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.
10
,
No.
3
,
June
2020
,
pp. 3
047~
3056
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v10
i
3
.
pp3047
-
30
56
3047
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
Evol
ut
ion
ary
al
go
ri
th
ms
-
ba
se
d
tu
nin
g of PID con
troll
er
fo
r
an
AVR sys
tem
Petchin
athan
Govind
an
Depa
rtment
o
f
E
le
c
tri
c
al a
nd
Co
m
pute
r
Engi
n
ee
r
ing,
Co
ll
eg
e
of
Engi
ne
eri
ng
D
eb
re
Berh
an
Univ
e
rsit
y
,
Et
hiop
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ve
d
Ma
r
14
, 201
9
Re
vised
N
ov
16
,
20
19
Accepte
d
Dec
1
, 2
0
19
In
thi
s
pape
r
,
a
n
evol
u
ti
ona
r
y
al
gorit
hm
-
base
d
opti
m
izati
on
a
lgori
thm
i
s
proposed
with
new
obj
ec
t
ive
func
ti
on
to
des
ign
a
PID
con
trol
ler
for
the
aut
om
atic
vo
lt
ag
e
r
egulator
(
AV
R)
s
y
stem.
T
he
n
ew
ob
j
ective
func
ti
on
is
proposed
to
improve
th
e
tra
nsi
en
t
response
of
the
AV
R
cont
ro
l
s
ystem
and
to
obta
in
the
opt
i
m
al
val
u
es
of
c
ontrol
ler
ga
in.
I
n
thi
s
p
ape
r
,
p
a
rti
cle
sw
arm
opti
m
iz
ation
(PS
O)
and
cuc
koo
sea
rch
(CS)
al
gor
it
hm
s
are
proposed
to
tun
e
the
par
ame
te
rs
o
f
a
PID
cont
ro
lle
r
for
th
e
con
trol
of
AV
R
s
y
stem.
Sim
ula
ti
on
result
s
ar
e
c
apable
and
illus
tra
t
e
the
eff
ec
t
ive
n
ess
of
the
propo
sed
m
et
hod
.
Num
eri
ca
l
and
s
imulat
ion
resul
ts
base
d
on
the
pr
oposed
tun
ing
a
pproa
ch
on
PID
cont
rol
of
an
AV
R
s
y
ste
m
for
servo
and
reg
ulator
y
co
ntrol
show
the
excel
l
ent
p
er
form
anc
e
of
PS
O a
nd
CS
opt
imiza
t
ion
a
lgori
thm
s.
Ke
yw
or
d
s
:
AV
R
syst
em
Cuck
oo sea
rch
Ev
olu
ti
onary
a
lgorit
hm
s
Objecti
ve f
un
c
ti
on
PI
D
contr
oller
PSO
Copyright
©
202
0
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Petc
hin
at
ha
n
G
ov
i
nd
a
n
,
Dep
a
rtm
ent o
f El
ect
rical
an
d
Com
pu
te
r
E
ng
i
neer
i
ng,
Coll
ege
of
En
gi
neer
in
g De
br
e
Berh
a
n
U
niv
e
rsity
,
Debre Be
rh
a
n,
Ethio
pia
.
Em
a
il
:
gp
et
chi@g
m
ai
l.co
m
1.
INTROD
U
CTION
Au
t
om
atic
volt
age
re
gu
la
t
or
pl
ay
s
a
c
ru
ci
al
r
ole
i
n
powe
r
s
yst
e
m
so
a
s
t
o
regulat
e
t
he
ou
tpu
t
volt
age
at
a
nom
inall
y
const
ant
desire
d
vo
lt
age
le
vel.
In
po
wer
ge
ne
rator,
the
f
unct
ion
of
A
VR
is
t
o
e
ns
ure
t
he
volt
age
from
the
po
we
r
ge
ner
at
or
s
t
o
be
r
unning
s
m
oo
thly
and
a
ls
o
to
m
ai
ntain
the
sta
bili
ty
of
the
volt
ag
e
from
the
ge
ner
at
or
s
.
The st
abili
ty
of
the
A
VR
co
ntr
ol
syst
e
m
is
an im
po
rtant
iss
ue
since
it
can
c
riti
cal
ly
i
m
pin
ge o
n
the
secu
rity
of
the
powe
r
syst
e
m
.
The
excit
a
ti
on
syst
em
ou
gh
t
to
res
pons
i
ble
for
the
e
ff
e
ct
ive
vo
lt
age
c
on
t
ro
l
and
im
pr
ovem
ent
of
t
he
syst
e
m
sta
bili
ty
[1]
.
T
he
e
xcita
ti
on
syst
em
no
t
on
ly
co
ntr
ols
t
he
outp
ut
volt
age
of
the
ge
ne
rator
a
nd
al
so
c
ontr
ols
the
powe
r
fa
ct
or
a
nd
m
agn
it
ud
e
of
the
c
ur
ren
t
.
I
n
m
os
t
of
the
e
xcite
r
sy
stem
a
thyrist
or
-
bas
ed
syst
em
is
e
m
plo
ye
d
to
pro
vid
e
a
co
nt
ro
ll
ed
outp
ut
vo
lt
age
t
o
t
he
e
xcite
r.
I
n
m
os
t
of
the
in
du
st
rial
app
li
cat
io
ns
th
e
pro
portio
nal
-
integ
ral
-
der
i
va
ti
ve
(PID
)
c
ontr
oller
has
be
en
c
omm
on
ly
us
e
d
because
of
it
s
si
m
ple
config
urat
ion,
tr
ouble
-
fr
ee
im
ple
m
entat
ion
an
d
go
od
perform
ance
in
a
la
rg
e
ra
nge
of
op
e
rati
ng
c
ondi
ti
on
s.
Ne
ve
rth
el
ess,
e
ff
ect
ive
an
d
s
uitable
t
un
i
ng
of
the
P
I
D
c
ontr
oller’s
par
am
et
ers
ha
s
bee
n
relat
ively
diff
i
cult
beca
us
e
m
any
in
du
st
rial
processes
are
f
reque
ntly
aff
ec
te
d
by
pro
blem
s
su
c
h
as
highe
r
ord
er,
tim
e d
el
ay
s an
d nonli
nea
riti
es [2
-
4].
The
t
un
i
ng
of
PI
D
co
ntr
oller
ha
d
been
do
ne
thr
ough
c
onven
ti
onal
m
et
ho
ds
ove
r
the
pa
st
deca
des.
The
c
onve
ntio
nal
m
et
ho
d
f
or
tu
ning
of
P
I
D
c
on
t
ro
ll
ers
su
c
h
as
Zie
gle
r
-
Nich
ols
(
Z
-
N)
an
d
C
ohen
-
Co
on
te
chn
iq
ues
a
re
pro
d
uci
ng
on
ly
sta
ble
tu
ned
pa
ram
et
ers
with
s
om
e
os
ci
ll
at
i
on
a
nd
ov
e
rs
hoot
outp
ut
res
ponse
.
In
or
der
to
av
oid
the
sho
rtcom
ing
s
of
the
co
nventio
nal
tun
in
g
m
et
ho
ds,
s
of
t
com
pu
t
ing
te
ch
niques
li
ke
Ar
ti
fici
al
Ne
ural
Networ
k
an
d
Fu
zzy
lo
gic
ap
proac
hes
ha
ve
been
pro
posed
in
the
li
te
ratur
e
[5,
6].
E
voluti
on
a
ry
al
gorithm
s
-
based
ap
proac
hes
are
al
so
propos
ed
to
tu
ne
the
par
am
et
ers
of
PI
D
c
ontr
oller
in
m
any
app
li
cat
ions
in
li
te
ratur
e.
P
et
chinatha
n
et
al
.
has
pro
pos
ed
a
c
om
bin
at
ion
of
PS
O
a
nd
bacteria
l
f
oragin
g
al
go
rith
m
for
op
ti
m
al
tun
in
g
of
t
he
P
I
con
t
ro
ll
er
[
7].
Re
centl
y,
Mult
idynam
ic
s
Algorithm
for
Global
Op
ti
m
iz
at
ion
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
30
47
-
3056
3048
(MA
GO)
[
8],
An
t
Col
on
y
O
pti
m
iz
at
ion
(
A
CO)
[
9],
Hyb
r
id
Bi
oge
ogra
p
hy
base
d
O
ptim
iz
at
ion
(BB
O)
[
10
]
,
Si
m
ultaneo
us
Pertu
rb
at
io
n
Stoch
a
sti
c
App
roxim
a
ti
on
[
11
]
,
Africa
n
buf
falo
op
ti
m
iz
a
ti
on
al
gorithm
(A
BO)
[
12
]
Invasive
wee
d
op
ti
m
iz
ation
(
I
WO)
[
13,
14]
and
PS
O al
gorit
hm
[1
5]
ha
s
pr
opos
e
d
for t
uni
ng
of
P
ID
c
ontrolle
r
par
am
et
er in
va
rio
us
c
on
t
ro
l
app
li
cat
io
ns
.
In
recent
li
te
ratur
e
m
any
evo
l
ution
a
ry
opti
m
iz
at
ion
al
gorith
m
s
are
propose
d
f
or
t
un
i
ng
PID
co
ntr
oller
in
the
A
VR
syst
e
m
su
ch
as
A
na
rch
ic
So
ci
et
y
Op
ti
m
iz
ation
[
16
]
,
rei
nfor
ce
m
ent
le
arn
i
ng
autom
at
a
op
ti
m
iz
at
ion
appr
oach
[
17
]
,
real
co
de
d
G
A
with
f
uzzy
log
ic
te
c
hn
i
qu
e
[18],
Ch
oatic
ant
swa
rm
al
go
rithm
[19],
A
rt
ific
ia
l
Be
e
Colo
ny
al
gorithm
[20],
Hybr
i
d
G
A
-
Ba
ct
erial
Fora
ging
(BF
)
al
gorith
m
[2
1]
a
nd
l
oc
al
unim
od
al
sa
m
pl
ing
al
gorithm
[2
2].
G
A
an
d
A
nt
Colo
ny
O
ptim
i
zat
ion
te
c
hniq
ues
are
pr
opose
d
t
o
t
un
e
the
par
am
et
ers
of
FO
P
I
D
con
t
ro
ll
er
in
c
on
t
ro
ll
in
g
of
A
VR
syst
em
.
In
so
m
e
of
the
re
search
pap
e
rs
novel
perf
or
m
ance
c
rite
ria
ha
s
bee
n
pro
po
se
d
f
or
op
ti
m
al
tun
ing
of
P
ID
an
d
F
OP
I
D
c
ontr
oller
in
A
VR
c
ontr
ol
syst
em
.
A
novel
perform
ance
crit
erion
c
om
pr
ise
s
of
ov
e
rs
hoot,
set
tl
ing
ti
m
e,
ste
ady
sta
te
er
ror
an
d
m
e
an
of
ti
m
e
weigh
te
d
i
nteg
ral
abs
olu
te
error
has
bee
n
pr
opos
e
d
f
or
optim
al
tu
ning
of
P
ID
con
t
ro
ll
er
i
n
AV
R
syst
em
us
in
g
c
ucko
o
searc
h
al
gorithm
[2
3].
A.
Sika
nder
e
t.
al
,
20
18
has
pro
po
se
d
a
cuc
koo
sea
rc
h
al
gorithm
based
f
racti
on
al
orde
r
PID
con
t
ro
ll
er
f
or
AV
R
syst
em
with
perf
or
m
ance
c
rite
rio
n
wh
ic
h
was
pro
po
s
e
d
by
Gai
ng
et
.
al
in
2004
[24].
In
t
his
resea
rch
wor
k,
Cuc
koo
search
(CS
)
a
nd
p
a
rtic
le
swa
rm
op
tim
iz
a
ti
on
(
PSO)
al
go
rithm
s
are
propos
ed
to
fin
d
the
op
ti
m
a
l
par
am
et
ers
of
PI
D
c
ontr
oller
in
the
co
ntr
ol
of
a
uto
m
at
ic
vo
lt
age
re
gula
to
r
(
AV
R
)
syst
em
with
new
pe
rfo
rm
a
nce
c
rite
rio
n
c
om
pr
ise
s
of
Int
egr
al
a
bs
ol
ute
erro
r,
rise
ti
m
e,
set
tl
ing
ti
m
e
an
d
peak
ove
r
sh
oot.
The
pe
rfo
rm
a
nce
of
this
ne
w
pro
pose
d
perform
ance
c
rite
rion
is
c
om
par
ed
with
perform
ance
of
oth
e
r
perform
ance
cr
it
er
ion
s
uc
h
as
ITAE,
ITSE
,
I
SE,
MSE
a
nd
I
AE.
The
pa
per
is
m
ai
nly
or
ga
ni
zed
su
c
h
t
hat
s
ect
ion
two
de
scribe
s
about
t
he
A
ut
om
atic
Vo
lt
a
ge
Re
gula
to
r
(
AV
R
)
syst
em
;
sect
io
n
t
hr
ee
exam
ines
the
Cuck
oo
search
(CS)
al
gorithm
and
pa
rt
ic
le
swar
m
op
ti
m
iz
at
ion
(P
S
O
)
al
gori
thm
s;
s
ect
ion
f
our
a
nd
five
co
nce
ntra
te
on
the
a
pp
li
cat
ion
of
CS
-
PID,
PS
O
-
P
I
D
a
nd
c
onven
ti
onal
t
un
i
ng
m
et
ho
d
(Zie
gl
er
-
Nic
ho
ls
)
i
n
op
ti
m
al
tun
ing
PI
D
con
t
ro
ll
er
f
or
both
se
r
vo
an
d
r
egu
la
to
ry
c
on
tr
ol
of
AV
R
syst
e
m
.
Additi
on
al
ly
,
sect
ion
si
x
de
scribes
c
oncl
us
ions
of the st
ud
y.
2.
AU
TO
M
ATI
C VOLT
A
GE
R
EG
ULAT
O
R
The
A
uto
m
at
ic
V
oltage
Re
gul
at
or
(
AV
R
)
is
a
ve
ry
im
po
rta
nt
m
od
ule
to
m
ain
ta
in
the
te
rm
i
nal
vo
lt
a
ge
of
a
ny
power
gen
e
rato
rs
sinc
e
it
ad
justs
t
he
excit
er
vo
lt
ag
e
of
the
po
wer
ge
ne
rators.
T
he
A
VR
syst
e
m
is
to
c
on
ti
nu
ously
observe the te
r
m
inal vo
lt
age
of
powe
r
ge
nerat
or
under
va
riou
s l
oad
i
ng cond
it
io
ns
at al
l t
i
m
es b
y
ens
ur
in
g
that
t
he
ge
ner
at
or
'
s
vo
lt
age
operat
es
within
the
predeterm
ined
lim
i
ts.
The
A
V
R
syst
e
m
con
sist
s
of
four
m
ai
n
par
t
s,
nam
el
y
a
m
p
li
fier,
excit
er,
gen
e
rato
r
a
nd
sens
or
.
The
re
al
m
od
el
of
A
VR
syst
em
[
2
0
]
is
il
lustrate
d
in
Fi
gure
1
.
In
or
de
r
to
m
od
el
the
f
our
af
or
e
sai
d
c
om
po
ne
nts
an
d
determ
ine
their
trans
fer
f
un
ct
ion
s
,
each
com
pone
nt
m
us
t
be
li
near
iz
ed
by
ignori
ng
t
he
sat
ur
at
ion
an
d
oth
e
r
no
nlinear
it
ie
s
and
al
s
o
c
on
s
ideri
ng
the
m
ajo
r
ti
m
e
co
ns
ta
nt.
Th
e
est
i
m
at
ed
transf
e
r
functi
ons
of
t
hese
c
om
po
ne
nts
m
a
y
be
re
present
ed
by
m
at
he
m
at
ic
a
l as f
ollow
s
[1
2
]
:
The
tra
nsfer
fu
nction m
od
el
of the
am
plifie
r
is represe
nted by
s
G
s
V
s
V
a
a
e
R
1
)
(
)
(
(
1)
wh
e
re
a
G
denotes
the
gain
of
the
a
m
plifie
r,
a
den
otes
t
he
tim
e
const
ant
of
a
n
a
m
plifie
r
m
od
el
,
e
V
and
R
V
represe
nts
the
error
volt
age
a
nd
am
plifie
d
volt
age
res
pecti
vely
.
Th
e
sta
ndard
values
of
a
G
are
bet
ween
10
and
400.
T
he
ti
m
e
const
ant
a
of an
a
m
plifie
r
ra
ng
e
from
0
.02 t
o 0
.1
s.
The
tra
nsfer
fu
nction m
od
el
of a
n
e
xcite
r
ca
n be
re
pr
ese
nted
by a
gain
e
G
and
tim
e co
ns
ta
nt
e
s
G
s
V
s
V
e
e
R
F
1
)
(
)
(
(2)
wh
e
re
F
V
denotes
the
fiel
d
vo
lt
a
ge.
T
he
sta
nd
a
r
d
val
ues
of
e
G
are
betwee
n
10
a
nd
40
0.
T
he
ti
m
e
const
ant
e
of an
am
plifie
r
ra
ng
es
f
ro
m
0
.
5
to
1.0 s.
Si
m
il
arly
,
the
li
near
iz
ed
tra
nsfer
f
unct
ion
m
od
el
of
the
ge
ne
rator
c
ou
l
d
be
re
pr
e
sente
d
by
a
gain
g
G
and tim
e con
st
ant
g
.
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
Evolutio
nary
al
go
rit
hms
-
base
d
tu
ning
of P
I
D
c
on
tr
oller f
or a
n
A
VR syste
m
(
Pet
c
hinat
han Go
vi
ndan
)
3049
s
G
s
V
s
V
g
g
F
T
1
)
(
)
(
(3)
wh
e
re
T
V
denotes
th
e
ge
ner
at
or
te
rm
inal
vo
lt
age.
T
he
c
onsta
nts
g
G
and
g
are
load
de
pende
nt,
g
G
coul
d
var
y
from
0.
7
to
1.0
an
d
g
bet
ween
10
a
nd
400.
T
he
ti
m
e
c
on
sta
nt
e
of
a
n
a
m
pl
ifie
r
ra
ng
e
s
from
1.
0
t
o
2.0 s
.
(fu
ll
l
oad to
no loa
d).
Finall
y,
the
tra
ns
fe
r
f
unct
ion
m
od
el
of
the
s
ens
or
ca
n
be
r
epr
ese
nted
by
a
si
m
ple
first
or
de
r
tran
sfe
r
functi
on, give
n by,
s
G
s
V
s
V
s
s
T
S
1
)
(
)
(
(4)
wh
e
re
S
V
de
not
es
the
se
nsor
out
pu
t
volt
age.
Tim
e
con
sta
nt
s
cou
ld
be
ve
ry
sm
al
l,
gen
e
rall
y
be
tween
0.
00
1
and
0.0
6
s
.
T
he
value
s
of
gai
n
a
nd
tim
e
const
ant
f
or
va
rio
us
c
om
po
ne
nts
in
AV
R
m
od
e
l
are
li
ste
d
in
T
able
1.
Ma
ke use
of th
e ab
ov
e
m
od
el
s,
the
AVR
bloc
k diag
ram
is con
t
ro
ll
ed
b
y
PID c
on
t
ro
ll
er is
sh
ow
n
in
Fi
gur
e 2
.
Figure
1.
Mo
de
l of A
VR syst
e
m
[
2
0
]
Table
1.
Gain
and tim
e con
st
ant of c
om
po
ne
nts
of
A
VR m
od
el
Co
m
p
o
n
en
ts
Gain
Ti
m
e
con
stan
t
A
m
p
lif
ier
10
a
G
1
.
0
a
Exciter
1
e
G
4
.
0
e
Gen
erator
1
g
G
1
g
Sen
so
r
1
s
G
01
.
0
s
Figure
2.
Bl
oc
k diag
ram
o
f
th
e AVR sy
ste
m
con
t
ro
ll
ed
b
y
P
ID
co
ntr
oller
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
30
47
-
3056
3050
3.
EVOLUTI
ONAR
Y
ALGO
R
ITHMS
Ma
ny
ev
olu
ti
onary
al
gorit
hms
are
re
ported
in
li
te
ratur
e
for
t
un
i
ng
of
PID
c
on
t
ro
ll
er
par
am
et
er
in
m
any
app
li
cat
io
ns
.
I
n
this
pap
e
r,
par
t
ic
le
swar
m
op
ti
m
iz
at
ion
(P
SO)
an
d
cuc
koo
se
arch
(CS)
al
go
r
it
h
m
s
are
propose
d
for op
ti
m
al
tun
ing
of P
ID co
nt
r
oller in
AVR
syst
e
m
. Th
e
ba
sic
s of these
pr
opos
e
d
al
gorithm
s ar
e as foll
ow
s
:
3.1.
P
art
ic
le
s
w
arm
op
timi
z
at
i
on
(PSO)
al
go
ri
th
m
PSO
al
gorithm
was
de
velo
pe
d
by
Mr.
Jam
es
Ke
nn
e
dy
a
nd
Russell
C.
Eberha
rt
in
19
95
[
2
5
,
2
6
]
.
PSO
is
im
ple
mented
base
d
on
the
bi
ol
ogic
al
be
hav
i
our
of
som
e
ani
m
al
s
to
chase
t
he
al
ive
hab
it
s
t
hroug
h
swar
m
intel
li
gen
ce.
S
om
e
of
the
at
tr
act
ive
featu
res
of
PSO
are eas
y
to
i
m
ple
m
ent
an
d
al
so
no g
r
adient
in
form
a
ti
on
is
require
d.
It
ca
n
be
us
e
d
to
so
l
ve
a
wi
de
ar
ray
of
dif
fer
e
nt
op
t
i
m
iz
ation
pr
ob
l
e
m
s.
It
has
the
popula
ti
on
of
s
war
m
s
that
is
al
lowed
to
m
ov
e
in
th
e
search
s
pace
accor
ding
to
a
fr
am
ed
form
ula.
The
m
ov
e
m
ent
of
the
s
war
m
popula
ti
on
res
ults
in
the
best
-
kn
own
posit
io
n
of
th
e
s
war
m
’s
popula
ti
on.
The
proce
ss
is
rep
eat
e
d
ti
ll
th
e
be
st
s
at
isfact
or
y
s
ol
ution
is
obta
ine
d.
T
he
c
hoic
e
a
nd
sel
ect
ion
of
PSO
pa
ram
et
er
s
ha
ve
a
la
r
ge
i
m
pact
on
opti
m
iz
ing
perform
ance an
d y
ie
lds the
best
r
es
ult.
The
PS
O
al
gor
it
h
m
con
sist
s
of
a
colle
ct
ion
of
pa
rtic
le
s
tha
t
m
ov
e
in
t
he
r
egio
n
of
the
se
arch
s
pace
influ
e
nce
d
by
their
ow
n
best
p
ast
l
ocati
on
a
nd
th
e b
est
pas
t
locat
io
n
of
t
he
w
hole
s
wa
rm
or
a
cl
os
e
n
ei
ghbo
ur.
Each ite
rati
on
a p
a
rtic
le
’s
vel
ocity
is upd
at
e
d usin
g:
))
)
(
(
*
)
(
*
(
)
)
)
(
(
*
)
(
*
(
)
(
)
1
(
2
1
t
P
P
r
a
n
d
C
t
P
P
r
a
n
d
C
t
V
t
V
i
g
b
e
s
t
i
b
e
s
t
i
i
i
(5)
wh
e
re
,
)
1
(
t
V
i
is
the
new
velocit
y
of
t
he
i
th
par
ti
cl
e,
C
1
an
d
C
2
are
the
weig
ht
fa
ct
or
f
or
t
he
loc
al
best
a
nd
global
best
pos
it
ion
s
res
pecti
ve
ly
.
P
i
(
t
)
is
the
i
th
par
ti
cl
e’s
po
sit
ion
at
tim
e
.
b
e
s
t
i
P
is
the
i
th
par
ti
cl
e’s
best
-
kn
own
po
sit
io
n
an
d
g
b
e
s
t
P
is
the
be
st
posit
ion
well
-
kn
own
to
the
s
war
m
.
T
he
r
and(
)
functi
on
ge
ner
at
es
a
unif
or
m
rand
om
value
betwee
n
0
a
nd
1.
T
he
vari
ants
in
the
(
5
)
co
ns
ide
r
bes
t
po
sit
ions
wit
hin
a
pa
rtic
le
s
local
neig
hbour
hood
in
ti
m
e. A
p
a
rtic
le
p
os
it
io
n
is
updated
usi
ng:
)
(
)
(
)
1
(
t
V
t
P
t
P
i
i
i
(6)
3.2.
C
uck
oo
s
earch
(
CS
)
al
gori
th
m
The
CS
al
gorit
hm
was
de
velo
ped
by
Mr.
Xi
n
–
Shan
g
a
nd
S
us
a
h
Deb
in
20
09.
It
is
im
ple
m
ented
base
d
on
t
he
uniq
ue
be
hav
i
our
of
the
bird
c
uc
koo.
T
he
init
ia
l
popu
l
at
ion
ta
ke
n
is
the
num
ber
of
c
ucko
os
an
d
it
s
eggs.
Cuck
oo
wil
l
search
a
nd
la
ys
it
s
egg
s
in
th
e
nests
of
ot
he
r
host
sp
eci
es
.
It
will
seek
f
or
the
best
ne
s
t
fr
om
the
acce
s
sible
nests.
It
reli
es
on
th
ree
at
ti
tude
s
nam
ely
it
la
ys
one
eg
g
at
a
tim
e,
and
the
ne
st
with
best
e
gg
s
ca
n
be
a
gitat
ed
ove
r
ne
xt
bear
in
g
for
ha
tc
h
in
g,
t
he
acce
ssi
ble
a
m
ou
nt
of
host
nests
is
fi
xed.
Ba
sed
on
t
his
cucko
o
hatchin
g
a
ddre
ss
the
optim
iz
e
d
so
l
ution
is
ac
qu
i
red
f
or
t
he
pro
blem
.
The
best
e
xe
rcise
a
m
ou
nt
is
cal
le
d
from
the
trave
rse
d
s
olu
ti
ons.
T
he
re
fore,
a
sit
uatio
n,
in
w
hich
t
he
gr
eat
est
num
b
ers
of
e
gg
s
a
re
s
aved,
is
the
para
m
et
er
wh
ic
h
the
c
uc
koo
sea
rch
int
ends
to
op
ti
m
i
ze
it
.
Fo
r
sim
plici
ty
in
descr
ibin
g
a
C
uckoo
Searc
h
al
gorithm
,
the foll
owin
g
t
hr
ee
ideali
zed
ru
le
s a
re
us
e
d [
2
7
]:
Each c
uc
koo
la
y
on
e
egg at a
t
i
m
e, an
d d
um
p
it
s egg in
the
r
andom
ly
ch
os
e
n nest;
The best
nests
with
high
qu
al
i
ty
o
f
e
ggs
will
carr
y
ov
e
r
t
o
th
e n
e
xt g
e
ne
rati
on
s;
The
num
ber
of
avail
able
ho
st
nests
is
fixe
d,
a
nd
the
eg
g
la
id
by
a
c
uc
koo
is
disco
ver
e
d
by
t
he
host
bi
rd
wi
th
a pro
ba
bili
ty
P
a
∈
[
0,
1].
Ba
sed
on
t
hes
e
th
ree
ru
le
s
,
t
he
basic
ste
ps
of
the
Cuc
koo
Searc
h
(CS
)
al
gorithm
s
ca
n
be
s
umm
ari
zed
a
s
the pseu
doco
de
shown i
n
Fig
ur
e
3
[2
7
].
W
he
n
gen
e
rati
ng
ne
w
so
l
utio
ns
X(t+
1) for
, sa
y, a cuc
koo
i
, a
Lévy flig
ht
is perfo
rm
ed.
Lév
y fli
ght i
s
on
e
of t
he wel
l
-
kn
own
fligh
ts
beh
a
viou
r of
m
any anim
al
s an
d
in
sect
s.
)
L
é
v
y
(
)
(
)
1
(
t
X
t
X
i
i
(7)
wh
e
re
,
α
>
0
is
the
ste
p
siz
e
w
hich
m
us
t
be
re
la
te
d
to
t
he
sca
le
s
of
the
pro
bl
e
m
of
i
nterests.
Ma
inly
,
we
ca
n
us
e
α
=
1.
I
n
t
he
(
7
)
is
basical
ly
a
r
andom
walk
s
t
och
a
sti
c
eq
uation.
In
m
os
t
cas
es,
a
ra
ndom
w
al
k
is
a
m
ark
ov
chai
n
whose
ne
xt
loc
at
ion
only
de
pe
nd
s
on
the
c
urr
ent
locat
io
n
)
(
t
X
i
and
the
tra
ns
it
io
n
pro
bab
il
it
y
(th
e
seco
nd
te
rm
in
the
(
7
)
).
The
pro
duct
⊕
m
eans
e
ntry
wise
m
ul
ti
plica
ti
on
s.
T
he
rand
om
walk
via
Lé
vy
fligh
t
is
m
or
e
e
ff
ic
ie
nt
in
e
xp
l
or
i
ng
th
e
searc
h
sp
ace
since
it
s
ste
p
l
eng
t
h
is
m
uch
longer
in
the
long
r
un.
Lévy
fligh
t
will
m
ake
sur
e
the
syst
e
m
will
no
t
be
trap
pe
d
in
a
l
ocal
op
tim
u
m
.
CS
al
go
rithm
is
a
popu
la
ti
on
-
base
d
al
gorithm
and
it
is
si
m
il
ar
to
GA
and
PSO,
but
i
n
C
S
al
gorith
m
the
num
ber
of
par
am
et
ers
to
be
tu
ne
d
is
l
ess
th
an
GA
a
nd
P
S
O
,
and th
us
it
is
pro
ba
bly m
or
e comm
on
to
ad
a
pt to
a
w
i
der cl
ass of
op
ti
m
iz
a
ti
on
pro
blem
s.
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
Evolutio
nary
al
go
rit
hms
-
base
d
tu
ning
of P
I
D
c
on
tr
oller f
or a
n
A
VR syste
m
(
Pet
c
hinat
han Go
vi
ndan
)
3051
Figure
3.
Pse
udoc
ode
of cuc
koo sea
rc
h (CS
)
algorit
hm
4.
FOR
M
ULAT
ION
OF
OBJ
ECTIVE F
U
N
CTIO
N
In
this
pa
pe
r,
var
i
ou
s
obj
ect
ive
functi
ons
a
re
im
ple
m
ente
d
t
o
fin
d
t
he
optim
al
par
am
e
te
rs
of
P
I
D
con
t
ro
ll
er
us
in
g
pro
pose
d
ev
olu
ti
onary
al
gorithm
s
and
al
so
pe
rfo
rm
ance
of
var
i
ou
s
obj
ect
iv
e
f
un
ct
i
on
s
a
re
com
par
ed. Th
e
va
rio
us
im
plem
ented
ob
j
ect
ive
functi
ons a
r
e:
Me
an
S
quare
Error
:
dt
t
e
t
M
S
E
0
2
))
(
(
1
(8)
In
te
gr
al
S
quare
Erro
r:
dt
t
e
I
S
E
0
2
)
(
(9)
In
te
gr
al
s Tim
e Square
Erro
r:
dt
t
e
t
I
T
S
E
0
2
))
(
(
(10)
In
te
gr
al
Tim
e w
ei
ghte
d A
bso
lute Er
ror:
dt
t
e
t
I
T
A
E
0
)
(
(11)
In
te
gr
al
Absol
ute Err
or
:
dt
t
e
I
A
E
0
)
(
(12)
wh
e
re
e(t)
is
t
he
er
ror
sig
nal
i
n
ti
m
e
do
m
ai
n.
I
n
proce
ss
of
t
un
i
ng
of
P
ID
c
on
t
ro
ll
er,
t
he
c
on
t
ro
ll
er
pa
ra
m
et
ers
are
a
dju
ste
d
t
o
m
ini
m
iz
e
t
he
e
rror
si
gnal
or
to
m
ini
m
iz
e
the
val
ue
of
a
bove
m
entioned
obje
ct
ive
functi
ons
(
8
-
12
)
.
I
n
order
to
i
m
pr
ove
the
ti
m
e
respon
se
a
naly
sis
by
re
du
ci
ng
rise
ti
m
e,
ov
e
rs
hoot
a
nd
set
tl
ing
tim
e
and
t
o
get
bette
r
res
pons
e
of
t
he
c
ontr
oller,
t
he
c
om
bin
e
d
obj
ect
ive
fun
ct
ion
has
be
en
form
ulate
d
by
us
in
g
IA
E
,
ti
m
e
do
m
ai
n
sp
e
ci
ficat
ion
s
s
uch
as
ris
e
tim
e,
set
tl
ing
tim
e
and
peak
overs
hoot
a
nd
the
weig
ht
fa
ct
or
s.
The
wei
gh
i
ng
f
act
or
s
are
be
nt
by
an
age
ncy
of
assum
ing
it
er
at
ion
s
f
or
al
te
r
ed
am
ou
nt
of
weig
hts
an
d
ac
cepti
ng
the
weig
ht
ag
e
ncy
with
bette
r
pe
rfor
m
ance.
The
perf
orm
ance
of
pro
po
s
ed
c
om
bin
ed
ob
j
ect
ive
f
un
ct
i
on
(
in
(
13
)
)
ha
s
be
en
c
om
par
ed
with
oth
e
r
e
xis
ti
ng
obj
ect
ive
functi
ons.
The
wei
gh
ti
ng
fact
or
s
f
or
this
res
earch
work
a
re
c
ons
idere
d
as
w
1
=
40
a
nd
w
2
=2
0
us
i
ng
tria
l
a
nd
e
rro
r
proce
dure
.
T
he
f
or
m
ulate
d
e
quat
io
n
f
or
the co
m
bin
e
d objecti
ve
fun
ct
ion
is
as
fo
ll
ow
s:
)
(
2
1
I
A
E
w
ov
e
r
s
ho
ot
P
e
ak
t
i
m
e
Se
t
t
l
i
ng
t
i
m
e
R
i
s
e
w
F
(13)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
30
47
-
3056
30
52
5.
IMPLEME
N
TATION
OF
TUNING
OF
PID CO
NTR
OLL
ERS
--
SE
TPOINT T
R
ACKIN
G
In
t
his
pa
per,
tun
i
ng
of
PID
Con
tr
ollers
has
been
im
ple
m
e
nted
a
nd
sim
ul
at
ed
us
i
ng
Zie
gler
-
Nich
ol
s
(Z
-
N), C
ucko
o Searc
h
a
nd Pa
rtic
le
Sw
arm
Optim
iz
at
ion
alg
or
it
hm
f
or the
con
t
ro
l
of AV
R sy
stem
.
5.1.
Z
ie
gler
-
nichols
(Z
-
N
) me
thod
of
t
un
ing
The
Z
-
N
m
et
ho
d
is
a
c
onve
nt
ion
al
cl
os
e
d
l
oop
m
et
ho
d
f
or
tun
in
g
of
PID
con
t
ro
ll
er.
T
hi
s
te
ch
n
iq
ue
is
al
so
cal
le
d
as
an
u
lt
im
a
te
c
yc
li
ng
m
et
ho
d
wh
ic
h
is
bas
ed
on
ad
justi
ng
a
cl
os
e
d
lo
op
ti
m
e
resp
ons
e
unti
l
su
sta
ine
d
os
ci
ll
at
ion
s
occur.
The
n
c
on
tr
olle
r
set
ti
ngs
are
c
om
pu
te
d
base
d
on
the
in
form
at
ion
from
the
cl
os
ed
loop
res
pons
e.
The
pe
rfor
m
ance
of
A
VR
sys
tem
has
been
a
naly
zed
by
i
m
plem
ent
ing
Z
-
N
tu
ni
ng
m
et
h
od
f
or
tun
in
g
the
par
a
m
et
ers
of
PID
con
t
ro
ll
er.
T
he
trans
fer
f
unct
io
n
m
od
el
of
am
plif
ie
r,
e
xcite
r,
gen
e
rato
r
an
d
s
ens
or
has bee
n
de
riv
ed
.
By
us
in
g
t
hi
s d
eri
ved tra
nsfer
functi
on,
th
e Ma
tl
ab
-
Sim
ulin
k
m
od
el
of
t
he
cl
ose
d l
oop
AV
R
syst
e
m
with
P
ID
co
ntr
oller
has
bee
n
dev
e
lop
e
d
as
s
how
n
in
the
Fig
ure
4
.
T
he
c
ontr
oller
gain
val
ue
s
an
d
the
de
rive
d
ti
m
e
respo
ns
e
s
pe
ci
ficat
ion
s
are
giv
e
n
in
T
a
ble
2
.
T
he
respo
nse
ob
ta
i
ned
us
in
g
the
Z
-
N
m
eth
od
is
represe
nted
i
n t
he
Fig
ure
5
.
Figure
4.
Ma
tl
ab
–
Sim
ulink
m
od
el
of close
d
l
oop A
VR syst
em
w
it
h
PI
D
contr
oller
Table
2
.
C
on
t
r
oller
gain
a
nd t
i
m
e resp
onse
s
pecifica
ti
ons for Z
-
N
m
et
ho
d o
f
tu
ning
PID
Co
n
troller
gain
K
p
K
i
K
d
0
.99
4
1
.56
7
0
.21
1
Ti
m
e
r
esp
o
n
se sp
ecif
i
catio
n
s
Ris
e ti
m
e
(
Tr
)
in s
ec
Settlin
g
ti
m
e
(Ts
)
i
n
sec
% Peak
ov
ersh
o
o
t (
% Mp)
Peak
valu
e
(Cp
)
Peak
ti
m
e
(T
p
)
in
s
ec
0
.21
2
8
1
.07
4
3
1
.42
1
6
1
.31
4
2
0
.55
3
5
Figure
5. Cl
os
e
d
lo
op
respo
nse
of
AV
R
syst
e
m
(Z
-
N
tu
ning
)
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
Evolutio
nary
al
go
rit
hms
-
base
d
tu
ning
of P
I
D
c
on
tr
oller f
or a
n
A
VR syste
m
(
Pet
c
hinat
han Go
vi
ndan
)
3053
5.2. C
S
algorithm
-
based
tuni
ng
In
th
is
resear
ch
wor
k,
t
he
cuc
koo
sear
ch
al
go
rithm
has
bee
n
im
ple
m
ented
f
or
i
m
pr
ovin
g
the
pe
rfor
m
ance
of
t
he
P
ID
c
on
t
ro
ll
er.
The
par
am
et
ers
consi
der
e
d
f
or
CS
al
gorithm
s
are
as
fo
ll
ows:
Num
ber
of
nests
(
n)
=
25,
Disc
overy
rate
of
al
ie
n
e
gg
s
(P
a
)
=
0.2
5,
T
otal
nu
m
ber
of
it
erati
on
s
=
100.
In
t
his
work,
var
i
ou
s
ob
j
ect
ive
functi
ons
a
r
e
co
ns
ide
re
d
for
op
ti
m
al
tun
in
g
of
P
ID
co
ntr
oller
as
m
entioned
in
the
(
8
)
to
(
12)
.
Fo
r
eac
h
obj
ec
ti
ve fu
nctio
n, the
cu
c
koo sea
r
ch
alg
or
it
hm
is
sim
ulate
d
f
or
10
num
ber
s
of
tim
es. Th
e
opti
m
iz
ed
PI
D
c
on
t
ro
ll
er
p
aram
et
ers
an
d
tim
e
do
m
ai
n
s
pecifica
ti
ons
a
r
e
rec
orde
d
f
or each run. A
fte
r
10
num
ber
o
f
r
uns,
the av
er
age
value
of PID c
on
t
ro
ll
er
par
am
et
e
rs
an
d
ti
m
e
d
om
ai
n
sp
eci
ficat
ion
s
for
each
obj
ect
iv
e f
un
ct
i
on
a
re
ta
bu
la
te
d
i
n
Ta
ble
3.
T
he
ste
p
respo
ns
e
of
cl
ose
d
lo
op
AV
R
con
t
ro
l
syst
em
for
eac
h
obj
ect
ive
f
unct
io
n
is
s
ho
w
n
in
Fig
ur
e
6.
From
the
Table
3
and
Fig
ur
e
6,
it
has
bee
n
co
nc
lud
e
d
that,
t
he
perform
ance
of
t
he
CS
al
go
rithm
base
d
P
ID
co
nt
ro
ll
er
with
I
AE
obj
ect
iv
e
f
unc
ti
on
is
pro
du
ce
d
bette
r
r
esp
on
se
with
le
ss
oversho
ot,
rise
ti
m
e
and
set
tl
ing
tim
e.
Table
3
.
C
on
t
r
oller
gain
a
nd t
i
m
e resp
onse
s
pecifica
ti
ons for va
rio
us
obj
e
ct
ive fun
ct
io
ns (CS
base
d
t
un
i
ng)
Ob
jectiv
e
f
u
n
ctio
n
Kp
Ki
Kd
Ris
e ti
m
e
(Tr
)
in
sec
Settlin
g
ti
m
e
(Ts
)
in
sec
% Peak
ov
ersh
o
o
t
(% M
p
)
Peak
valu
e
(Cp
)
Peak
ti
m
e
(T
p
)
in
se
c
MSE
0
.99
9
0
.46
0
9
0
.2
0
.22
6
2
1
.41
5
5
1
9
.58
5
2
1
.18
7
8
0
.50
6
3
ISE
0
.87
6
0
.46
0
9
0
.2
0
.24
1
2
1
.40
0
7
1
4
.59
0
7
1
.14
0
6
0
.55
6
1
IT
SE
0
.86
1
0
.47
1
8
0
.19
9
8
0
.24
3
5
1
.37
7
1
1
4
.09
1
1
1
.13
6
5
0
.55
6
1
IT
A
E
0
.67
0
.47
3
6
0
.19
9
8
0
.28
0
2
0
.77
7
1
5
.43
3
1
.05
4
4
0
.55
8
1
IAE
0
.65
1
0
.46
0
4
0
.19
9
9
0
.28
4
7
0
.74
2
3
4
.43
4
6
1
.04
4
6
0
.57
3
3
Co
m
b
in
ed
0
.55
4
3
0
.39
0
4
0
.18
2
0
.33
4
0
.53
0
.08
5
3
1
.00
2
0
.65
8
1
Ba
sed
on
t
he c
om
par
at
ive
perform
ance
anal
ysi
s
an
d
perfor
m
ance
of
va
rio
us
obj
ect
i
ve
functi
ons
(8
)
t
o
(
12
),
the
ne
w
com
bin
ed
obj
ect
iv
e
f
un
ct
i
on
has
been
form
ulate
d
as
m
entione
d
in
(
13
)
.
I
n
orde
r
to
i
m
pr
ove
the
perform
ance
of
PID
co
ntr
oller,
in
this
c
om
bin
ed
obj
ec
ti
ve
f
unct
ion
ti
m
e
respo
ns
e
s
pecifica
ti
ons
a
re
al
s
o
consi
der
e
d
al
ong
with
IAE
a
nd
weig
ht
fact
or
.
T
he
C
S
al
gorithm
based
PID
con
t
ro
ll
er
with
com
bin
ed
obj
e
ct
ive
functi
on
is
sim
ula
te
d
for
th
e
10
num
ber
of
it
erati
ons.
The
perform
ance
of
c
om
bin
ed
ob
j
ect
ive
f
unct
ion
i
s
com
par
ed
with
the
pe
r
form
ance
of
I
AE
obj
e
ct
ive
functi
on.
Fr
om
the
com
par
at
ive
analy
s
i
s
show
n
in
Ta
ble
3
and
Fig
ur
e
6 i
t
has
been
c
oncl
ud
e
d t
hat,
the
perform
ance
of
com
bin
ed
obje
ct
ive
functi
on
base
d P
ID
c
on
trolle
r
has
yi
el
de
d
a
be
tt
er r
es
pons
e
with im
pr
oved
peak o
ve
rsho
ot,
rise ti
m
e and
set
tl
ing
tim
e.
Figure
6.
The
c
losed
lo
op r
es
ponse
of
t
he AV
R sy
stem
f
or va
rio
us
obj
ect
iv
e f
un
ct
io
ns
(CS
al
go
rithm
-
base
d
tu
ning)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
30
47
-
3056
3054
5.3.
PS
O
algo
ri
th
m
-
ba
se
d
t
uning
Fr
om
the
CS
al
go
rit
hm
-
based
tu
ning
it
has
bee
n
prov
e
d
that,
IAE
and
c
om
bin
ed
obj
ect
iv
e
functi
on
-
base
d
tun
in
g
has
pro
du
ce
d
bette
r
pe
rfor
m
ance
f
or
the
set
point
t
rack
i
ng
pro
ble
m
.
In
order
t
o
ens
ur
e
the
ef
fecti
ven
e
ss
of
the
CS
a
lgorit
hm
in
tun
in
g
of
P
I
D
c
on
t
ro
ll
er,
the
perform
ance
of
the
CS
al
gor
it
h
m
is
com
par
ed
a
nd
analy
zed
wit
h
PS
O
al
gorit
hm
in
this
wo
r
k.
T
he
P
SO
par
am
et
ers
are
sel
ect
ed
as
f
ollows
:
dim
ension
(
d)
=
3,
popula
ti
on
siz
e
=
25,
m
axim
u
m
nu
m
ber
of
bir
d
ste
p
=
100,
c
ogniti
ve
fact
or
(C
1
)
=
1.2,
so
ci
al
accel
erat
ion
facto
r (C
2
) =
1
.
2, ine
rtia
wei
gh
t
facto
r
(
w
)
=
0.9.
The
c
om
par
at
ive
pe
rfo
rm
ance
analy
sis
of
CS
an
d
PS
O
bas
ed
P
ID
t
un
i
ng
with
Z
-
N
t
un
i
ng
is
s
hown
in
Table
4.
F
ro
m
t
he
Ta
ble
4
a
nd
the
Fig
ur
e
7,
it
has
been
c
oncl
ud
e
d
that
,
the
e
vo
l
ution
a
ry
al
gorithm
has
pro
du
c
e
d
bette
r
perform
a
nce
tha
n
t
he
Zi
egler
-
Nich
ols
m
et
ho
d.
Am
ong
the
pro
pose
d
evo
l
ution
a
ry
al
gorithm
s,
the
Cuck
oo
Searc
h
al
gorith
m
has
yi
el
ded
a
bette
r
res
ponse
than
the
PS
O
al
gorithm
with
bette
r
ti
m
e
respon
se
s
pecifi
cat
ion
s
in the se
t p
oin
t
tracki
ng probl
e
m
.
Figure
7
.
Cl
os
e
d respo
ns
e
of
AV
R
syst
em
f
or CS a
nd PSO
b
ase
d PI
D
tu
ni
ng
Table
4
.
C
on
t
r
oller
gain
a
nd t
i
m
e resp
onse
s
pecifica
ti
ons
f
or CS a
nd PSO
b
ase
d
t
un
i
ng
with Z
-
N
tu
ning
Co
n
troller para
m
e
t
ers
&
ti
m
e
resp
o
n
se Sp
ecif
icatio
n
s
Ob
jectiv
e Fun
ctio
n
Z
-
N
PSO
-
I
AE
PSO
Co
m
b
in
ed
CS
-
IA
E
CS
Co
m
b
in
ed
K
p
0
.99
4
0
.97
9
2
0
.54
5
5
0
.65
1
4
0
.55
4
3
K
i
1
.56
0
.73
7
4
0
.41
2
7
0
.46
0
3
7
0
.39
0
4
K
d
0
.21
1
0
.37
4
9
0
.19
5
7
0
.19
9
9
4
0
.18
1
7
9
Ris
e ti
m
e
(
T
r
)
in
s
ec
0
.21
2
8
0
.16
8
9
0
.32
7
1
0
.18
4
7
0
.
3
4
4
Settlin
g
ti
m
e
(T
s
)
i
n
sec
1
.07
4
0
.89
8
2
1
.13
7
5
0
.74
2
3
0
.52
9
9
% Peak
ov
ersh
o
o
t (
% Mp)
3
1
.42
1
6
1
3
.87
2
0
.45
3
2
4
.43
4
6
0
.08
5
3
Peak
valu
e
(Cp
)
1
.31
4
2
1
.14
1
.00
6
1
.04
4
6
1
.00
15
Peak
ti
m
e
(
T
p
)
in
s
ec
0
.55
3
5
0
.35
5
2
.33
2
6
0
.57
3
3
0
.65
8
1
6.
IMPLEME
N
TATION
OF
TUNING
OF
PID CO
NTR
OLL
ERS
–
DI
STUR
B
A
NCE
R
EJE
CTIO
N
The
AV
R
syst
e
m
has
bee
n
i
m
ple
m
ented
s
o
fa
r
i
n
this
pa
per
f
or
t
he
s
et
po
i
nt
trac
kin
g
pro
blem
.
In
ord
e
r
to
ana
ly
se
the
pe
rform
an
ce
of
ev
ol
ution
a
ry
al
gori
thm
-
based
tu
ni
ng
of
P
ID
co
nt
ro
ll
er,
the d
ist
urba
nce
rej
ect
io
n
pro
bl
e
m
al
so
reali
ze
d
in
this
resea
r
ch
wor
k.
The
CS,
P
SO
an
d
Z
-
N
m
et
ho
ds
base
d
P
ID
c
ontrolle
rs
hav
e
bee
n
im
ple
m
ented
to
so
l
ve
the
dist
urba
nce r
ejecti
on problem
.
In
this
work, I
AE
an
d
com
bin
ed
obje
ct
ive
functi
on
ha
s
be
en
co
ns
i
der
e
d
as
an
obj
ect
ive
functi
on
f
or
C
S
an
d
PS
O
bas
ed
tu
ning
of
PID
co
ntr
oller.
B
ecaus
e
these
t
wo
ob
j
e
ct
ive
f
unct
ion
s
we
re
pro
duce
d
bette
r
pe
rform
ance
in
set
point
t
rack
i
ng
pro
ble
m
.
The
Sim
ul
ink
m
od
el
o
f
the
A
VR
syst
e
m
fo
r
the
distu
rb
a
nce
rej
ect
io
n
pr
ob
l
e
m
is
sh
own
in
Figure
8.
I
n
th
is
si
m
ulati
on
stud
y,
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
Evolutio
nary
al
go
rit
hms
-
base
d
tu
ning
of P
I
D
c
on
tr
oller f
or a
n
A
VR syste
m
(
Pet
c
hinat
han Go
vi
ndan
)
3055
bo
t
h
posit
ive
a
nd
ne
gative
in
put
distu
rb
a
nce
has
been
i
ntr
oduced
i
n
the
A
V
R
syst
e
m
after
reachi
ng
ste
a
dy
sta
te
in
bet
ween
3
to
4
sec
as
s
hown
in
Fig
ur
e
9
.
The
outp
ut
re
sp
onse
of
distu
rb
a
nce
rej
ect
i
on
prob
le
m
us
in
g
CS,
PSO
an
d
Z
-
N
m
et
ho
d
-
base
d
tun
in
g
of
P
ID
con
t
ro
ll
ers
f
or
pro
po
se
d
obj
e
c
ti
ve
f
un
ct
io
ns
is
sho
wn
in
Fig
ur
e
9.
Fr
om
the
Fig
ure
9
it
h
as
been
con
cl
ud
e
d
th
at
,
the co
m
bin
ed object
ive
funct
ion
-
base
d
C
S a
lgorit
hm
tun
ing
h
a
s
pro
du
ce
d bett
er
perform
ance than
PS
O
a
nd
Z
-
N
b
a
sed
tu
nin
g.
Figure
8.
Ma
tl
ab
-
sim
ulink
m
od
el
of
AV
R
s
yst
e
m
w
it
h
PID c
on
t
ro
ll
er
f
or
disturba
nce rejecti
on
prob
le
m
Figure
9
.
The
Cl
os
ed
l
oop re
sp
onse
of st
he AVR
syst
em
f
or d
ist
ur
ban
ce
rej
ect
io
n
us
in
g
Z
N, CS
and PS
O base
d t
un
i
ng
7.
CONCL
US
I
O
N
In
this
w
ork,
tun
in
g
of
P
I
D
con
t
ro
ll
er
para
m
et
ers
us
in
g
CS,
PS
O
an
d
Z
-
N
m
et
ho
ds
pr
ese
nt
in
the
c
on
t
ro
l
of
AV
R
Syst
em
.
The
pro
posed
m
et
ho
d
fi
nd
s
out
the
op
ti
m
al
par
am
et
ers
of
PID
c
ontrolle
r
by
so
l
ving
the
opti
m
iz
at
io
n
pr
ob
le
m
fo
r
m
ini
m
iz
ing
the
obj
ect
ive
func
ti
on
com
pr
isi
ng
I
AE,
rise
ti
m
e,
set
tl
ing
tim
e
an
d
peak
over
s
hoot.
From
the
num
ero
us
re
su
lt
s
of
sim
ulati
on
,
it
has
bee
n
con
cl
ud
e
d
that
,
CS
al
go
rithm
ba
sed
tun
in
g
yi
el
ds
bette
r
co
ntr
oller
pe
r
f
or
m
ance
than
P
SO
a
lgorit
hm
-
based
tun
in
g
a
nd
it
is
far
bette
r
tha
n
conve
ntion
al
Z
-
N
m
et
ho
d.
Fro
m
the
Figure
7
and
Ta
ble
4
it
ha
s
bee
n
sho
wn
that,
the
pe
rce
nt
age
peak
ove
rs
hoot
has
bee
n
a
bundantly
re
duced
to
0.085
3
%
in
CS
with
com
bin
e
d
obj
ect
iv
e
f
un
ct
io
n
t
han
32.
421%
in
t
he
Z
-
N
m
et
ho
d.
Sim
i
l
arly
,
the
ti
m
e
do
m
ai
n
sp
eci
fi
cat
ion
s
hav
e
be
en
ric
hly
co
unte
r
balance
d
by
the
ap
plica
tio
n
of
the CS al
gorith
m
an
d
the
perf
or
m
ance of the
AVR co
ntr
ol s
yst
e
m
h
as also
been ab
unda
ntl
y im
pr
oved
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
30
47
-
3056
3056
REFERE
NCE
S
[1]
Vijay
V
it
t
al
and
Jam
es
D
M.
,
“
Pow
er
S
y
s
te
m
Co
ntrol
and
Stabilit
y
,
”
Third
edi
t
ion
IE
EE
Press
Po
wer
Eng
ineeri
ng
Serie
s,
A
Jhon wil
e
y
&
sons
,
Inc
.
,
publ
icati
ons,
2
020.
[2]
A.
Visioli
,
“
Tu
ning
of
PID
cont
roll
e
rs
with
f
uzzy
log
ic
,”
Pr
oc.
Inst.
Elec
t
.
Eng.
Control
T
heory
Applicati
on
,
vol.
148
,
no
.
1
,
p
p.
1
-
8
,
2001
.
[3]
T.
L
.
Seng
,
M.
B.
Khali
d
,
and
R.
Yus
of
,
“
Tuning
of
a
n
eur
o
-
fu
zzy
con
troller
b
y
g
ene
t
ic
al
gor
ithm
,”
IEEE
Tr
ans.
Syst.
,
Man,
Cyb
e
rn.
B
,
vol
.
29
,
no
.
2
,
pp
.
226
-
236
,
1999
.
[4]
R.
A.
Krohl
ing
and
J.
P.
Re
y
,
“
Design
of
opti
m
al
d
isturb
ance
re
je
c
ti
on
PID
controlle
rs
using
g
e
net
i
c
a
lgori
thm,
”
IEE
E
Tr
ans.
Ev
o
l.
Comput
,
vo
l. 5
,
no
.
1
,
pp
.
78
-
82
,
2001
.
[5]
Petc
hin
at
han
,
e
t
al.
,
“
Lo
ca
l
li
ne
ar
m
odel
tr
ee a
nd
Neuro
-
Fuzz
y
s
y
s
te
m
for
m
odellin
g
and
cont
ro
l
of
an
expe
r
imenta
l
pH ne
utralizat
io
n
p
roc
ess
,”
Braz
il
ian Journal
o
f Chemic
al
Engi
n
ee
ring
,
vol
.
31
,
no.
2
,
pp
.
483
-
4
95
,
2014
.
[6]
Binod
Kum
ar
Sahu,
et
a
l.
,
“
T
ea
ch
ing
–
learni
n
g
base
d
opt
imiza
t
ion
a
lgori
th
m
base
d
fuz
z
y
-
PID
cont
roll
e
r
for
aut
om
at
i
c
g
ene
r
at
ion
cont
ro
l
of
m
ult
i
-
area
powe
r
s
y
st
em
,”
Appl
i
e
d
Sof
t
Computi
ng
,
vol
.
27
,
pp
.
2
40
–
249
,
2015
.
[7]
Petc
hin
at
han
,
et
al.
,
“
H
y
br
id
P
SO
-
Bac
te
ri
al
Fo
rag
ing
B
ase
d
In
te
lligent
PI
Con
trol
ler
Tun
ing
f
or
pH
Proce
ss
,”
Proce
ed
ings
of
the
Inte
rnat
iona
l
Con
fe
renc
e
on
Information
S
y
stems
Design
a
nd
Int
el
l
ige
nt
A
ppli
cation
s
201
2
(
INDIA
2012)
,
pp.
515
-
522
,
201
2
.
[8]
G.J.
A.
Soto
,
J.
M.G.
López
,
J.
A
Herná
ndez
-
R
ive
ros
,
“
Coupled
evol
ut
iona
r
y
tuni
ng
of
PID
Control
l
ers
fo
r
the
B
enc
hm
ark
on
Vapor
Com
pr
ession
Refr
ig
erat
ion
,”
IF
AC
-
Pap
ersO
nLine
,
vol
.
51,
no
.
4
,
pp
.
50
9
-
514
,
2018
.
[9]
Moham
ed
Y,
M
oez
G
.
,
“
A
Novel
Te
chn
ique
fo
r
Tuni
ng
PI
-
con
trol
ler
in
Sw
it
ch
ed
Re
luc
t
ance
Motor
Drive
for
Tra
nsporta
ti
on
S
y
stem
s,
”
Inte
r
nati
onal
Journa
l
of
El
e
ct
ri
cal
a
nd
Computer
E
ngine
ering
(
IJECE)
,
vol
.
8
,
no.
6,
pp.
4272
-
4281
,
2018
.
[10]
Khanduja
,
N.
an
d
Bhushan,
B
.
,
“
CS
TR
Control
Us
ing
IMC
-
PID
,
PS
O
-
PID
,
and
H
y
brid
BBO
-
FF
-
PID
Control
le
r
,”
In
Appl
i
cat
ions
of
Arti
f
ic
ia
l
Int
ellige
n
ce
Techn
iq
ues
in
Engi
n
ee
ri
ng.
Adv
an
ce
s
in
Inte
lligen
t
System
s
and
Computing
,
Springer,
Sing
ap
ore
,
pp
.
519
-
526
,
2019
.
[11]
Shukor,
N.S.A.
,
Ahm
ad,
M.A.
and
Tumari
,
M.
Z.
M.
,
“
Data
-
dr
i
ven
PID
tuni
ng
base
d
on
sa
fe
expe
riment
at
io
n
d
y
nami
cs
for
c
ontrol
of
li
qu
id
slosh
,”
In
Proce
ed
ings
of
8
th
IE
EE
Control
and
Syst
em
Gr
aduate
Re
sear
c
h
Coll
oquium,
pp.
62
-
66
,
2017
.
[12]
Odili
JB,
e
t
al
.
,
“
Para
m
et
ers
-
tu
ning
of
PID
c
ontrol
ler
for
aut
o
m
at
ic
vol
ta
ge
re
gula
tors
using
t
he
Afric
an
buff
al
o
opti
m
iz
ation
,
”
P
LoS
ONE
,
vol
.
1
2,
no
.
4
,
pp
.
1
-
1
7,
2
017
.
[13]
RK
Manda
va
,
P
R
Vundavilli
,
“
Tuni
ng
of
PID
cont
roller
par
am
et
ers
o
f
a
B
iped
Robot
using
I
W
O
al
gorit
hm
,
”
in
proce
edi
ngs o
f
th
e
4th
in
te
rnat
ional
conf
ere
n
ce on
me
chat
ronic
s and
robotic
s
e
ngine
ering
,
pp
.
90
-
94,
2018
.
[14]
Badriy
ah
A
O,
Am
ee
r
L
S,
Abb
as
K
K
.
,
“
Resol
ving
of
opt
imal
fra
ctional
PID
c
ontrol
ler
for
DC
m
otor
dr
ive
b
ase
d
on
ant
i
-
windup
b
y
inva
sive
w
e
ed
opti
m
izati
on
te
chn
ique
,
”
In
d
onesian
Journ
al
of
E
lectric
a
l
Engi
ne
ering
a
nd
Computer
Scien
ce
(
IJEECS)
,
vol
.
15
,
no
.
2
,
pp
.
9
5
-
103,
2019
.
[15]
Aza
r
A.T.,
Ser
r
ano
F.E
.
,
“
Frac
ti
onal
Order
Sli
ding
Mode
PID
Control
l
er/
Obs
erv
er
for
Con
tinuous
Nonline
a
r
Sw
it
che
d
S
y
s
tem
s
with
PS
O
Pa
ramet
er
Tuni
ng
,”
in
pro
ce
ed
ings
of
in
te
rnationa
l
Confe
renc
e
on
A
dvanc
ed
Mac
hin
e
Learning
Techn
ologi
es
and
Applicat
ions
(
AMLT
A2018)
,
pp.
13
-
2
2,
2018
.
[16]
Sha
y
egh
i
H,
Dada
shpour
J
.
,
“
Anarc
hic
Soc
ie
t
y
Optimiz
at
ion
Based
PID
C
ontrol
of
an
Autom
at
ic
Volt
a
g
e
Regul
at
o
r
(AV
R)
S
y
st
em
,
”
Journal
of
Elec
tric
al
a
nd
Elec
troni
c En
gine
ering
,
vol
.
2
,
no
.
4
,
pp
.
199
-
207,
2012
.
[17]
M.
Kashki,
et
al
.
,
“
A
r
ei
nfor
ce
m
ent
learni
ng
au
to
m
at
a
op
ti
m
iz
a
ti
o
n
appr
o
ac
h
for
o
pti
m
um
tuni
ng
o
f
PID
con
trol
l
er
in
AV
R
s
y
s
te
m
,”
in
Int
ernati
on
al
Conf
ere
nc
e
o
n
Intelli
g
en
t
Co
mputing,
Adv
an
ce
d
Intelli
g
ent
C
omputing
Theori
es
and
Applications
,
pp
.
684
–
692
,
2
008
.
[18]
Deva
raj
D,
Sel
vaba
l
a
B.
,
“
Re
al
-
cod
ed
gen
et
i
c
al
go
rit
hm
an
d
fuz
z
y
logic
appr
oac
h
for
re
al
-
ti
m
e
tuni
ng
o
f
pro
-
porti
on
al
-
in
te
gra
l
-
der
iv
at
iv
e
cont
rol
le
r
in
au
tomati
c
vo
ltage
reg
ulator
s
y
st
em,
”
I
ET
Gene
r.
T
rans
m.
Distrib.
,
vol.
3
,
no
.
7
,
pp
.
641
-
649,
2009
.
[19]
H.
Zhu,
et
al
.
,”
CAS
al
gorit
hm
-
base
d
opti
m
um
design
of
PID
cont
roll
e
r
in
AV
R
s
y
stem
,
”
Chaos Soli
tons
Fract
a
l
s.
,
vol.
42
,
no
.
2,
pp
.
792
–
800
,
2009
.
[20]
H.
Gozde
,
M.C.
Ta
pl
a
m
ac
iogl
u
,
“
Com
par
at
ive
pe
rform
anc
e
an
alys
is
of
art
if
ic
i
al
b
e
e
col
on
y
al
gor
it
h
m
for
aut
o
-
m
at
i
c
volt
ag
e
reg
u
lator
(AV
R)
s
y
stem
,”
J. F
rank
l. Inst.
,
vol.
348
,
no
.
8
,
p
p.
1927
-
1946
,
2
011
.
[21]
D.H.
Kim
.
,
“
H
y
brid
GA
–
BF
base
d
intelligen
t
PI
D
cont
roller
tuning
for
AV
R
s
y
st
em
,”
App
l.
Soft
Comput
,
vol
.
11,
no.
1
,
pp
.
11
-
22
,
2011.
[22]
P.K.
Mohant
y
,
et
al.
,
“
Tun
ing
and
assess
m
ent
of
proporti
onal
–
integ
ral
–
der
iva
t
ive
co
ntrol
ler
for
an
au
tomati
c
vo
lt
ag
e
reg
ulator
s
y
stem
emplo
y
ing
lo
cal
unimoda
l
sam
pli
ng
al
gor
it
hm
,”
El
e
ct
r.
Pow
er
Compon.
Syst
.
,
vol
.
42
,
no
.
9
,
pp.
959
-
969
,
20
14
.
[23]
Z.
Bingu
l,
O. K
a
rah
an
,
“
A
Novel
per
form
anc
e
crit
eri
on
appr
oac
h
t
o
opti
m
um
design of
PID
cont
rol
l
er using
cuc
koo
sea
rch
al
gor
it
hm
for
AV
R
s
y
s
te
m
,”
Journal
o
f
th
e Frankl
in
Instit
u
t
e
,
vo
l. 355, no. 1
3,
pp
.
55
34
–
555
9,
2018
.
[24]
A.
Sikand
er,
et
a
l
.
,
“
A
nove
l
tech
nique
to
design
c
uckoo
se
arc
h
bas
ed
FO
PID
cont
r
oll
er
for
AV
R
in
power
s
y
s
te
m
s
,”
Computers and Electric
a
l
Eng
in
ee
ring
,
vol
.
70
,
pp.
261
–
274
,
20
18.
[25]
Kenne
d
y
J.
,
Eb
erh
art
R.
C
.
,
“
Parti
cles
Sw
arm
Optimiza
ti
o
n
,”
Pr
oc.
IEEE
In
t
ernati
onal
Conf
ere
nce
on
N
eural
Net
works, Perth
Australi
a,
IEEE Service
C
ent
er,
Pi
scataway,
N
J
.
,
pp.
1942
-
1948
,
1995
.
[26]
Ebe
rha
r
t R
.
C,
Kenne
d
y
J
.
,
“
A Ne
w
Optimizer
Us
i
ng
Particle
s
Swarm
Theory
,”
Pro
c.
S
ixth
Int
ernational
Symposium
on
Mic
ro M
achin
e
and
Hum
an
S
ci
en
ce,
Nagoya
-
Japan,
I
EEE
S
erv
ice
Cen
te
r,
P
isc
ataway
,
NJ
,
pp
.
39
-
43
,
1995
.
[27]
X.
-
S.
Yang,
S.
D
eb.
,
“
Cuckoo
se
a
rch
via
L´
ev
y
f
lights
,”
Proc.
of
World
Congress
on
Nature &
Bi
o
logi
cally
Inspire
d
Computing
(
NaBIC
2009)
,
pp.
21
0
-
214,
2009
.
Evaluation Warning : The document was created with Spire.PDF for Python.