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.
4
,
Decem
be
r 202
0
, p
p.
2164
~
217
2
IS
S
N:
20
88
-
8694
,
DOI: 10
.11
591/
ij
peds
.
v11.i
4
.
pp
2164
-
217
2
2164
Journ
al h
om
e
page
:
http:
//
ij
pe
ds
.i
aescore.c
om
PID opti
mal co
ntrol to r
educ
e ene
rgy c
ons
umptio
n in DC
-
drive
system
Ha
ri
Maghfi
r
oh
1
,
Muha
mm
ad
Niza
m
2
, Su
priy
an
t
o
Pr
apto
diy
ono
3
1,2
Depa
rtment
of
Elec
tr
ical Engi
n
ee
ring
,
Seb
el
as
Mare
t
Univ
ersity
, I
ndonesi
a
3
Depa
rtment of
El
e
ct
ri
ca
l
Eng
in
ee
ring
,
Sult
an
A
geng
T
irt
ay
asa
Univer
sity
,
Indo
nesia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
M
a
y
9
, 2
0
20
Re
vised
Ju
n
10
, 20
20
Accepte
d
J
ul
9
, 20
20
The
cont
rol
sys
te
m
that
is
wid
el
y
used
in
ind
ustry
is
PID
(Proportional
Inte
gra
l
Der
iva
t
i
ve)
.
A
lm
ost
90
%
of
industr
ie
s
stil
l
us
e
PID
co
ntrol
sys
tems
bec
ause
of
i
ts
simpl
icity,
ap
pli
c
abi
lity,
and
re
li
ab
il
i
ty.
H
oweve
r,
th
e
wea
kness
of
PI
D
is
tha
t
it
t
akes
a
long
ti
m
e
to
tune.
PID
cont
r
ol
with
good
per
forma
n
ce
and
low
ene
rgy
con
sumpti
on
ca
n
be
a
chi
ev
ed
using
GA
tuni
n
g
with
th
e
appr
opr
ia
t
e
ob
je
c
ti
v
e
fu
nct
ion
.
The
con
t
ribut
ion
of
thi
s
pape
r
is
to
propose
the
i
mp
le
m
ent
a
ti
on
of
LQR
cont
ro
l
in
the
for
m
of
PID
using
GA
tuni
ng
wi
th
L
QR
objecti
v
e
func
ti
on
.
The
proposed
al
g
orit
hm
was
im
plemented
bo
th
in
the
si
mul
a
t
ion
and
har
dwar
e
whi
ch
is
a
m
i
ni
conve
yor
with
a
DC
mot
o
r.
Th
e
resul
t
sho
ws
tha
t
th
e
prop
osed
al
gor
it
hm
i
s
bet
ter
in
both
IAE
and
en
erg
y
consu
mpt
ion
co
mpa
r
ed
wi
th
other
PID
tuni
ng,
Zi
egler
–
Nicho
ls
(ZN),
and
GA
with
IAE
obj
ec
t
iv
e
func
t
ion.
Com
par
ed
with
PID
ZN,
it
h
as
I
AE
and
ene
rgy
r
educ
t
ion
by
2
.
76
%
and
16
.
07
%
r
espe
ctivel
y
.
Although
it
s
p
er
forma
nc
e
is
lower
tha
n
th
e
LQR,
i
t
h
as
o
the
r
adv
ant
ag
es
th
at
use
fewe
r
senso
rs.
The
o
the
r
ad
vant
ag
e
of
th
e
proposed
me
tho
d
is,
PID
is
more
f
am
i
li
ar
u
sing.
Th
ere
for
e,
it
ea
sy
to
be
i
mpl
ement
ed
in
the
ex
isti
ng
sys
te
m
withou
t
a
lot of
cha
nges
.
Ke
yw
or
d
s
:
DC m
otor
Energ
y
GA
LQR
PI
D
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
:
Har
i
M
a
ghfir
oh,
Dep
a
rtme
nt of
Ele
ct
rical
En
gi
neer
i
ng,
Sebelas
M
a
ret
Un
i
ver
sit
y,
Jl.Ir. S
utami
36A
, Sur
a
ka
rta
–
57126,
Ind
on
e
sia
.
Emai
l:
h
ari.
ma
ghfir
oh@
gm
ai
l.com
1.
INTROD
U
CTION
On
e
te
ch
nolo
gy
t
o
s
upport
pro
duct
ion
s
pee
d
is
el
ect
ric
m
otors
with
hi
gh
perf
or
m
ance
,
ef
fici
enc
y,
sp
ee
d
dyna
mic
s,
an
d
go
od
l
oa
d
res
pons
e
.
T
he
re
are
tw
o
kinds
of
el
ect
ric
mo
to
r
w
hic
h
ar
e
AC
an
d
DC
mo
to
r.
DC
mo
t
or
has
so
me
a
dvanta
ge
s
su
c
h
as
eas
y
to
co
ntr
ol
the
sp
ee
d
or
posit
ion
a
nd
wi
de
ad
justable
ra
ng
e
[1
-
2].
Howe
ver,
it
also
ha
s
some
draw
bac
ks
,
on
e
of
the
m
is,
it
us
es
mec
ha
nical
commuta
tor
(bru
s
h)
w
hich
cau
s
e
hig
h
mainte
na
nce
c
os
t
[
3].
DC
mo
t
or
s
ar
e
wi
dely
use
d
as
in
ste
el
r
ol
li
ng
mil
ls,
el
ect
ric
trai
ns,
e
le
ct
ric
veh
ic
le
s,
a
nd
rob
otics
act
uator
s
[4].
Since
energ
y
is
an
importa
nt
iss
ue
tod
a
y,
a
nd
a
ccordin
g
to
[5],
the
el
ect
ric
mo
t
or
is
one
of
t
he
ap
pliances
th
at
co
ns
ume
c
onside
rab
le
ene
rgy,
the
c
ontr
ol
met
hod
w
hich
ca
n
reduce e
nerg
y consu
mp
ti
on
w
it
h
bette
r pe
rfo
rma
nce is
nee
de
d.
The
s
peed
c
on
trol
of
the
DC
mo
to
r
is
ge
neral
ly
ob
ta
ine
d
by
cha
ngin
g
it
s
te
rmin
al
volt
age
[
6].
PID
con
t
ro
l
is
one
of
the
well
-
known
al
gorith
m
s
for
sp
ee
d
c
ontr
ol.
Ba
sed
on
[
7
,
8]
,
al
m
ost
90
%
of
in
dus
trie
s
us
e
PI
D
c
on
tr
ol
be
cause
of
it
s
si
mp
li
ci
ty,
a
ppli
cabil
it
y,
a
nd
re
li
abili
ty.
H
ow
e
ver,
the
wea
kn
ess
of
PID
is
t
hat
it
ta
kes
a
lo
ng
ti
me
to
tu
ne
[
9
,
10]
.
Seve
ral
methods
of
tu
ning
PID
c
ontr
ols
ha
v
e
bee
n
pro
po
se
d.
T
his
tu
ning
method
ca
n
be
cat
egorized
int
o
i
)
e
mp
i
rical
methods
s
uc
h
as
Zie
gler
-
Nic
ho
ls
(
ZN
)
an
d
Coh
e
n
-
C
oon
(
CC
),
ii
)
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
PID opti
m
al c
on
tr
ol to
reduc
e ener
gy
c
onsumptio
n
i
n DC
-
dr
iv
e
syste
m
(
Ha
ri
Ma
ghfi
roh)
2165
analyti
cal
met
hods
su
c
h
as
r
oo
t
loc
us
(RL)
an
d
fr
e
quenc
y
res
pons
e
(
FR)
,
a
nd
ii
i)
opti
m
iz
at
ion
met
hods
su
c
h
as
us
in
g
G
enet
ic
Algori
thm
(
GA),
Partic
le
S
wa
rm
O
ptimi
zat
ion
(PSO
),
an
d
An
t
Colo
ny
Op
ti
miza
ti
on
(
ACO
)
[
8
,
10
,
11].
T
he
em
pi
ric
meth
od
is
easy
to
a
pp
l
y,
bu
t
the
resu
lt
s
are
not
ve
ry
good.
An
al
ytica
l
methods
ca
n
be
use
d
to
imp
rove
so
me
sp
eci
fic
par
a
mete
rs
of
the
syst
em
s
uc
h
as
sta
bili
ty,
r
isi
ng
ti
me an
d st
ead
y
-
sta
te
e
rror
s
. While
the
opti
miza
ti
on
m
et
hod i
s no
w wide
ly u
se
d beca
use
o
f
it
s ab
il
it
y
i
n
te
r
ms
of opti
miza
ti
on as in
[12] a
nd
[13].
Com
par
at
ive
stud
ie
s
on
t
he
us
e
of
opti
miza
ti
on
meth
od
s
f
or
P
I
D
tun
in
g
ha
ve
be
en
ca
r
rie
d
ou
t
by
[
14].
A
rtific
ia
I
I
ntell
igent
(AI)
met
hods
us
e
d
a
re
GA,
P
SO,
A
CO
an
d
E
volu
ti
on
ar
y
P
r
ogra
mmin
g
(EP),
w
hile
th
e
pa
rameter
s
c
ompare
d
a
re
s
et
tl
ing
ti
me,
ri
se
ti
me,
a
nd
overs
hoot.
Ba
se
d
on
the
th
ree
models
us
e
d,
G
A
exc
e
ls
in
al
l
par
ts
e
xcep
t
at
t
he
set
tl
ing
ti
me
in
one
m
od
el
.
T
he
us
e
of
G
A
f
or
PI
D
t
un
i
ng
has
be
en
done
a
s
in
[
15
,
16]
,
a
nd
[
17]
.
H
oweve
r,
al
l
these
st
ud
ie
s
us
e
c
os
t
functi
on
s
relat
ed
to
performa
nce
s
uch
a
s
IA
E
(Integ
ral of Abs
olu
te
M
a
gn
it
ude
of
E
rro
r)
a
nd
IT
AE
(
I
nteg
ral o
f Ti
m
e mu
lt
ipli
e
d
by
the Abso
l
ute Er
r
or).
In
[
18]
s
om
e
performa
nce
obje
ct
ive
functi
on
s
a
re
c
ompa
red
w
hic
h
is
M
SE
(
M
ea
n
of
the
S
quare
Error),
ITAE,
I
AE,
I
S
E
(
I
ntegral
of
the
S
quare
E
rror)
a
nd
ITSE
(
In
te
gr
al
of
Time
m
ulti
plied
by
the
S
quare
Error).
The
c
omparis
on
res
ults
f
r
om
ov
e
rs
hoot
an
d
set
tl
ing
ti
me
ar
e
show
n
that
t
he
I
AE
obje
ct
ive
f
unct
ion
give
s
the
lowest
overs
ho
ot
(
OS),
w
hile
IT
AE
giv
e
s
t
he
s
mall
est
set
tl
ing
ti
me.
I
n
[
19],
I
AE,
ISE,
an
d
M
SE
ob
j
ect
ive
functi
on also
c
ompare
d. The
re
su
lt
sho
ws
t
ha
t IA
E
h
a
s lo
we
r ov
e
r
-
s
ho
ot a
nd
IS
E
has
the l
ow
est
sett
li
ng
t
ime.
So
me
re
searc
he
rs
al
s
o
pro
pos
e
the
co
mb
i
nation
of
s
om
e
obje
ct
ive
f
un
ct
io
n
t
o
im
pro
ve
t
he
res
ult
of
GA
op
ti
mizi
ng
PID
c
ontr
ol.
A
rtu
ro
[10]
c
ombines
I
SE,
OS
,
a
nd
M
SE
as
an
ob
je
ct
ive
f
un
ct
io
n
wi
th
the
weig
hting
valu
e
to
en
ha
nce
c
orres
pondin
g
pe
rformance
cri
te
ria.
Wh
il
e
[20]
pr
opos
es
th
e
obje
ct
ive
fun
ct
ion
s,
wh
ic
h
are
the
com
bin
at
io
n
of
I
AE,
rise
ti
m
e
and
c
ontrolle
r
outp
ut.
H
ow
ever,
the
y
did
no
t
e
xp
la
in
t
he
eff
ect
of
c
ontrolle
r
outp
ut
in
the
obje
ct
ive
functi
on
af
fect
the
e
ne
rgy
us
ed
by
the
co
ntr
oller.
I
t
on
ly
c
ompa
r
es
the
performa
nce
be
tween
GA
a
nd
simple
x
m
et
hod
in
PID
tun
in
g.
Zahi
r
et
al
.
[21]
pr
opos
e
d
the
m
od
i
fied
ob
je
ct
ive
funct
ion
wh
ic
h
a
dding
ov
e
rs
hoot,
ste
ady
-
sta
te
error,
set
tl
ing
ti
me,
an
d
rise
ti
me
into
IT
AE,
IA
E
,
IS
E,
a
nd
ITSE
.
The
y
co
ncl
ud
e
t
hat
th
e
modifie
d
ob
je
ct
ive
f
unct
io
n
giv
es
im
provement
to
the
con
t
ro
ll
e
r
performa
nce.
PI
D
co
ntr
ol
wi
th
go
od
pe
rformance
a
nd
lo
w
e
nergy
co
nsum
ption
can
be
achieve
d
us
i
ng
G
A
tu
nin
g
with
t
he
a
ppr
opriat
e
obje
ct
ive
f
un
ct
i
on.
In
[
22
]
LQR
obje
c
ti
ve
f
unct
ion
a
s
G
A
obje
ct
ive
f
un
ct
i
on
to
tu
ne
t
he
PI
D
co
ntr
oller
is
pro
posed
.
LQR
is
on
e
of
the
opti
mal
con
t
ro
ls
w
hich
ca
n
be
us
e
d
to
re
du
ce
e
nerg
y
consu
mp
ti
on
by
a
dju
sti
ng
a
c
os
t
functi
on
[23].
Howe
ver,
[
22],
di
d
not
giv
e
a
cl
ear
anal
ys
is
of
t
he
e
ne
rgy.
Ther
e
f
or
e,
in
t
his
resea
rc
h,
t
he
L
QR
obje
c
ti
ve
functi
on
t
o
tu
ne
the
PID
via
G
A
will
be
performe
d
in
real
hard
war
e
t
o
prov
e
it
s
abili
ty
t
o
c
on
t
ro
l
bo
t
h
performa
nce
a
nd
ene
r
gy.
T
he
pro
posed
al
go
rithm
will
be
t
est
ed
in
a
mini
co
nv
eyor
dri
ve
n
by
a
D
C
mo
t
or
.
The
c
on
tri
bu
ti
on
of
t
his
pa
pe
r
is
to
pro
pose
the
impleme
nt
at
ion
of
LQR
co
ntr
ol
in
the
f
orm
of
PI
D
with
GA
tun
in
g
us
in
g
LQR
obje
ct
ive
functi
on.
T
he
adv
a
ntage
s
of
this
method
a
re
it
use
s
fe
wer
s
ens
or
s
co
mp
a
red
with
the
ori
gina
l
LQR
meth
od
w
hic
h
use
s
s
ens
or
s
as
ma
ny
as
t
he
sta
te
in
the
sy
s
te
m
model.
F
urt
hermo
re,
t
he
pro
po
se
d
al
go
r
it
hm
is
simple
r
co
mp
a
red
wi
th
Linea
r
Q
ua
dr
at
ic
Gau
s
sia
n (L
Q
G)
w
hich
can
e
li
minate
so
me
or all
sen
s
ors i
n
L
QR a
nd r
e
pl
ace t
hem b
y
a
n ob
s
er
ver.
2.
RESEA
R
CH MET
HO
D
2
.
1
.
Syste
m
D
esi
gn
The
s
ys
te
m
use
d
in
this
st
udy
is
a
mini
c
onveyo
r
wit
h
a
DC
m
otor.
T
he
DC
m
otor
use
d
has
a
12
V
5.5A
s
pecifica
ti
on
with
a
m
ax
imum
sp
ee
d
of
250
r
pm
a
nd
a
ma
ximum
to
r
qu
e
of
10.
6
kg/c
m.
Fi
gure
1
s
hows
the
mi
ni
c
onve
yor
us
e
d.
T
he
sy
ste
m
m
od
el
is
obta
ine
d
by
ta
kin
g
m
otor
i
nput
a
nd
ou
t
pu
t
data
i
n
the
form
of
vo
lt
age
a
nd
spe
ed
us
i
ng
a
da
ta
log
ge
r.
Using
this
data,
t
he
transf
e
r
f
unct
ion
of
t
he
sy
ste
m
is
der
iv
ed
usi
ng
M
A
TLAB
S
yst
em
I
de
ntific
at
ion
.
T
he
tra
ns
f
er
functi
on
of
the
s
ys
te
m
is
s
how
n
i
n
(
1)
.
T
he
c
ontr
ol
al
go
rith
m
is t
est
ed
in
MA
TLAB/ Sim
ulink be
f
or
e im
pl
emented
in
the
real ha
rdwa
re syste
m.
Figure
1. M
i
ni
conve
yor
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
D
ri
S
ys
t,
V
ol
.
11
, N
o.
4
,
D
ecembe
r
2020
:
2164
–
2172
2166
(
)
=
1
.
094
2
2
−
0
.
91
−
0
.
04
(1)
2
.
2
.
PID
Tuni
ng
and O
pt
im
al Con
tr
ol
2
.
2
.
1.
Zi
egler
Nichols
M
e
thod
Zie
gler
Nich
ol
s's
meth
od
is
div
ide
d
into
two
wa
ys
:
c
ur
ve
reacti
on
th
at
done
in
op
en
l
oop
a
nd
ulti
mate
cycle
wh
ic
h
done
in
cl
os
e
lo
op.
T
he
first
met
hod,
par
a
mete
rs
a
re
rathe
r
di
ff
ic
ult
to
est
imat
e
in
no
is
y
env
i
ronme
nts,
wh
il
e
the
seco
nd
one,
can b
e q
uite de
trime
ntal
to
the
s
ys
te
m
beca
us
e not al
l
of
the
syst
ems
ca
n
tolerat
e
su
sta
i
ning
osc
il
la
ti
on
co
ndit
ion
s
[
24]
.
In
t
his
re
search
,
the
sec
ond
met
hod
is
us
e
d
w
hich
done
i
n
simulat
ion.
PID
f
ormula
is
s
how
n
in
(
2).
T
he
ulti
mate
cy
cl
e
is
based
on
increase
Kp
from
0
to
crit
ic
al
gain
value
(
Kc
r
)
in
w
hich
the
out
pu
t
ex
hib
it
s
s
ust
ai
ned
os
ci
ll
at
ion
w
he
n
Ti
=
∞
a
nd
Td
=
0.
The
Kc
r
a
nd
t
he
corres
pondin
g peri
od (
Pcr
)
t
he
n use
d
to
calc
ulate
PID
gain
us
in
g
t
he
f
orm
ula in T
able
1.
(
)
=
(
1
+
1
+
)
(2)
Table
1.
Zie
gle
r
Nic
hols ulti
m
at
e cycle tu
ning
form
ula
[
25]
Co
n
troller
Kp
Ti
Td
P
0
.5
Kcr
-
-
PI
0
.45
Kcr
Pcr
/ 1.2
-
PID
0
.6
Kcr
0
.5
Pcr
0
.
1
2
5
Pcr
2
.
2
.
2.
Gene
tic
Algori
t
hm
M
eth
od
The
gen
et
ic
al
gorithm
(
GA)
i
s
a
r
obus
t
op
ti
miza
ti
on
te
ch
ni
qu
e
base
on
D
arw
i
n’
s
pr
i
ncip
le
of
natu
ral
sel
ect
ion
[26]
.
The
basic
goal
of
GA
is
to
opti
mize
the
fitness
f
un
ct
io
n.
This
meth
od
is
intr
oduce
d
by
Joh
n
Ho
ll
an
d
at
t
he Un
i
ver
sit
y o
f
M
ic
hi
gan in
19
70. Acco
r
ding
to
[
27]
G
A has
the
fo
ll
owin
g adv
a
ntage
s:
a.
It is
a sim
ple al
gorithm
that is
easi
ly un
der
st
ood an
d
im
plem
ented
b.
The
al
gorithm
is rob
us
t
c.
GA is a
non
-
li
near p
ro
ce
ss th
at
co
ul
d be a
pp
li
ed
to m
os
t i
ndus
t
rial
pro
ces
ses w
it
h g
ood r
esults
d.
GA searc
h
a
populat
ion o
f p
oi
nts instead
of
a sin
gle so
l
utio
n
e.
GA does
not
ne
ed
in
f
or
mati
on a
bout the
s
yst
em ex
ce
pt fo
r t
he
fitnes
s fu
nc
ti
on
.
Figure
2
s
how
s
the
flo
w
c
hart
to
im
plement
ing
G
A.
Starti
ng
f
rom
i
niti
al
i
zi
ng
popula
ti
on
wh
ic
h
is
a
set
of
so
l
utions
re
pr
ese
nted
by
c
hrom
osome
.
I
n
this
ste
p,
t
he
popula
ti
on
i
s
ge
ner
at
e
d
ra
ndom
l
y.
T
he
s
econ
d
ste
p
is
e
valuat
ing
t
he
fitness
base
on
the
obje
ct
ive
f
unct
ion.
In
a
c
ontr
ol
s
ys
te
m,
the
r
e
are
ma
ny
ob
je
ct
ive
functi
ons
us
e
d
as
sh
ow
n
in
Table
2
[
18]
.
The
thi
rd
st
ep
is
evaluati
ng
t
he
te
rmina
ti
on
crit
eria.
If
the
te
rmin
at
io
n
cri
te
ria
are
sat
isfi
ed
the
outp
ut
is
the
opti
mal
s
olu
ti
on.
O
n
th
e
oth
e
r
ha
nd,
t
he
ge
netic
ope
rati
on
will
be
pe
rformed
w
hich
is
r
epro
du
ct
io
n,
cr
os
s
ov
e
r,
an
d
m
utati
on
[
28]
.
R
epro
du
ct
io
n
is
simply
retai
ni
ng
a
fit
string
i
n
the
fol
low
in
g
ge
ne
ra
ti
on
,
c
ro
s
sover
involves
s
wa
ppin
g
pa
rtia
l
string
of
rand
om
le
ng
th
betwe
en
two
-
par
e
nt
stri
ngs,
an
d
mu
ta
ti
on
invol
ves
flippi
ng
a
ra
ndom
bit
in
t
he
stri
ng.
Re
peat
t
o
t
he
sec
ond
ste
p
un
ti
l
te
rmin
at
io
n
c
rite
ria are sati
sfi
ed.
Table
2.
T
ype
of f
it
ness
fu
nction
Fitn
ess
Fun
ctio
n
Equ
atio
n
IAE
∫
|
(
)
|
0
ISE
∫
(
)
2
0
M
SE
1
∫
(
(
)
)
2
0
IT
AE
∫
|
(
)
|
0
IT
SE
∫
(
)
2
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
PID opti
m
al c
on
tr
ol to
reduc
e ener
gy
c
onsumptio
n
i
n DC
-
dr
iv
e
syste
m
(
Ha
ri
Ma
ghfi
roh)
2167
Figure
2.
Ge
ne
ti
c algo
rith
m
proces
s f
l
ow cha
rt
2
.
2
.
3.
Op
tima
l Contr
ol
LQR
is
one
of
the
opti
mal
con
t
ro
l
te
c
hn
i
ques
w
hich
c
on
sider
t
he
sta
te
s
of
the
s
ys
te
m
an
d
c
ontrol
input
t
o
mak
e
op
ti
mal
co
ntr
ol
decisi
ons
.
Ac
cordin
g
to
[29]
,
this
meth
od
is
si
m
ple
a
nd
rob
us
t.
Sup
pos
e
the
sta
te
eq
uatio
n of t
he
s
ys
te
m
is:
̇
(
)
=
(
)
+
(
)
(3)
(
)
=
(
)
+
(
)
The
sta
te
fee
db
ack
c
ontr
ol
is
U
(t)
=
-
K
LQR
X
(t)
,
w
her
e
K
LQR
is der
i
ved
f
r
om
the
mi
nimiza
ti
on
of
c
os
t
functi
on as
sho
wn in (
4)
.
=
∫
(
(
)
(
)
+
(
)
(
)
)
(4)
The
matri
ces
Q
an
d
R
deter
mine
the
relat
ive
imp
ort
ance
of
t
he
er
ror
an
d
the
e
xp
e
ndit
ur
e
of
e
ne
rg
y
[30]
.
T
heref
ore,
the
tra
de
-
off
betwe
en
pe
rfo
rma
nce
a
nd
e
ne
rgy
us
e
d
ca
n
be
set
by
c
hoosi
ng
an
ap
pro
pri
at
e
el
ement
of
m
at
rices
Q
a
nd
R
. T
he bloc
k diag
ram of
L
QR contr
ol is s
how
n i
n
Fig
ure
3.
Figure
3. Bl
oc
k diag
ram of
L
QR c
on
tr
ol
2
.
2
.
4.
PI
D
O
p
timal
Contr
ol
The
pro
po
se
d
meth
od
is
P
ID
t
un
e
d
by
GA
us
i
ng
the
LQR
c
os
t
f
unct
i
on.
T
his
a
lgorit
hm
is
com
bin
in
g
t
he
adv
a
ntage
of
P
ID
,
G
A,
an
d
L
QR
to
el
imi
nat
e
the d
isa
dvant
age o
f
eac
h
of them
a
nd
to
i
nc
rease
the
pe
rforma
nc
e
with
l
ow
e
nerg
y
co
nsum
ption.
P
ID
is
simple
an
d
ea
sy
to
be
i
mp
l
emented
,
GA
has
the
powe
rful
sea
rc
hing
ca
pa
bili
ty
to
opti
mize
t
he
obje
ct
ive
f
unct
ion
,
w
hile
L
QR
is
the
op
ti
mal
c
on
tr
ol
m
et
hod
wh
ic
h
well
pe
r
forme
d
with
c
on
t
ro
ll
ed
e
ne
rgy
co
nsum
ptio
n.
In
[
22]
L
QR
ob
je
ct
ive
funct
ion
as
GA
obje
ct
ive
functi
on
t
o
t
une
t
he
PID
c
on
t
ro
ll
er
is
use
d.
H
oweve
r,
the
a
nalysis
of
performa
nc
e
an
d
e
nergy
is
not
exp
la
ine
d.
Fig
ur
e
4
is
the
dia
gr
a
m
of
the
propose
d
met
hod
wh
ic
h
us
i
ng
L
QR
obje
ct
ive
f
un
ct
io
n
in
t
he
GA
t
o
offli
ne
t
u
ne
th
e
PID.
Offli
ne
tu
ning
is
ch
ose
n
beca
us
e
it
is
si
m
pler
an
d
ca
n
be
im
plemented
in
lo
w
-
c
ost
hard
war
e
co
m
par
e
with
onli
ne
t
un
i
ng.
The
tu
ning
proces
s
ca
n
be
do
ne
in
a
po
werfu
l
com
pu
te
r
t
o
ge
t
fast
resu
lt
s,
the
n
th
e PID
par
a
mete
r value im
ple
mented
in real
hard
war
e.
Figure
4. Bl
oc
k diag
ram of
pro
posed
PID
opti
mal co
ntr
ol
3.
RESU
LT
S
AND
DI
SCUS
S
ION
Thr
ee
al
go
rith
ms
a
re
te
ste
d
and
c
ompare
d
with
the
pro
pose
d
al
gorithm
w
hich
a
re
PID
t
un
e
d
by
Zie
gler
-
Nich
ol
s (
PID Z
N), P
I
D
tu
ned
by GA w
it
h I
AE
ob
je
ct
ive f
unct
io
n
(PI
D
G
A
I
A
E)
an
d
L
QR. While
the
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
D
ri
S
ys
t,
V
ol
.
11
, N
o.
4
,
D
ecembe
r
2020
:
2164
–
2172
2168
pro
po
se
d
al
go
rithm
is
P
ID
tun
e
d
by
GA
with
L
QR
obje
ct
ive
funct
ion
name
d
as
PI
D
GA
L
QR.
Bot
h
performa
nce
and
e
ne
rgy
c
on
s
umpti
on
a
re
co
mp
a
re
d
by
us
i
ng
I
A
E
an
d
total
energ
y
co
nsu
mp
ti
on,
resp
ect
ivel
y.
B
efore
i
mp
le
me
nted
in
real
ha
rdwar
e
,
t
he
al
gorithm
is
sim
ulate
d
us
i
ng
M
A
TLAB
Si
mu
li
nk.
Af
te
r
a
good
resu
lt
is
fou
nded
in
the
sim
ulati
on
sta
ge,
the
al
gorithms
are
te
ste
d
in
real
hard
war
e
.
Two
conditi
on
te
sti
ng
is
ca
rr
ie
d
out
wh
ic
h
a
re
st
ep
res
pons
es
a
nd
s
peed
var
ia
t
ion
.
T
he
c
ontr
oller
par
a
mete
r
s
val
ue
sh
ow
n
in
Tabl
e
3.
The
val
ue
of
Q
a
nd
R
matri
x
is
t
he
s
ame
f
or
PID
GA
LQR
a
n
d
LQR.
In
this
t
est
,
the
tolerance
of
± 2
%
is
us
e
d.
Table
3.
C
on
t
r
oller
par
a
mete
r
s
Metho
d
Kp
Ki
Kd
K
LQR
PID
ZN
0
.3
1
.3
0
.3
-
PID
G
A
LQR
1
0
.4
0
.2
-
PID
G
A I
AE
2
0
.8
0
.9
-
LQR
-
-
-
[0.9
0.0
4
]
3.1. Ste
p res
p
on
ses
Step
res
pons
e
s
are
t
he
basic
t
est
ing
to
k
no
w
the
p
er
forma
nc
e o
f
the
c
on
t
r
oller
i
nclu
ding
sett
li
ng
ti
me
and
ove
rsho
ot.
Tw
o
oth
e
r
par
a
mete
rs
w
hi
ch
are
I
AE
a
nd
total
ene
r
gy
c
on
s
umpti
on
al
so
incl
ud
e
d
f
or
com
par
is
on.
Fi
gure
5
s
hows
the
ste
p
re
spo
nse
of
the
syst
em
with
a
diff
e
ren
t
c
ontrolle
r
.
Fig
ur
e
5
(a
)
i
s
the
sp
ee
d
pr
of
il
e
wh
ic
h
s
how
n
t
hat
PID
Z
N,
P
ID
G
A
L
QR,
a
nd
PID
GA
I
A
E
ha
ve
near
l
y
the
same
rise
ti
me
but
the
seco
nd
m
et
hod
ha
s
lo
w
er
over
sho
ot
without
un
dershooti
ng.
Fro
m
the
set
tl
ing
ti
me
par
am
et
er,
the
pro
po
se
d
meth
od
has
t
he
lo
w
est
se
tt
li
ng
ti
m
e.
Wh
il
e
P
I
D
GA
I
AE
is
f
as
te
r
tha
n
PID
Z
N.
Step
res
ponse
s
o
f
the
PID
meth
od
with
th
ree
di
ff
ere
nt
t
un
i
ng
met
hods
s
hown
th
at
the
G
A
LQR
tu
ning
met
hod
ha
s
t
he
bes
t
performa
nce
i
n
te
rm
s
of
set
t
li
ng
ti
me
a
nd
ov
e
rs
hoot.
T
he
la
st
method,
op
ti
mal
c
on
t
rol
us
in
g
L
QR,
has
the
best
performa
nc
e compa
red w
it
h
three
o
t
her
methods
.
(a)
(b
)
(c)
Figure
5. Sim
ul
at
ion
step
r
es
ponse
s
,
(a) S
pee
d prof
il
e
,
(
b)
I
AE pro
file
,
(c
) Ener
gy
prof
il
e
In
te
rm
s
of
I
A
E,
it
has
the
sa
me
resu
lt
a
s
se
tt
li
ng
ti
me
a
nd
over
sho
ot
paramet
er
w
hich
is
L
QR
has
the
lowe
st
IAE.
W
hile
the
pro
posed
al
gorithm
is
in
sec
on
d
place
a
fter
L
QR
w
hich
bette
r
am
ong
tw
o
oth
e
r
PI
D
tu
ning
me
thods,
as
sho
w
n
in
Fig
ur
e
5
(
b).
I
n
the
e
n
e
r
gy
point
of
vie
w,
Fi
gure
5
(c
),
P
ID
G
A
L
Q
R
has
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
PID opti
m
al c
on
tr
ol to
reduc
e ener
gy
c
onsumptio
n
i
n DC
-
dr
iv
e
syste
m
(
Ha
ri
Ma
ghfi
roh)
2169
the
lowe
st
ene
rgy
c
on
s
umpti
on.
I
n
the
be
gin
ni
ng,
L
QR
ha
s
the
lo
west
en
ergy
c
on
s
umpt
ion
.
H
ow
e
ve
r,
wh
e
n
ti
me
increase,
it
s
ener
gy
c
ons
umpti
on
al
s
o
i
ncr
ease
s.
T
his
is
du
e
t
o
integ
r
al
con
tr
ol
w
hic
h
ad
de
d
in
th
e
LQR
method t
o
el
im
inate
the stea
dy
-
sta
te
e
rror.
Table
4
s
how
s
the
detai
le
d
resu
lt
w
hich
r
esumes
of
ste
p
res
pons
e
pa
rameters
.
The
r
e
al
so
%
I
AE
and
%
Ene
r
gy
wh
ic
h
a
re
t
he
c
omparati
ve
be
tween
f
our
met
hods
with
P
ID
ZN
as
the
base
.
T
her
e
fore,
th
e
val
ue
of
%
IAE
an
d
%
Energ
y
is
100%
for
PID
Z
N.
It
is
cl
early
seen
that
the
pro
po
s
ed
met
hod
is
in
the
seco
nd
plac
e
of
%
IAE
with
lowe
r
by
17.
54
%
co
mp
a
red
with
PID
ZN
.
In
te
rm
s
of
e
ne
rgy,
it
sta
nds
in
the
lo
west
e
nerg
y
consu
mp
ti
on
w
it
h
4.7
1
%
a
nd
1.12
%
lo
wer c
ompare
d wit
h PID Z
N
a
nd L
QR r
es
pecti
vel
y.
Af
te
r
t
he
si
mu
l
at
ion
te
st,
t
he
a
lgorit
hm
is
im
plemente
d
in
r
eal
ha
rdwa
re.
The
co
ntr
oller
par
a
mete
r
is
the
sa
me
as
th
e
sim
ulati
on
te
st
with
out
fi
ne
-
tu
ning.
Fig
ur
e
6
s
hows
the
ha
rdwar
e
ste
p
r
esp
on
ses
.
Fig
ure
6
(a
)
is
the
s
pee
d
prof
il
e.
It
s
hows
that
al
l
the
al
gorith
m
perfor
ms
well
exce
pt
PID
G
A
I
AE
wh
ic
h
os
ci
ll
at
e
ar
ound
the
set
po
i
nt.
This
is
due
to
the
higher
val
ue
of
PID
paramet
ers
an
d
fi
ne
-
t
un
i
ng
did
no
t
a
pp
l
y.
T
he
oth
er
method
ca
n
pe
rform
w
el
l
an
d
track
the
set
point
with
the
s
ame
set
tl
ing
ti
me
w
hich
is
8
s.
H
oweve
r,
th
e
PID
ZN met
hod ha
s a
4
%
ov
e
rs
hoot. Whilst
, t
he pr
opos
e
d met
hod ha
s the
sam
e p
e
rformance
as LQ
R
.
In
te
rms
of
IAE,
L
QR
has
t
he
lowe
st
I
AE,
the
sa
me
resu
lt
as
the
si
mu
la
ti
on
te
st.
Wh
il
e
the
pr
opos
e
d
method
has
I
AE
highe
r
by
6.43%
a
nd
lo
wer
by
1.7
6%
co
mp
a
red
with
L
QR
a
nd
PID
Z
N
res
pecti
vely.
I
n
te
rms
of
e
nerg
y
c
onsum
ption,
L
QR
is
in
the
first
place
.
T
he
integ
ral
c
ontr
ol
a
dd
e
d
in
L
Q
R
has
a
small
eff
ect
on
t
he
ha
rdwa
r
e.
PID
G
A
L
Q
R
is
in
second
place
in
both
I
AE
an
d
e
nerg
y
p
ara
mete
rs.
It
has
lo
wer
IAE
and
energ
y
co
mp
a
r
ed
with
PID
Z
N
by
1.7
6%
a
nd
15.
65%
res
pe
ct
ively.
P
ID
GA
IAE
has
t
he
higher
IA
E
be
cause
it
s
sp
ee
ds
re
spon
s
e
has
os
ci
ll
at
ed
w
hile
it
s
energ
y
is
lo
we
r
t
han
PID
ZN
since
w
hen
it
s
s
peed
belo
w
the
set
po
i
nt
it
co
nsume
s
lo
wer
energ
y.
Table
5
res
um
e
s
t
he
res
ults
of
the
ha
rdware
impleme
ntati
on
of
ste
p
re
spo
ns
es.
Table
4.
Res
ume
of ste
p respon
s
es
par
a
mete
rs
Metho
d
S
ettlin
g
Time
(secon
d
)
% OS
IAE
% IA
E
Total Energy
Used
(
Jo
u
le)
% Energy
PID
ZN
12
49
3
0
9
.
6
2
100
3
0
7
5
.
7
100
PID
G
A
LQR
11
32
2
5
5
.
3
0
8
2
.
4
6
2
9
3
0
.
8
9
5
.
2
9
PID
G
A I
AE
13
46
3
0
1
.
8
9
9
7
.
5
0
3
0
2
9
.
2
9
8
.
4
9
LQR
8
0
2
4
0
.
6
8
7
7
.
7
3
3
0
4
1
.
3
9
8
.
8
8
(a)
(b)
(c)
Figure
6. Ha
rdwar
e
step
res
ponse
s
,
(a) S
pee
d prof
il
e
,
(
b)
I
AE pro
file
,
(c
)
Energ
y prof
il
e
Table
5.
Res
ume
of
hard
ware i
mp
le
me
ntati
on of ste
p respon
s
es
par
a
mete
rs
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
D
ri
S
ys
t,
V
ol
.
11
, N
o.
4
,
D
ecembe
r
2020
:
2164
–
2172
2170
Metho
d
S
ettlin
g
Time
(secon
d
)
% OS
IAE
% IA
E
Total Energy
Used
(
Jo
u
le)
% Energy
PID
ZN
8
4
1538
100
4
4
0
.
4
5
100
PID
G
A
LQR
8
0
1511
9
8
.
2
4
3
7
1
.
5
4
8
4
.
3
5
PID
G
A
I
AE
o
scilate
o
scilate
3161
2
0
5
.
5
3
3
8
6
.
6
3
8
7
.
7
8
LQR
8
0
1412
9
1
.
8
1
2
9
9
.
7
1
6
8
.
0
5
3.2. Spee
d
Vari
at
io
n
Resp
onses
Finall
y,
the
la
s
t
te
st
is
t
he
sp
e
ed
va
riat
ion
.
In
this
te
st,
t
he
s
peed
ref
e
ren
ce
is
va
ried
f
rom
10
0
r
pm
to
120
rpm
the
n
t
o
80
r
pm
.
Fig
ur
e
7
is
s
howi
ng
the
s
ys
te
m
respo
ns
e
wh
ic
h
is
s
peed
pr
ofi
le
,
IA
E
pro
file
,
an
d
energ
y
pro
file
.
S
peed
prof
il
e
sh
ows
near
l
y
t
he
same
res
ult
with
ste
p
res
ponse
w
hich
PID
G
A
I
AE
os
c
il
la
te
s
arou
nd
sp
e
ed
r
efere
nce.
The
o
the
r
th
ree methods
can
trac
k
the
s
peed
va
ria
ti
on
w
el
l. In
te
rms
of I
AE
, F
igure
7
(b),
PID
G
A
I
AE
has
a
high
er
IA
E
becau
s
e
the
sp
ee
d
res
pons
e
is
os
ci
ll
at
ing
.
LQR
ha
s
the
lo
west
I
AE,
the
n
fo
ll
owe
d
by
P
I
D
G
A
L
QR,
a
nd
PID
Z
N.
In
energ
y
co
nsu
mp
ti
on,
Fig
ur
e
7
(c
),
it
is
see
n
that
L
QR
ha
s
th
e
lowest
e
nerg
y
co
nsum
ption,
w
hile
the
pro
po
s
ed
meth
od
is
in
s
eco
nd
place
be
tt
er
th
an
P
I
D
GA
I
AE
a
nd
PI
D
Z
N.
(a
)
(b
)
(c)
Figure
7. Ha
rdwar
e
sp
ee
d va
r
ia
ti
on
r
es
ponse
s
,
(a
)
S
pee
d va
riat
ion
pro
file
,
(b) IAE
pr
of
il
e
,
(c
)
E
nerg
y prof
il
e
The
qu
a
ntit
at
ive
re
su
lt
of
IAE
an
d
e
nerg
y
is
in
Ta
ble
6.
PI
D
Z
N
as
the
base
c
ompa
rison.
L
QR
i
s
su
pe
rio
r
w
hic
h
has
the
lo
wes
t
bo
th
IAE
an
d
ene
r
g
y
c
ons
umpti
on.
T
he
pro
po
se
d
met
hod
is
in
seco
nd
place
after
L
QR
both
in
I
AE
an
d
energ
y
c
on
s
umpti
on.
Com
pa
rin
g
L
QR
a
nd
P
I
D
GA
L
Q
R,
the
seco
nd
on
e
ha
s
adv
a
ntage
s w
hi
ch
are
only
use
d
on
e
se
ns
or
i
n
this case, w
hi
le
LQR
usi
ng two
se
nsor
s
f
or
eve
ry
sta
te
w
hi
ch
i
s
sp
ee
d
se
nsor
a
nd
c
urren
t
sen
so
r
.
T
he
ot
her
ad
va
ntages
of
the
pro
posed
method
are
,
P
ID
is
m
ore
fa
mil
ia
r
us
in
g.
The
refo
re,
it
eas
y
to
be
impleme
nted
in
the
e
xisti
ng
s
ys
te
m
with
ou
t
a
l
ot
of
c
ha
ng
e
s.
T
he
tra
de
-
off
betwee
n
pe
rfo
rma
nce
a
nd
e
nerg
y
ca
n
be
adj
us
t
ed
by
set
ti
ng
wei
ghte
d
val
ue
i
n
obje
ct
ive
f
unct
ion
corres
pondin
g wit
h
mat
rix
Q
dan
R
in t
he
L
QR met
hod.
Table
7
resum
es
the
ene
rgy
and
IS
E
redu
ct
ion
f
rom
ha
r
dw
a
re
te
sti
ng.
The
pr
opos
e
d
al
go
rit
hm
reduces
bo
t
h
I
AE
an
d
e
nergy
con
s
umpti
on
com
par
e
d
wit
h
PI
D
Z
N.
T
he
aver
a
ge
of
al
l
the
res
ults
is
IAE
and
energ
y
re
duct
ion b
y 2.7
6
%
a
nd 16.0
7
%
res
pecti
vely
.
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
PID opti
m
al c
on
tr
ol to
reduc
e ener
gy
c
onsumptio
n
i
n DC
-
dr
iv
e
syste
m
(
Ha
ri
Ma
ghfi
roh)
2171
Table
6.
Res
ume
of
hard
ware i
mp
le
me
ntati
on of s
pee
d var
ia
ti
on
r
es
ponse
s p
a
rameters
Metho
d
IAE
% IA
E
Total Energy
Used
(
Jo
u
le)
% Energy
PID
ZN
2638
100
1
3
4
4
.
4
0
100
PID
G
A
LQR
2539
9
6
.
2
5
1
1
2
2
.
7
3
8
3
.
5
1
PID
G
A I
AE
7152
2
7
1
.
1
1
1
1
8
3
.
5
5
88
LQR
2488
9
4
.
3
1
9
2
8
.
2
2
69
Table
7.
Res
ume
of
IAE a
nd en
e
rgy red
ucti
on of PI
D GA
LQR c
ompare
with P
ID Z
N
Data so
u
rce
% IA
E
redu
ctio
n
% Energy
r
ed
u
ctio
n
Table 5
1
.76
1
5
.65
Table 6
3
.75
1
6
.49
Av
erag
e
2
.76
1
6
.07
4.
CONCL
US
I
O
N
The
pro
po
se
d
al
gorithm
is
si
mp
le
a
nd
ea
sy
to
be
i
mp
le
m
ented
si
nce
it
bases
on
a
we
ll
-
know
n
al
gorithm,
PID
.
It
was
su
cces
sfu
ll
y
im
pleme
nted
bo
t
h
in
t
he
simulat
ion
a
nd
ha
rdwar
e
s
ys
te
ms.
The
te
st
was
carried
out
with
tw
o
sc
hem
es
w
hich
a
re
ste
p
re
spo
ns
es
a
nd
s
peed
va
riat
ion
res
pons
e
s.
The
res
ult
sho
ws
that
the
pr
opos
e
d
al
gorithm
is
be
tt
er
in
bo
t
h
I
AE
a
nd
ene
r
gy
c
on
s
umpti
on
com
pa
red
wit
h
oth
e
r
P
ID
tun
i
ng
methods
wh
ic
h
a
re
Zie
gle
r
–
Nic
ho
ls
,
a
nd
GA
with
I
AE
ob
je
ct
ive
f
unct
ion
.
C
ompare
d
with
PID
ZN
,
it
has
IA
E
a
nd
e
nerg
y
re
duct
ion
by
2.76
%
an
d
16
.07
%
re
sp
ect
iv
el
y.
Alt
hough
it
s
per
f
orma
nc
e
is
lowe
r
tha
n
the
LQR,
it
has
ot
her
a
dv
a
ntage
s
that
us
e
fewer
sen
sors,
w
hi
le
LQR
us
es
sens
or
s
f
or
e
ve
ry
sta
te
.
T
he
oth
e
r
adv
a
ntage
of
the
pro
pose
d
m
et
hod
is,
P
ID
i
s
more
famil
ia
r
us
i
ng.
The
refor
e
,
it
easy
to
be
imple
me
nte
d
in
the
existi
ng syst
em w
it
ho
ut a l
ot of c
hanges.
REFERE
NCE
S
[1]
A.
Far
am
ar
zi a
n
d
K.
Sab
ahi
,
“Recurre
nt
Fuz
zy
Neura
l
Ne
twork
f
or
DC
-
mot
or
co
ntrol
,
”
in
2011
F
if
th
Int
ernati
ona
l
Confe
renc
e
on
Gene
tic
and
Ev
o
lut
ionary
Computing
,
2011,
pp.
9
3
–
96
,
2011
.
[2]
U.
K.
Bansal
an
d
R.
Narve
y
,
“S
pee
d
Contro
l
of
DC
Motor
Us
ing
Fuzzy
PID
Control
ler,”
Adv.
El
e
ct
ron.
E
lectr.
Eng.
,
vol
.
3
,
no
.
9,
pp
.
1209
–
122
0,
2013
.
[3]
C.
A.
Kumar
,
B
.
R.
Hari
j
an,
M.
K.
Kumar
,
and
M.
Bha
r
at
hi
,
“B
LDC
Motor
Spe
ed
Contro
l
using
Fuzzy
Logic
PI
D
Control
le
r
and
C
ompa
ring
It
Wit
h
PI Controlle
r
,
”
Int. J. Eng. Ad
v. Tec
hnol
.
,
vol
.
9
,
no.
2,
pp.
4917
–
4922,
2019
.
[4]
A.
Hughes a
nd
B.
Drury,
Elec
tri
c
Motor
and
Dr
i
ve
s: F
undam
ental,
Typ
es
and
Ap
pli
catio
ns
,
5
th e
d.
E
lsevier, 2019
.
[5]
J.
De
Ol
iveira
,
A.
Nied
,
M.
H.
F.
Santos,
and
R
.
P.
Dias,
“Stud
y
on
the
Ene
rgy
Eff
i
cienc
y
of
S
oft
Star
ti
ng
of
a
n
Induc
ti
on
Motor
with Torque Co
ntrol
,
”
in
Ad
vasc
es
in
Mo
tor Torque
Control
,
In
te
c
h,
2011
,
pp
.
33
–
46
,
2011
.
[6]
S.
N.
Al
-
bar
go
th
i,
G.
M.
Qaryout
i,
and
Q.
M.
Jab
er,
“Spe
ed
cont
r
ol
of
DC
mot
or
using
conv
ent
io
nal
and
ad
apt
iv
e
PID
cont
rollers,
”
Indone
s
.
J
.
Ele
ct
r. Eng.
Comput.
S
ci.
,
vol. 16, n
o.
3
,
pp
.
1221
–
1
228,
2019
.
[7]
S.
M.
Rakh
ta
l
a
and
E.
S.
Rou
dbar
i,
“Applica
t
ion
of
PEM
Fu
el
Ce
ll
for
Stan
d
-
al
one
Based
on
a
Fuz
zy
PI
D
Control
,
”
Bu
ll. E
le
c
tr. Eng. Inf
or
matic
s
,
vo
l. 5, n
o.
1
,
pp
.
45
–
61
,
2016.
[8]
K.
H.
Ang,
G
.
Chong,
S.
Me
m
ber
,
and
Y.
Li,
“PID
Control
S
ystem
Analysis
,
Design
,
and
T
ec
hnology
,
”
I
EEE
Tr
ans.
Control
S
yst.
Te
chnol.
,
vo
l.
13
,
no
.
4
,
pp
.
5
59
–
576,
2005
.
[9]
A.
Bal
estr
ino,
A.
Cai
t
i,
V.
C
alabró,
E
.
Crisost
omi
,
and
A.
L
a
ndi,
“From
B
asi
c
to
Advanc
ed
PI
Control
le
rs :
A
Compl
exity
vs
.
Perform
ance
Co
mpa
rison,
” in
Ad
vanc
e
in
PID
Co
ntrol
,
In
tech, pp. 87
–
100
,
201
1
.
[10]
A.
Y.
Jae
n
-
cue
l
l
ar,
R
.
de
J.
Rome
ro
-
Tron
coso,
L.
Mora
le
s
-
velaz
que
z,
and
R.
A.
Os
ornio
-
rios,
“PID
-
Control
ler
Tuni
ng
Opt
im
i
z
at
ion
with
G
ene
t
ic
Algor
it
hms
in S
erv
o
Sys
te
ms,”
Int. J. Adv
.
R
ob
ot.
S
yst.
,
vo
l. 10, pp. 1
–
14,
2013.
[11]
J.
M.
Herr
ero
,
X
.
B
l
asc
o
,
M.
Ma
rti
ne
z,
and
J
.
V
Salc
edo
,
“Optim
al
PID
Tun
ing
with
Gen
et
i
c
Al
gorit
hm
for
Non
-
Li
ne
ar
Proc
ess Model
s,”
in
15
th
Tr
ie
nial World Congr
ess
,
pp.
31
–
36
,
2002
.
[12]
A.
A.
M
.
Za
hir
,
S.
S.
N.
Alhady
,
W.
A
.
F.
W
.
Oth
ma
n,
and
M.
F
.
Ahmad,
“Ge
ne
tic
Algori
thm
Opt
im
izati
on
of
PID
Control
le
r
for
Br
ushed
DC Motor
,
”
In
tell. Manuf. Mec
hatroni
cs
,
p
p.
427
–
437
,
201
8.
[13]
C.
Li
n
,
S.
Mem
ber
,
H.
Jan,
and
N.
Shieh
,
“GA
-
Based
Multi
ob
j
ec
t
ive
PID
Cont
rol
for
a
L
ine
ar
Brushless
DC
Motor,
”
I
EEE/ASME
Tr
ans.
Me
c
hatronic
s
,
vo
l. 8
,
no.
1,
pp.
56
–
65
,
2003
.
[14]
B.
Naga
r
aj
and
N.
Murugana
nth
,
“A
Co
mpa
ra
ti
v
e
Study
of
PID
Control
le
r
Thn
in
g
Us
ing
GA
,
EP,
PS
O
and
ACO
,
”
in
2010
In
t. Con
fe
renc
e
on
Comm
unic
ati
on
Cont
rol an
d
Computi
ng
Technol
og
ie
s
,
pp
.
305
–
313
,
2
010
.
[15]
U.
Ans
ari
,
S.
Al
am
,
and
S.
M.
u
n
N.
J
afr
i,
“Mo
del
ing
and
Cont
rol
of
Thr
ee
Pha
se
BLDC
Motor
using
PID
wi
th
Gene
tic
Algori
thm
,
”
in
201
3
UKSim
13th
Inte
rntiona
l
Confe
renc
e
on
Mode
lling
and
Simulat
i
on
,
pp.
189
–
194
,
20
11
.
[16]
V.
Vishal,
V.
Kumar
,
K.
P.
S
.
Ran
a,
and
P.
Mishra,
“
Co
mp
ara
t
ive
Study
of
Some
Opti
mi
z
at
ion
Techni
que
s
Applie
d
to
DC
Motor
Cont
rol,
”
in
2014
IEE
E
Inte
rnat
i
onal
Adv
an
ce
Computing
Co
nfe
renc
e
(IA
CC
)
,
pp.
1342
–
1347
,
2014
.
[17]
M.
A.
Ibra
h
im,
A.
K.
Mahmood
,
and
N
.
S.
Sul
tan,
“Opt
im
a
l
PID
cont
rol
le
r
of
a
b
rushless
DC
motor
using
geneti
c
al
gorit
h
m,”
In
t.
J.
Powe
r
El
e
ct
ro
n.
Dr
iv
e
Syst
.
,
v
ol.
10
,
no
.
2
,
pp
.
822
–
830,
2019
.
[18]
A.
Mirza
l
,
S.
Yos
hii
,
and
M
.
Furukawa,
“
PID
Para
me
te
rs
Op
ti
mization
by
Us
ing
Gene
tic
Algorit
h
m,”
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
D
ri
S
ys
t,
V
ol
.
11
, N
o.
4
,
D
ecembe
r
2020
:
2164
–
2172
2172
ISTECS
J.
,
2012
.
[19]
S.
K.
Suman
an
d
V.
K.
Gir
i,
“
Gene
tic
Algori
t
hms
Techni
ques
Based
Opti
mal
PID
Tuni
ng
For
Speed
Contro
l
of
DC Motor,
”
Am. J
.
Eng
.
Te
chnol.
Manag.
,
vol
.
1
,
no.
4
,
pp
.
59
–
64
,
2016.
[20]
H.
Zha
ng
,
Y
.
C
ai
,
and
Y.
Ch
en
,
“Par
amete
r
Optim
i
za
t
ion
of
PID
Control
l
ers
B
ase
d
on
Gen
etic
Algorit
hm
,
”
in
2010
Inte
rnat
ion
al
Conf
ere
nce o
n
E
-
Heal
t
N
et
wo
rking,
Dig
it
al
E
c
osystem
and
Te
c
hnologi
es
,
pp.
4
7
–
49
,
2010
.
[21]
A.
A.
M.
Z
ahi
r
,
S.
S.
N.
Alhady
,
W
.
A.
F.
W.
Ot
hma
n,
A.
A.
A.
Wa
hab
,
and
M.
F.
Ahmad
,
“Obj
ec
t
ive
func
ti
ons
modi
ficat
ion
of
GA
opti
mized
PID
cont
rol
le
r
for
brushed
DC
mo
tor,
”
Int
.
J.
Elec
tr.
Comput.
Eng
.
,
vol
.
10
,
no
.
3
,
pp.
2426
–
2433
,
2020.
[22]
S.
L. Brunt
on
an
d
J.
N.
Kutz
,
Dat
a
-
Dr
iv
en
Sc
ie
n
c
e
and
Engi
n
e
eri
ng
,
1st
ed. Camb
ridge
Univ
ersit
y
Press
,
2019.
[23]
H.
Maghfir
oh
,
C
.
Her
ma
nu,
M.
H.
Ibra
h
im
,
M.
Anw
ar,
and
A.
Ram
elan,
“Hybr
id
fuz
zy
-
PID
li
k
e
opt
im
a
l
cont
ro
l
to
red
uce e
n
erg
y
c
onsumpti
on,
”
T
ELKOMNIKA
,
v
ol.
18
,
no
.
4
,
pp
.
2053
–
2061,
202
0.
[24]
H.
Maghfir
o
h,
O.
W
ahyunggor
o,
A
.
I
.
Cahy
adi,
and
S.
Pr
apt
od
iyono,
“PID
-
Hybrid
Tuni
ng
to
I
mprove
Control
Perform
ance in
Speed
Contro
l
o
f
DC Motor
B
ase
on
PLC
,
”
in
3
rd ICA
,
2013
,
pp.
233
–
238.
[25]
A.
Abdulam
e
er,
M.
Sulai
m
an,
M.
S.
M.
Aras,
and
D.
Sal
ee
m
,
“GUI
Based
Con
trol
Sys
te
m
An
al
ysis
Us
ing
PID
Control
le
r
for
E
duca
t
ion,
”
Indon
es.
J
.
Elec
tr.
En
g.
Comput.
Sc
i.
,
vol.
3
,
no
.
1
,
pp
.
91
–
101,
2016
.
[26]
G.
Li
n
and
G
.
L
iu,
“
Tuni
ng
PID
Control
ler
Us
ing
Adapti
ve
Gen
et
i
c
Algorit
h
ms,
”
in
5th
In
te
rnat
ional
Confe
ren
c
e
on
Computer
Sc
i
enc
e
and
Educ
at
ion
,
pp
.
519
–
523
,
2010
.
[27]
L.
Fan
and
E
.
M
.
Joo,
“De
sign
fo
r
Auto
-
tuni
ng
PI
D
Control
l
er
B
a
sed
on
Gen
et
i
c
Algorit
hms,
”
in
IEE
E
Confe
r
ence
on
Industrial Elect
ronics
and
Ap
pli
cations
,
pp
.
1
924
–
1928
,
2009
.
[28]
M.
Gani,
S.
Isl
a
m,
and
M
.
A.
Ulla
h
,
“Optimal
PID
tuni
ng
for
co
ntrol
li
ng
the
temper
a
ture
of
el
e
ct
ri
c
furna
ce
by
gene
t
ic
al
gor
it
h
m,
”
SN
Appl
.
Sc
i
.
,
vol. 1, no. 8, p
p.
1
–
8
,
2019
.
[29]
L.
B
.
Prasad
,
B
.
Tya
g
i,
and
H.
O.
Gupta
,
“Opt
i
ma
l
Con
trol
of
Nonline
ar
Inve
rt
ed
Pendulum
D
ynam
i
ca
l
Sys
tem
with
Disturb
ance
Inpu
t
using
PID
Controller
and
LQR,
”
in
2011
IEE
E
In
t.
Con.
o
n
Control
S
ystem
,
Computing
a
nd
Engi
ne
ering
,
pp.
540
–
545
,
2011
.
[30]
K.
Ogat
a,
Mode
r
n
Control Engin
ee
ring
,
5th
ed
.
N
ew
Jersey:
Prent
ic
e
Hall,
2010
.
Evaluation Warning : The document was created with Spire.PDF for Python.