d on m
od
e
l refe
ren
ce
techn
i
qu
e
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
Dr
i
S
ys
t
,
V
ol
.
1
1
, N
o.
3
,
Se
ptembe
r
2020
:
13
3
3
–
1
343
1338
The
tra
nsfer
fu
nction o
f FO
PID c
on
t
ro
l ca
n be
descr
i
bed as
foll
ow
s
[
29]
.
(
)
(
)
=
+
1
+
(
4
)
=
−
(
5
)
Assume t
hat th
e p
la
nt c
an
b
e
simpli
fied
t
o
a
first
order sy
ste
m as
obvi
ou
s
in the f
ollow
i
ng e
qu
at
io
n.
(
)
(
)
=
+
1
(
6
)
Wh
e
re
are
un
known
par
a
me
t
ers.
Als
o,
ass
um
e
that
the
model
ref
e
ren
c
e
ta
kes
a
f
orm
fir
st
order s
ys
te
m a
s the
fo
ll
owin
g rela
ti
on
s
hi
p.
(
)
(
)
=
+
1
(
7
)
Wh
e
re
are
sel
ect
ed
by
desi
gner
.
Fr
om e
quat
ions
[4
-
6
]
can
con
cl
ud
e t
hat
=
+
1
(
+
1
+
)
(
−
)
(
8
)
→
=
+
1
+
+
1
−
+
1
+
+
1
(
1
+
+
.
1
+
+
1
)
=
+
.
1
+
+
1
(
+
1
+
+
.
1
+
+
1
)
=
+
.
1
+
+
1
=
+
.
1
+
+
1
+
+
.
1
+
(
9
)
=
−
(
10
)
=
[
+
.
1
+
+
1
+
+
.
1
+
−
+
1
]
(
1
1
)
=
[
+
+
+
.
1
+
1
−
(
+
.
1
+
)
(
+
+
+
.
1
+
1
)
2
]
(
1
2
)
The
(
1
2
)
ca
n b
e re
wr
it
te
n
=
[
(
+
+
.
1
+
+
1
−
−
.
1
−
)
(
+
+
.
1
+
+
1
)
2
]
(
13
)
=
[
(
+
1
)
(
+
+
.
1
+
+
1
)
2
]
(
1
4
)
=
[
(
+
1
)
(
+
+
.
1
+
+
1
)
(
+
.
1
+
)
]
(
1
5
)
Fr
om
(
4
)
a
nd (
6
)
=
[
2
(
+
+
.
1
+
+
1
)
]
(
1
6
)
To
ac
hi
eve
the
desire
d per
for
mance,
the
foll
ow
i
ng con
diti
on
m
us
t
be h
old.
+
+
.
1
+
+
1
=
+
1
(
1
7
)
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
P
ow Elec
& Dri S
ys
t
IS
S
N:
20
88
-
8
694
Mo
del refe
renc
e self
-
tun
i
ng fr
actional
order
PID co
ntr
ol ba
sed o
n
…
(
M.
A. A
bd
el
G
hany
)
1339
=
2
+
1
(
1
8
)
Fr
om t
he MI
T
ru
le
ca
n o
btain
the
fo
ll
owin
g rela
ti
on
s
hip
=
−
.
.
2
+
1
(
1
9
)
=
−
1
.
.
+
1
(
21
)
1
=
.
2
(
2
1
)
)
=
∫
+
(
0
)
(
2
2
)
Wh
e
re
(
0
)
is t
he
i
niti
al
v
al
ue of
pro
portion
al
ga
in
, By
t
he
sam
e steps.
)
=
∫
+
(
0
)
(
2
3
)
Wh
e
re
(
0
)
is t
he
i
niti
al
v
a
lue
of
pro
portion
al
ga
in
.
)
=
∫
+
(
0
)
(
2
4
)
Wh
e
re
(
0
)
is t
he
i
niti
al
v
al
ue of
der
i
vative
gain
.
=
ln
(
)
[
(
+
1
+
)
(
+
+
1
+
+
1
)
2
−
1
(
+
+
1
+
+
1
)
]
(
2
5
)
=
ln
(
)
[
+
1
+
−
−
−
1
−
−
1
(
+
+
1
+
+
1
)
2
]
(
2
6
)
=
ln
(
)
[
−
(
+
1
)
(
+
+
1
+
+
1
)
2
]
(
2
7
)
=
ln
(
)
[
−
(
+
1
)
(
+
+
1
+
+
1
)
(
+
1
+
)
]
(
2
8
)
Also
,
f
rom
(
4
) a
nd (
6
)
=
−
2
ln
(
)
[
(
+
+
1
+
+
1
)
]
(
2
9
)
=
−
2
ln
(
)
.
+
1
(
30
)
=
.
.
2
ln
(
)
.
+
1
(
31
)
=
4
.
.
+
1
(3
2
)
4
=
.
2
(
0
)
ln
(
)
(
0
)
=
2
.
(
0
)
.
ln
(
)
(
3
3
)
)
=
∫
+
(
0
)
(
3
4
)
=
[
.
.
ln
(
)
+
+
1
+
+
1
−
.
.
ln
(
)
(
+
1
+
)
(
+
+
1
+
+
1
)
2
]
=
[
.
.
ln
(
)
(
+
+
1
+
+
1
−
−
1
−
)
(
+
+
1
+
+
1
)
2
]
(
3
5
)
=
[
.
.
ln
(
)
(
+
1
)
(
+
+
1
+
+
1
)
2
]
(
3
6
)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
Dr
i
S
ys
t
,
V
ol
.
1
1
, N
o.
3
,
Se
ptembe
r
2020
:
13
3
3
–
1
343
1340
=
[
.
.
ln
(
)
(
+
1
)
(
+
+
1
+
+
1
)
(
+
1
+
)
]
(
3
7
)
Also
,
from
(
4
) a
nd (
6
)
=
[
2
.
.
ln
(
)
.
(
+
+
1
+
+
1
)
]
(
3
8
)
=
2
.
.
ln
(
)
.
+
1
(
3
9
)
=
−
.
.
2
.
.
ln
(
)
.
+
1
(
40
)
=
−
5
.
.
+
1
(4
1
)
5
=
.
2
.
(
0
)
.
(
0
)
.
ln
(
)
=
3
.
(
0
)
.
ln
(
)
(
42
)
)
=
∫
+
(
0
)
(
43
)
4.
SIMULATI
O
N RESULTS
This
sect
ion
de
monstrate
s
th
e
simulat
ion
r
esults
of
fixe
d
structu
re
TS
M
F
OPID
a
nd
sel
f
-
tu
ning
TSMF
O
PID
ba
sed
on
M
RA
S
a
pp
li
ed
to
P
SS
with
dif
fere
nt
op
e
rati
ng
po
i
nts
(h
ea
vy
and
li
ght
pa
ra
mete
rs)
thr
ough se
ver
a
l t
yp
es
of
distu
rb
a
nces.
Ca
se
1: Fi
xed s
tructu
re T
SMF
OP
I
D pe
rform
ance at
dif
fer
e
nt ope
rati
ng cond
it
io
n.
The
mec
han
ic
a
l
tor
que
Tm
a
nd
V
ref
inc
reas
es
s
udde
nly
with
ste
p
val
ue
5%
in
case
hea
vy,
li
ght
a
nd
normal
pa
ram
et
ers
values
.
The
res
ults
a
r
e
de
m
on
st
rated
i
n
Fig
ur
e
.
8.
It
is
cl
ea
r
tha
t
fix
ed
st
ru
ct
ur
e
TSMFO
PID
re
sp
onse
ca
nnot
adap
t
the
c
hanges
in
ope
rati
ng
c
onditi
ons.
S
o,
the
sel
f
-
tu
ni
ng
bec
om
es
es
sentia
l
to obtai
n hi
gh
performa
nce t
hro
ugh seve
ral
op
e
rati
ng c
ondi
ti
on
s a
nd d
ist
urba
nces.
(a)
(b)
Figure
8. The
s
ys
te
m
dyna
mic
res
pons
e
w
it
h
ste
p
c
hange
0.05
Ca
se 2
:
Co
mpa
rison
betwee
n t
he fixe
d para
mete
rs
T
SMF
OP
I
D
a
nd sel
f
-
tun
in
g f
or
TS
M
F
OPID
at li
ght l
oa
d
conditi
on.
In
case
of
li
ght
pa
rameters
va
lues,
the
mec
han
ic
al
t
orq
ue
Tm
a
nd
Vr
e
f
e
xpos
es
to
ste
p
change
5%
high.
T
he
res
ul
ts
a
re
dem
onstr
at
ed
in
Fig
ur
e
. 9
.
It
is
note
d
t
ha
t
sel
f
-
tu
ning
TSMFO
PID
(
M
RA
S)
r
esp
on
se
has
low fluct
uations, small
ov
e
rs
hoot a
nd it
r
eac
h
to
r
e
f
ere
nce
po
i
nt in smal
l t
ime
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
P
ow Elec
& Dri S
ys
t
IS
S
N:
20
88
-
8
694
Mo
del refe
renc
e self
-
tun
i
ng fr
actional
order
PID co
ntr
ol ba
sed o
n
…
(
M.
A. A
bd
el
G
hany
)
1341
(a)
(b)
Figure
9. The
s
ys
te
m
dyna
m
ic
r
es
pons
es
at li
gh
t c
onditi
on
Ca
se
3
:
Co
m
par
is
on
betwe
en
the
fi
xe
d
structu
re
TS
M
F
OPID
a
nd
sel
f
-
tu
ning
TSMFO
PID
(
M
RA
S)
performa
nce at
h
ea
vy con
diti
on.
The
pro
po
se
d
con
t
ro
ll
ers
we
r
e
inv
est
i
gated
by
co
mp
a
rin
g
the
dy
namic
re
sp
onses
of
the
PSS
at
ste
p
disturba
nce
5%
in
the
mech
anical
to
rque
∆Tm
a
nd
∆
V
r
ef.
Fig
ur
e
.
10
il
lustrate
s
the
r
esults
of
this
c
ase.
It
is
obvious
t
hat
th
e
dynamic
respon
s
e
of
sel
f
-
tu
ning
TS
M
F
OPID
(M
R
AS)
ha
s
a
good
perfor
mance
c
ompar
ed
t
o
fixe
d param
et
ers
TS
MFO
PID
contr
oller whe
re it has
small
e
r ov
e
rs
hoot a
nd s
mall
sett
li
ng ti
me
.
(a)
(b)
Figure
10. Syst
em dy
namic
re
sp
onses
w
it
h
r
efere
nce trac
ki
ng.
Ca
se
4:
T
he
performa
nce
of
t
he
fi
xe
d
st
ru
ct
ur
e
TS
MF
OP
I
D
a
nd
sel
f
-
tu
ning
TS
MFOP
ID
durin
g
the
par
a
mete
rs va
r
ia
ti
on
s a
nd h
ea
vy loa
d
c
onditi
on.
To
i
nv
e
sti
gate
the
r
obus
tne
s
s
of
pro
posed
con
t
ro
ll
ers
,
t
he
inerti
a
c
oe
ff
ic
ie
nt
inc
reased
to
become
M
=1
.5
of
nor
mal
value
an
d
the
distu
r
ban
ce
for
∆T
m
an
d
∆Vr
e
f
re
presen
te
d
by
a
0.0
5
s
te
p
cha
nge
f
rom
zer
o
to
2
s
eco
nds,
then
,
dec
reased
by
0.0
3
f
rom
2
se
co
nds
t
o
4
seco
nd
s
a
nd
finall
y
dec
rease
d
by
0.0
1
as
s
how
n
i
n
Figure.
11.
Fi
gure.
12
sho
w
s
the
syst
em
r
esp
on
ses
dr
i
ve
n
by
sel
f
-
tun
i
ng
TS
MFO
PID
(
M
RAS
)
a
nd
fixe
d
par
a
mete
rs
TS
M
F
OPID
co
nt
ro
ll
er
w
hen.
I
t
is
cl
early
s
een
that
the
sel
f
-
tu
ning
TS
M
F
OPID
(
MR
AS
)
ov
e
rc
om
es
th
ese
var
ia
ti
ons
a
nd
giv
e
good
re
spo
ns
e
wit
h
a
s
ma
ll
set
tl
ing
ti
me,
th
us
in
di
cat
ing
the
ef
fecti
ven
e
ss
of
the
sel
f
-
tun
in
g
T
S
M
F
OP
I
D
(
M
RAS
)
ove
r
a
wide
range
of
pa
r
amet
er
var
ia
ti
on
an
d
change
of
op
e
r
at
ing
c
onditi
on
s
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
Dr
i
S
ys
t
,
V
ol
.
1
1
, N
o.
3
,
Se
ptembe
r
2020
:
13
3
3
–
1
343
1342
Figure
11. Vari
abl
e d
ist
urba
nc
e f
or
∆
Tm a
nd
∆Vr
e
f
(a)
(b)
Figure
12. Syst
em dy
namic
re
sp
onses
w
it
h
r
efere
nce trac
ki
ng
5.
CONCL
US
I
O
N
A
no
vel
sche
m
e
of
sel
f
-
t
un
i
ng
Ta
kaji
-
S
ug
e
no
Mo
dified
F
racti
on
al
Order
PI
D
(T
SMFO
PI
D
)
baas
on
the
M
odel
Re
fer
e
nce
a
dap
ti
ve
s
ys
te
m
(
MR
AS
)
ap
plied
on
P
ower
S
ys
te
m
Stabil
iz
er
(
PSS
).
TS
Fu
zz
y
te
chn
iq
ue
is
use
d
t
o
c
onstr
uct
a
(TSMF
OPI
D)
.
T
he
obje
ct
ive
of
M
odel
R
efere
nce
A
da
ptive
Syst
em
(
M
RA
S
)
tun
es
t
he
five
par
a
mete
rs
of
Takaji
-
S
ug
e
no
Mod
i
fied
FOP
ID
c
ontr
oller
onli
ne.
Dif
fer
e
nt
op
e
rati
ng
po
i
nt
s
f
or
PSS
we
re
im
pl
emented
to
in
vestigat
e
t
he
r
obus
t
ness
of
pro
posed
c
on
tr
ollers.
The
ha
rm
ony
opti
miza
ti
on
te
chn
iq
ue
use
d
to
obta
in
the
op
ti
mal
pa
ram
et
ers
of
propos
ed
c
ontr
ollers.
The
simulat
io
n
res
ults
pro
vide
that
Self
-
T
unin
g
T
SM
F
OPI
D
bas
ed
on
(
M
RA
S)
ha
v
e
bette
r
pe
rformance
tha
n
the
fi
xed
par
a
mete
rs
TS
M
O
FO
P
ID
Con
tr
oller.
REFERE
NCE
S
[1]
C.
Ch
en,
“Coor
dina
t
ed
Synthesis
of
Mult
im
a
ch
ine
Pow
er
Sys
t
em
Stabilizer
U
sing
An
Eff
i
cie
nt
De
ce
n
tra
l
ized
Modal
Contro
l
(
DM
C
)
Algorit
h
m,
”
IE
EE
Tr
ans.
Powe
r S
yst.
,
vo
l
.
2
,
no
.
3
,
pp
.
54
3
–
550,
1987
.
[2]
A.
M.
A
.
Gh
any, “
Pow
er
Sys
te
m A
utom
at
i
c
Vol
tage
Regulator
De
sign Ba
sed
on
St
at
i
c
Output
Feed
bac
k
PID
Us
ing
Ite
ra
ti
ve
Li
n
ea
r
Matri
x
Ine
qu
al
i
t
y,
”
2008
12
th
In
t.
Midd
le
East
P
ower
Sy
st.
Con
f.
MEP
CON
2008
,
vol
.
1,
pp
.
441
–
446,
2008
.
[3]
M.
J.
Prabu,
P.
Poongodi,
and
K.
Premkumar,
“Rot
or
Pos
it
ion
Control
of
Brus
hle
ss
DC
Motor
using
Adapti
v
e
Neuro
Fuzzy
Inf
ere
nc
e
Sys
tem,
”
Middl
e
-
East
J
.
S
ci
.
R
es.
,
vol
.
24
,
no.
7
,
pp
.
2395
–
2403,
2016
.
[4]
Y.
Ta
ng
,
M.
Cui
,
C.
Hua
,
L
.
Li,
and
Y.
Yang,
“Opti
mum
Design
Of
Frac
ti
ona
l
Order
PI
Λd
Μ
Control
ler
for
AV
R
Sys
te
m
Us
ing
C
haot
i
c
Ant
Sw
arm
,
”
E
xpe
rt S
yst.
Appl
.
,
vo
l. 39, n
o.
8
,
pp
.
6887
–
6
896,
2012
.
Evaluation Warning : The document was created with Spire.PDF for Python.