TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 13, No. 1, Janua
ry 201
5, pp. 114 ~
123
DOI: 10.115
9
1
/telkomni
ka.
v
13i1.700
8
114
Re
cei
v
ed Se
ptem
ber 20, 2014; Revi
se
d No
vem
ber
10, 2014; Accepted Decem
ber 8, 201
4
Improvement of Fuzzy Based Practical Controller for
Continuous Motion Control
Purtojo
*1,2
, Heru S. B. Rochardjo
2
, Ge
sang Nugro
h
o
2
, Herianto
2
1
Dept.of Mech
anic
a
l Eng
i
n
e
e
r
ing, Un
iversita
s Islam Indone
sia,
Jl. Kaliur
ang k
m
14.5 Yog
y
a
k
arta, Indon
esia
5528
4
2
Dept.of Mech
anic
a
l an
d Indu
strial Eng
i
ne
eri
ng, Univ
ersitas
Gadjah Ma
da,
Jl. Grafika 2, Yog
y
ak
arta, Ind
ones
ia 5
528
1
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: purtojo
@ui
i
.a
c.id
A
b
st
r
a
ct
T
h
is articl
e pr
e
s
ents a
deve
l
o
p
ment of
a fu
zz
y
bas
e
d
n
o
m
i
nal c
har
acteris
t
ic trajectory fo
llow
i
n
g
(NCT
F
)
control
l
er for co
ntinu
o
u
s motio
n
cont
rol. A new
stru
cture is pro
pos
ed i
n
ord
e
r to
achi
eve exc
e
ll
ent
perfor
m
a
n
ce o
f
tracking to a contin
uo
us referenc
e
inp
u
t and als
o
for poi
nt-to-poi
nt posi
t
ioni
ng task. Th
e
prop
osed struc
t
ure ma
inta
ins
the NC
T
F
controll
er simpl
e
config
uratio
n
w
h
ich is comp
osed of a no
mina
l
character
i
stic trajectory (NCT
) and a co
mp
e
n
sator.
T
he co
mp
ens
ator is b
a
sed o
n
a Ma
md
an
i type fu
z
z
y
control
l
er. Its
me
mbers
h
ip fu
nctions ar
e de
sign
ed acc
o
rdi
ng to the ava
i
l
abl
e infor
m
ati
o
n provi
ded
by NC
T
and th
e hardw
are spec
ificati
o
n.
Control
l
er p
e
rformanc
e w
a
s evalu
a
ted
thr
oug
h si
mul
a
tio
n
by co
mp
arin
g it
w
i
th the ex
isti
ng
metho
d
pr
evio
usly
prop
osed
fo
r th
e
fu
zz
y
bas
ed
NCT
F
control
l
e
r. T
he track
i
ng
perfor
m
a
n
ce w
a
s eva
l
uate
d
b
y
me
asuri
ng r
e
spons
es of
the
system pr
ovid
i
ng co
ntinu
ous
sinus
oid
a
l si
gn
a
l
inp
u
t. T
he res
u
lt in
dicat
e
s th
at substa
ntial
i
m
pr
ove
m
ent
is
achi
eve
d
in
trackin
g
of co
nti
nuo
us refer
enc
e
inp
u
ts. Moreov
er, a better result is also o
b
tai
ned i
n
perfor
m
i
ng po
int-to-p
oi
nt positio
ni
ng task.
Ke
y
w
ords
: nctf, fuz
z
y control,
tracking, traj
ectory following,
continuous motion
Copy
right
©
2015 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
Motion
control syste
m
s
are appli
ed in
variou
s
ind
u
strial
appli
c
ati
ons su
ch as machi
n
e
tools, rob
o
tic system
s an
d measuri
ng
machi
n
e
s
. Industri
a
l ma
ch
ines a
r
e req
u
ired to pe
rf
orm
high pe
rform
ance tasks
such a
s
po
siti
oning, tr
a
c
kin
g
and
conto
u
ring. Th
ese
tasks are of
te
n
use
d
to evalu
a
te perfo
rma
n
ce
of su
ch
systems.
In o
r
der to reali
z
e
high pe
rform
ance for all t
h
e
types of motions, a ge
neral-pu
r
p
o
se se
rvo cont
ro
lle
r is desi
r
ed. T
he co
ntrolle
r i
s
also expe
cted
to satisfy su
ch requi
reme
nts as hi
gh
accuracy
, fast resp
on
se a
nd rob
u
st. F
o
r moto
r se
rvo
system
s, the
po
sition tra
cki
ng
pre
c
i
s
i
on i
s
con
s
id
ered
a
s
the
main in
dex
of evaluatio
n
[1].
Inaccurate tracking m
a
y lead to a fail
u
r
e of a
sy
ste
m
. More
over,
the practi
cal
cont
rolle
r de
sign
method is al
so requi
re
d for practi
cal ap
p
lication
s
.
Variou
s
cont
rollers h
a
ve b
een d
e
velop
e
d
such
a
s
a
PID co
ntrol
system which
provide
s
the si
mple
st
and yet
mo
st efficient
sol
u
tion to m
any
control p
r
o
b
le
ms [2,
3]. Various PID tu
ni
ng
have been p
a
tented an
d many softwa
r
e packa
ge
s are
availa
ble.
However, accurate mod
e
l
of
the system is indispe
n
sabl
e to
perform
satisfa
c
to
ry perform
an
ce. More
over, co
mplexity of the
model in
crea
se
s du
e to th
e different b
e
haviour
of the macro
-
regi
on to the mi
cro-regi
on. T
w
o
-
step controll
e
r
s have b
een
extensiv
ely st
udied in o
r
de
r to overcom
e
the probl
em, particula
rly the
obje
c
t param
eters
differe
n
c
e of that two regi
on
s [3
]. Since fu
zzy
control is l
e
ss sen
s
itive to the
cha
nge
s
of p
a
ram
e
ters, fu
zzy
co
ntrol
a
ppro
a
ch i
s
co
mbined
with
PID control.
The le
ss a
ccurate
model is then
required. Th
us, it offers a
d
vant
age
s in pra
c
tical a
ppl
ication
s
and i
m
pleme
n
tatio
n
of fuz
z
y
c
o
ntrol [4-6].
For
p
r
a
c
tical appli
c
ation
s
, a
nomi
nal ch
ara
c
teri
stic
t
r
ajecto
ry
follo
wing (NCT
F) controlle
r
has
bee
n p
r
o
posed [7]. It has
advanta
g
e
s of
simpl
e
stru
cture an
d
ease of d
e
si
gn p
r
o
c
ed
ure
s
.
The stru
ctu
r
e
con
s
ist
s
of
a nomin
al cha
r
a
c
teri
s
t
ic tr
a
j
ec
to
r
y
(N
C
T
)
a
n
d
a PI c
o
mp
en
sa
to
r
(NCTF
-
PI). NCT i
s
con
s
tructed from
a simp
le o
pen loo
p
experim
ent and
PI compen
sator
para
m
eters
a
r
e ta
ken
fro
m
NCT i
n
form
ation. Howe
ver, the
drawb
a
ck of th
e
NCTF
-PI is i
n
the
pro
c
e
ss
of decidi
ng PI pa
ramete
rs. Sin
c
e an
unlim
it
ed co
mbin
ation of PI para
m
eters, this l
ead
to desig
ner ju
dgeme
n
t or trial and erro
r approa
ch [8].
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Im
provem
ent of Fuzzy Ba
sed Pra
c
tical
Contro
lle
r for Contin
uou
s
Motion Co
ntrol (Purtojo
)
115
NCT
F co
ntrol
l
er with fuzzy
compe
n
sato
r (NCTF
-
Fu
zzy) ha
s bee
n
propo
se
d to repla
c
e
PI compe
n
sator. The i
n
tenti
on is to
avoid
tr
ial and
erro
r app
ro
ach a
nd ma
ke the
NCT
F controller
more
pra
c
tical. For poi
nt-to-point p
o
siti
oning ta
sk, NCT
F-F
u
zzy controller
is effective
since
it
gives a simil
a
r re
sp
on
se
comp
ared wit
h
the NCT
F-PI controlle
r [9]. Howeve
r, the perfo
rma
n
ce
of NCTF
-
Fu
zzy cont
rolle
r perfo
rming
co
ntinuou
s moti
on tasks
requ
ire dee
p inve
stigation
s
.
This study prese
n
ts
the
in
vestigation of
NCTF-Fu
zzy
cont
rolle
r d
e
a
ling
with
co
ntinuou
s
motion tasks.
The
con
c
ept
of the NCTF
cont
rolle
r is
reviewe
d
an
d
the pro
p
o
s
ed
desi
gn meth
od
is explaine
d in se
ction 2. In
se
ction 3, the co
ntrolle
d
object is
de
scribe
d and i
t
s controlle
r is
desi
gne
d. The performan
ce evaluation
throug
h si
mu
lation is then
presented i
n
sectio
n 4 by
comp
ari
ng wi
th the existin
g
method. Fi
nally, t
he con
c
lu
sion a
nd f
u
ture
works
are d
e
scri
be
d in
se
ction 5.
2. NCTF
Con
t
roller
Con
c
e
p
t and desi
gn proce
dure of the NCTF c
ontrol
system have b
een explain
e
d
in [7
-
9]. It is co
mprised
of three
step
s: (i
) Th
e
co
ntro
lle
d ob
ject i
s
d
r
iven
with an
op
en-loop
step i
n
p
u
t
and its
displa
ceme
nt and
velocity are m
easure
d
. (
ii)
Usi
ng the
displacement a
n
d
velocity of the
mechani
sm
during the decel
e
ration, the NCT i
s
constructed on the ph
ase-plane. (iii)
T
he
comp
en
sato
r is then de
sig
ned u
s
ing the
open-l
oop
re
spo
n
se and t
he NCT information.
The structu
r
e
of fuzzy ba
sed
NCT
F control
sy
stem
con
s
ist
s
of an NCT and
a fuzzy
comp
en
sato
r. Figure 1
sh
ows the p
r
op
ose
d
fu
z
z
y
b
a
se
d N
C
TF
cont
r
o
l sy
st
e
m
.
The st
ru
ct
ure
slightly differe
nt in fee
d
ing
the in
puts.
Th
e first inp
u
t is the e
r
ror an
d
se
co
nd i
s
th
e obj
ect
moti
on
in the o
r
igin
al NCTF
structure in
stea
d of the e
r
ror rate in th
e propo
se
d
stru
cture. Thi
s
i
s
con
s
id
erin
g that in the co
ntinuou
s moti
on, cont
rolle
rs usually act
near the ref
e
ren
c
e a
nd the
ac
tion far from the referenc
e is
not important [10].
x
x
e
r
r
Figure 1. The
stru
cture of
NCT
F
co
ntrol
system with f
u
zzy comp
en
sator fo
r co
ntinuou
s motion
The typical NCT is sho
w
n in Figure 2. Im
porta
nt NCT param
eters are maximu
m erro
r
rate, h
max
, NCT maximum e
rro
r, A, and gradie
n
t at orig
in, m.
Figure 2. Typical NCT an
d its para
m
eters
0
0
Er
r
o
r
E
rro
r ra
t
e
h
max
A
m
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 13, No. 1, Janua
ry 2015 : 114 –
123
116
2.1. Compen
sator S
t
ruc
t
ure
The
stru
cture
of the fu
zzy
comp
en
sato
r
is sh
o
w
n i
n
F
i
gure
3. Th
e fuzzy co
mpe
n
s
ator is
a Mamdani t
y
pe fuzzy co
mpen
sato
r wi
th two inputs, u
p
which is the differen
c
e bet
ween t
h
e
obje
c
t motion
and NCT, an
d u
i
whi
c
h is t
he integral of u
p
. The outpu
t is the contro
l signal u.
Figure 3. The
stru
cture of the
Mamd
ani type fuzzy co
mpen
sato
r
2.2. Rule De
sign
Con
s
tru
c
tion
of the rule b
a
s
e i
s
de
sign
e
d
acco
rdin
g to the obje
c
t
motion in rea
c
hin
g
and
followin
g
NCT as sho
w
n
in Figure 4. The re
aching
phase re
gio
n
is wh
en th
e obje
c
t motion
approa
che
s
the NCT and
the following
pha
se re
gion
is when the
obje
c
t follows the NCT wit
h
in
boun
ded spe
c
ified
a
c
cura
cy.
Figure 4. The
controll
ed ob
ject motion
Based
o
n
th
e obj
ect
moti
on in
rea
c
hin
g
an
d follo
wi
ng the
NCT,
the fu
zzy ru
les
are
summ
ari
z
ed
as sho
w
n in
Table 1.
Table 1. Fu
zzy rule base
u
i
N
Z
P
u
p
N N
N
Z
Z N
Z P
P Z
P
P
2.2. Member
ship Func
tion Design
1In the fuzzifi
cation p
r
o
c
e
s
s, cri
s
p si
gnal
u
p
and u
i
is converted into
fuzzy varia
b
le
s. The
membership func
tion of input u
p
is sho
w
n in Figure 5(a).
u
p
Fuzz
ification
I
n
f
e
r
e
nce
Me
cha
n
i
s
m
De
fuzzi
ficati
on
u
Kno
w
l
e
dg
e
Base
p
u
e
e
u
p
(+
)
NC
T
u
p
(-
)
e
1
h
A
e
0
R
eac
hi
ng phas
e
r
egi
on
Fol
l
o
w
i
n
g
ph
as
e
r
e
g
i
on
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Im
provem
ent of Fuzzy Ba
sed Pra
c
tical
Contro
lle
r for Contin
uou
s
Motion Co
ntrol (Purtojo
)
117
p
u
p
u
(a) Me
mbe
r
s
h
ip of u
p
p
u
p
u
(b) Me
mbe
r
s
h
ip of u
i
p
u
p
u
(c) Memb
ership of u
Figure 5. The
membershi
p
function
s de
si
gn
There are three triang
ular-sha
ped mem
bership
fun
c
ti
ons which are Negative (N), Zero
(Z) a
nd Po
sitive (P). Signal
u
p
only varie
s
in the
rang
e of ±h
max
. T
he mem
bership fun
c
tion o
f
u
p
Zero
(Z) h
a
s t
he value of error
rate of se
nso
r
re
sol
u
tio
n
, h
s-res
.
The me
mbe
r
ship fu
nctio
n
of input u
i
i
s
sho
w
n i
n
Figure 5
(
b).
There a
r
e al
so th
ree
triangul
ar-sha
ped m
e
mbe
r
ship fu
nctio
n
s
which a
r
e
N, Z an
d P. T
he rang
e of t
h
is in
put i
s
±u
i-ma
x
cal
c
ulate
d
ba
sed o
n
the followin
g
equati
on:
max
0
max
5
.
0
Ah
de
u
u
A
p
i
(
1
)
In the follo
wi
ng p
h
a
s
e, ob
ject motio
n
o
scill
ates withi
n
±
se
nsor resol
u
tion, a
s-r
e
s
. Thus
the membership func
tion of u
i
Zero (Z
) can be sim
p
lified ba
sed o
n
the followi
ng e
quation:
res
s
res
s
res
s
res
s
i
h
a
h
a
u
res
s
4
(
2
)
In the d
e
fuzzification
pro
c
ess, fuzzy va
riable
s
are
converted
into
a
cri
s
p
si
gn
al u. T
he
membe
r
ship
function
of o
u
tput u a
r
e
sho
w
n
i
n
Fi
gure
5(c). T
here
are three mem
bership
function
s whi
c
h a
r
e z-sha
p
ed N, si
ngleto
n
Z and
s-sh
aped P. The
range of fu
zzy
variable o
u
tp
u
t
is ±u
r
, whic
h is
the rated motor input.
The fuzzy out
put variabl
e, u, is co
nverte
d to
the crisp control outp
u
t using
ce
nter
of are
a
(COA) defu
zzification meth
od. The
com
b
ination
of CO
A method, z-sha
ped a
nd
s-sh
ape
d ma
kes
the crisp o
u
tp
ut never rea
c
h the m
a
ximu
m rate
d moto
r inp
u
t. Ho
we
ver, sin
c
e
the
co
ntrolle
rs a
c
t
near the
reference, maximum cont
rol si
gnal value m
a
y not nece
s
sary.
3. Results a
nd Analy
s
is
3.1. Sy
stem
Des
c
ription
In order to
e
v
aluate the
e
ffectiveness
of t
he
pro
p
o
s
ed
controlle
r d
e
si
gn, a
serie
s
of
simulatio
n
is ca
rrie
d
out
based o
n
th
e dynami
c
m
odel of the
e
x
perime
n
tal li
near
po
sitioni
ng
system a
s
a controlled o
b
j
e
ct as
sho
w
n in Fi
gure 6
(
a). The
syst
em con
s
i
s
ts
of a DC mot
o
r
cou
p
led
with
ball
s
cre
w
to
drive
a li
ne
ar
slid
e
movi
ng tabl
e. Dy
namic mo
del
of the
sy
ste
m
is
derived a
c
co
rding to the informatio
n fro
m
NCT to co
nstru
c
t sim
p
lified model of the syste
m
[7].
s
s
K
s
G
(
3
)
Whe
r
e
r
u
h
K
max
and
m
.
Followi
ng the
pro
c
edu
re gi
ven in [7], th
e sy
stem i
s
d
r
iven by a st
epwi
s
e in
put and the
system re
spo
n
se
s
a
r
e
m
e
a
s
ured as sh
o
w
n
in Figu
re
6(b
)
. The
ste
p
wi
se in
put value i
s
the
rat
ed
motor input, u
r
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 13, No. 1, Janua
ry 2015 : 114 –
123
118
(a) T
he controlled obje
c
t
(b)
Re
spo
n
se
to a stepwi
s
e
input
(c) NCT of the controlled o
b
ject
Figure 6. Single-axi
s
linea
r positio
ni
ng
system a
s
th
e controlled o
b
ject
The
NCT
of the co
ntroll
ed obj
ect i
s
then con
s
tru
c
ted a
c
cordi
ng to the d
a
t
a within
deceleration rang
e,
as sh
own
in Figu
re
6(c).
Requi
red p
a
ra
mete
rs to
con
s
tru
c
t the sim
p
lified
model an
d to desig
n the compen
sato
r
are m, A, a
s-res
, h
ma
x
and h
s-res
. The simplified dyna
mic
model of the
controlled
obj
ect de
rived from NCT
info
rmation is
cal
c
ulated a
c
cord
ing to equ
atio
n
(3) i
s
,
2152
.
0
2152
.
0
672
.
5
s
s
G
(
4
)
3.2. Compen
sator
Desig
n
Followi
ng the
propo
se
d d
e
sig
n
pro
c
e
d
u
re, the me
mbershi
p
fun
c
tion of the controlled
obje
c
t u
p
is shown in Figu
re 7(a
)
. The ra
nge of u
p
is ±h
max
= ±56.72
. The membe
r
shi
p
fun
c
tion
of
u
p
Zero (Z
) h
a
s the
value
of ±h
s-res
=
±0
.2152. Th
e m
e
mbe
r
ship fu
nction
of the
controlled
obj
ect
u
i
is sh
own in
Figure
7(b
)
. The ra
nge
of the me
mbe
r
ship functio
n
follows eq
uati
on (1
), ±u
i-max
= ±
38.4. Th
e me
mbershi
p
fun
c
tion
of u
i
Zero
(Z
) ha
s
the
value
of ±u
i,s-r
e
s
=
±0.00
08
according
to t
h
e
equatio
n (2).
The m
e
mb
e
r
shi
p
fun
c
tion
of the
c
ontrolled o
b
je
ct
output u
are
sho
w
n i
n
Fig
u
re
7(c). The ran
ge of u is ±u
r
= ± 10V.
p
u
p
u
p
u
p
u
p
u
p
u
(a) Me
mbe
r
s
h
ip of u
p
(b) Me
mbe
r
s
h
ip of u
i
(c) Memb
ership of u
Figure 7. Membershi
p
fun
c
tion of the controlle
d obje
c
t
1.
5
2
2.
5
3
3.
5
4
-1
0
0
10
20
30
40
50
60
Ti
m
e
,
s
In
p
u
t, V
D
i
spl
a
cem
e
nt
,
m
m
V
e
l
o
c
i
ty
, m
m
/
s
in
p
u
t
di
spl
a
c
e
m
en
t
ve
l
o
c
i
t
y
d
ece
l
e
r
a
t
i
on
ra
n
g
e
-1
-0
.5
0
0.
5
1
-6
0
-4
0
-2
0
0
20
40
60
Er
r
o
r
,
m
m
E
r
r
o
r
ra
te
, m
m
/s
A =
1.
354
h
ma
x
=
56.72
m
=
-
0
.
2152
h
s
-re
s
=
0.2152
a
s
-re
s
=
0.001
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Im
provem
ent of Fuzzy Ba
sed Pra
c
tical
Contro
lle
r for Contin
uou
s
Motion Co
ntrol (Purtojo
)
119
4. Ev
aluation
4.1. Positioning Perform
a
nce
In ord
e
r to e
v
aluate its p
e
rform
a
n
c
e, t
he p
r
op
osed
co
ntrolle
r i
s
com
p
a
r
ed
with the
existing p
r
op
ose
d
metho
d
in de
signin
g
fuzzy b
a
sed
NCT
F controllers. The
p
e
rform
a
n
c
e o
f
the
previou
s
m
e
thod of fu
zzy based
NC
TF (F-NCTF-exst
)
controll
er p
e
rf
ormi
ng
a p
o
int-to-point t
a
sk
has bee
n rep
o
rted
overco
me the
nor
m
a
l NCTF
co
ntrolle
r [9]. In th
is evalu
a
tion,
the pe
rform
a
nce
of the prop
o
s
ed
cont
rolle
r (F-NCTF
-
p
r
op
) is com
pare
d
to tha
t
of the con
t
roller th
roug
h
simulatio
n
usi
ng dynami
c
model of an e
x
perime
n
tal linear p
o
sitio
n
i
ng syste
m
.
The F-NCTF-exst cont
rolle
r is
stri
ctly de
sign
ed a
c
cording to
the proce
dure ment
ioned in
[9]. It has the
obje
c
t velo
ci
ty as the
NCT se
co
nd in
p
u
t. The
stru
cture
of its fu
zzy co
mpe
n
sa
tor
inclu
d
e
s
a g
a
in Kf to amplify the control si
g
nal u, and its fuzzy
membershi
p
function
s may
provide m
a
ximum actu
ator rated by usin
g singl
eton
s as fuzzy outp
u
t variable
s
.
Figure 8 sh
o
w
s the
re
spo
n
se
s of
the F-NCT
F-ex
st a
nd the F-
NCT
F
-p
rop
contro
llers to a
1mm step inp
u
t. It is sho
w
n that both controlle
rs
pro
v
ide same ri
se time, but significa
nt amo
unt
of overshoot.
The F-NCT
F
-exst
contro
ller produ
ce
overshoot te
n times hi
gh
er than F
-
NCTF-
prop. T
he F
-
NCT
F-ex
st o
v
ersh
oot i
s
1
0
%, while
th
e F-NCTF-prop ove
r
s
hoot
is only
1% of a
referen
c
e inp
u
t. Since the oversh
oot is high, F-
NCT
F
-exst re
ach
e
s 2% stea
d
y
state condit
i
on
slo
w
er tha
n
the F-NCTF-p
r
op of abo
ut 0.03 s.
Figure 8. Positioning pe
rformance to a 1 mm step inp
u
t
The p
o
sitio
n
i
ng p
e
rfo
r
man
c
e to
a 5
mm
step
input,
which
is fa
r fro
m
NCT, a
r
e
shown in
Figure 9. As
sho
w
n, the
F
-
NCTF
-prop
controlle
r out
perfo
rmed
th
e F-NCTF-exst. The
rise time
achi
eves by
F-NCTF
-p
rop
is
slig
htly faster. M
o
reov
er, the
overshoot i
s
o
n
ly
0.2%, less th
an
4.6% of F-NCTF-ex
st. Similar to the pre
v
ious resu
lt for a small input, the s
e
ttling time res
p
ons
e
of F-NCTF
-prop co
ntrolle
r i
s
faste
r
than
F-NCTF
-ex
s
t of about 0.16
s.
From Fi
gure
8 and 9, it is
sho
w
n that
si
gnifi
ca
nt oversho
o
t red
u
cti
on and fa
ste
r
settling
time ha
s
bee
n a
c
hieve
d
while the
p
r
op
ose
d
cont
roll
er m
a
intain
t
he
rise time.
This sho
w
s that
the positio
nin
g
perfo
rman
ce improvem
e
n
t has be
en a
c
hieve
d
.
Figure 9. Positioning pe
rformance to a 5 mm step resp
onse.
0
0.
2
0.
4
0.
6
0.
8
1
0.
7
5
0.
8
0.
8
5
0.
9
0.
9
5
1
1.
0
5
1.
1
1.
1
5
Ti
m
e
,
s
P
o
siti
o
n
, mm
re
f
e
r
e
nc
e
F-
N
C
TF-
ex
s
t
F-
N
C
TF-
pr
op
0
0.
5
1
1.
5
2
4.
7
4.
8
4.
9
5
5.
1
5.
2
Ti
m
e
,
s
P
o
s
i
t
i
on,
m
m
re
f
e
re
n
c
e
F
-
N
C
TF-
e
x
st
F-
N
C
T
F
-
pr
op
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 13, No. 1, Janua
ry 2015 : 114 –
123
120
4.1. Tracking
Performanc
e
The
trackin
g
perfo
rman
ce
evaluation
i
s
con
d
u
c
ted using sinu
soid
a
l
input with
v
a
rying
amplitude a
n
d
frequ
ency.
Re
spo
n
ses t
o
a 1 mm a
m
plitude at freque
ncy of
π
rad/s
sinu
soi
dal
input is
sh
own in Figu
re
1
0
. It is sho
w
n
that
the F-NCTF
-prop im
mediately forced th
e obj
ect to
follow the ref
e
ren
c
e in
put, but this is not
the case
fo
r the F-NCTF
-exst controlle
r. By varying
the
amplitude,
Fi
gure
1
1
, 12
a
nd 1
3
sho
w
the p
o
siti
o
n
tracking
e
rro
r t
o
a
si
nusoida
l input
at
π
ra
d/s
freque
ncy
wit
h
amplitu
de
of 1 mm,
1.5
mm a
nd
5
mm re
sp
ectiv
e
ly. The n
u
m
e
rical
re
sults
are
summ
ari
z
ed i
n
Table 2.
Table 2. Tracking p
e
rfo
r
ma
nce
s
with F
-
NCTF
-exs
t an
d
F-NCTF
-pro
p controlle
rs t
o
a sinu
soi
dal
referen
c
e inp
u
t at frequen
cy
π
rad/s
Amplitude Controller
x
r
-x
max|
x
r
-x|
(µm)
rms
(x
r
-x
)
(µm)
1 mm
F-NC
TF
-e
xst
34.3
19.4
F-NC
TF
-pro
p
4.1
1.2
1.5 mm
F-NC
TF
-e
xst
51.4
28.8
F-NC
TF
-pro
p
4.5
1.2
5 mm
F-NC
TF
-e
xst
169.6
90.1
F-NC
TF
-pro
p
27.9
4.2
From the nu
meri
cal re
sult
it is shown th
at t
he tracki
n
g
error of the F-NCTF
-ex
s
t tends to
increa
se p
r
o
portion
ally to the input
a
m
plitude.
Its
values are
also mu
ch highe
r
than
the
prop
osed co
ntrolle
r.
Th
e
tra
cki
ng error of
the
propo
sed
controller, F
-
NCT
F-p
r
op,
relati
vely
con
s
tant for
an oscillatio
n
amplitude
wi
thin and n
ear NCT. Th
ere is a small variation of max|x
r
-
x|, but
the rms(x
r
-x) valu
es confirm that
there is n
o
si
gni
fica
nt variation. Ho
wev
e
r, for an inp
u
t
amplitude
far
from
NCT, th
ere
is an
indi
cation
that th
e tra
c
king
error te
nd
s to i
n
cre
a
se
while
the
amplitude in
crea
se
s.
Figure 10. Re
spo
n
ses to a
1 mm amplitu
de
π
rad/
s fre
quen
cy sin
u
soidal si
gnal in
put
Figure 11. Position tra
cki
ng
erro
r to a 1 mm amplitud
e and
π
rad/
s frequen
cy si
nusoidal inp
u
t
0
1
2
3
4
5
6
7
8
-1
.
5
-1
-0
.
5
0
0.5
1
1.5
Ti
m
e
,
s
P
o
s
i
t
i
on,
m
m
0
0.
0
2
0.
04
0.
06
0.
08
0.
1
0
0.
0
5
0.
1
0.
1
5
0.
2
0.
2
5
re
f
e
re
n
c
e
F-
N
C
T
F
-
e
x
s
t
F
-
N
C
T
F
-p
ro
p
z
oom
v
i
ew
0
1
2
3
4
5
6
7
8
-
0
.04
-
0
.03
-
0
.02
-
0
.01
0
0.
01
0.
02
0.
03
Ti
m
e
,
s
T
r
a
cki
n
g
e
rro
r,
m
m
0
0.
1
0.
2
0.
3
0.
4
-0
.
0
3
-0
.
0
2
-0
.
0
1
0
0.
01
F-
N
C
T
F
-
e
x
s
t
F-
N
C
T
F
-
p
r
o
p
z
oom
v
i
ew
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Im
provem
ent of Fuzzy Ba
sed Pra
c
tical
Contro
lle
r for Contin
uou
s
Motion Co
ntrol (Purtojo
)
121
Figure 12. Position tra
cki
ng
erro
r to a 1.5
mm amplitud
e and
π
rad/
s frequen
cy si
nusoidal inp
u
t
Figure 13. Position tra
cki
ng
erro
r to a 5 mm amplitud
e and
π
rad/
s frequen
cy
In order to in
vestigate the influen
ce of the
input freq
uen
cy, the syst
em is drive
n
by a 1
mm amplitu
d
e
sin
u
soidal
input at vari
ous f
r
eq
uen
cies. Respon
ses to
a 1 m
m
amplitud
e
at
freque
ncy of
2
π
rad/s
si
nusoidal i
npu
t is sho
w
n in
Figur
e 1
4
. The p
o
sitio
n
tracking
erro
r is
sho
w
n in Fi
g
u
re 1
1
, 15 an
d 16 for fre
q
uen
cy
π
, 2
π
and 4
π
respe
c
tively. The numeri
c
al
re
su
lts
are summa
ri
zed in Table 3.
Table 3. Tracking p
e
rfo
r
ma
nce
s
with F
-
NCTF
-exs
t an
d
F-NCTF
-pro
p controlle
rs t
o
a sinu
soi
dal
referen
c
e inp
u
t with amplitude of 1 mm
Freque
nc
y Controller
x
r
-x
max|
x
r
-x|
(µm)
rms
(x
r
-x
)
(µm)
π
rad/s
F-NC
TF
-e
xst
34.3
19.4
F-NC
TF
-pro
p
4.1
1.2
2
π
rad/s
F-NC
TF
-e
xst
68.2
39.7
F-NC
TF
-pro
p
4.4
1.2
4
π
rad/s
F-NC
TF
-e
xst
134.1
76.5
F-NC
TF
-pro
p
16.3
1.5
Figure 14. Re
spo
n
ses to a
1 mm amplitu
de 2
π
rad/
s freque
ncy si
nu
soid
al sig
nal input
0
1
2
3
4
5
6
7
8
-0.06
-0.04
-0.02
0
0.
02
0.
04
0.
06
Ti
m
e
,
s
E
rro
r
,
m
m
0
0.
1
0.
2
0.
3
0.
4
-0
.0
6
-0
.0
5
-0
.0
4
-0
.0
3
-0
.0
2
-0
.0
1
0
0.
01
F-
N
C
T
F
-
e
x
s
t
F
-
N
C
T
F
-p
ro
p
Zo
om
v
i
e
w
0
1
2
3
4
5
6
7
8
-0.
2
-0
.
1
5
-0.
1
-0
.
0
5
0
0.0
5
0.
1
0.1
5
Ti
m
e
,
s
E
rro
r,
m
m
0
0.
1
0.
2
0.
3
0.
4
-0
.
1
8
-0
.
1
6
-0
.
1
4
-0
.
1
2
-0
.1
-0
.
0
8
-0
.
0
6
-0
.
0
4
-0
.
0
2
0
F-
N
C
T
F
-
e
x
s
t
F
-
N
C
T
F
-p
ro
p
Zo
om
v
i
ew
0
1
2
3
4
5
6
7
8
-1.
5
-1
-0.
5
0
0.
5
1
1.
5
Ti
m
e
,
s
Er
r
o
r
,
m
m
0
0.
02
0.
04
0.
06
0.
08
0.
1
0
0.
05
0.
1
0.
15
0.
2
0.
25
0.
3
0.
35
0.
4
r
e
f
e
r
enc
e
F
-
N
C
T
F
-
e
xst
F-
N
C
T
F
-
p
r
o
p
Zoom
v
i
ew
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 13, No. 1, Janua
ry 2015 : 114 –
123
122
Figure 15. Position tra
cki
ng
erro
r to a 1 mm amplitud
e and 2
π
ra
d/s frequ
en
cy sinusoidal inp
u
t
Figure 16. Position tra
cki
ng
erro
r to a 1 mm amplitud
e and 4
π
ra
d/s frequ
en
cy sinusoidal inp
u
t
Similar to th
e re
sult
sh
o
w
n i
n
Ta
ble
2, the tra
c
kin
g
erro
r of th
e F-NCTF
-exst sho
w
s
tenden
cy to increa
se p
r
op
ortionally to t
he input
freq
uen
cy. Its tracki
ng e
rro
r in
cre
a
ses
abo
ut
twice
while th
e fre
que
ncy i
n
crea
se
s
by
2 time
s. Fo
r t
he p
r
o
p
o
s
ed
controlle
r, F-NCT
F-p
r
o
p
, t
he
tracking
erro
r is rel
a
tively con
s
tant. Ho
wever, at the
highe
r freq
u
ency, max|x
r
-x| is a little bit
highe
r at initial obje
c
t movement a
s
sh
own in Fi
gu
re 16, but the
rms(x
r
-x) val
ue is o
n
ly sli
ghtly
increa
se. T
h
us, the
tra
c
ki
ng e
r
ror pe
rforma
nce of t
he p
r
op
osed
controlle
r i
s
l
e
ss influ
e
n
c
e
d
by
the input.
4. Conclusio
n
An improvem
ent of a fuzzy based
NCTF cont
rolle
r for contin
uo
us motion
h
a
s be
en
pre
s
ente
d
in
this study. Minor
stru
ctu
r
e ch
ang
e h
a
s si
gnifica
ntly influenced
the controll
er
perfo
rman
ce
on perfo
rmi
ng co
ntinuo
u
s
motion
a
nd point-to
-
p
o
int positio
ni
ng tasks. T
h
e
simulatio
n
re
sult sho
w
s th
at the pro
p
o
s
ed controll
er
i
s
effective fo
r tracking ta
sks to a
sinu
soi
dal
input. Th
e tracking
erro
r to inp
u
ts within an
d n
e
a
r
NCT a
r
e
not
depe
ndi
ng o
n
the
i
nput
amplitude and less i
n
fluenced by
t
he oscillation frequency.
Moreover, for
point-to-point
positio
ning ta
sks, less oversho
o
t, faster settling
time and ri
se time were achieve
d
. The pro
p
o
s
e
d
controlle
r is
effective for
point-to
-
poi
nt positio
ni
ng
wheth
e
r th
e i
nput is
clo
s
e
or fa
r from
NCT
.
Since
the
on
ly minor cha
nge i
n
the
structu
r
e,
the
prop
osed
co
ntrolle
r m
a
int
a
ins the
NCTF
simple
stru
ct
ure an
d ea
se
of desig
n pro
c
ed
ure
s
.
Ove
r
all, the pro
p
o
s
ed
controlle
r is effective and
signifi
cantly outperfo
rmed t
he existing m
e
thod as a p
r
actical cont
ro
ller. As the future wo
rks, the
prop
osed
co
ntrolle
r will
b
e
implem
ent
ed in a
re
al mechani
sm
to validate
and eval
uate
its
robu
stne
ss to disturb
a
n
c
e
experim
entall
y
.
Ackn
o
w
l
e
dg
ements
This
re
sea
r
ch wa
s
su
ppo
rted by
a re
sea
r
ch g
r
ant
from the
Di
recto
r
ate
Ge
neral
of
High
er Edu
c
a
t
ion, Ministry of Education
and Culture I
ndon
esi
a
.
0
1
2
3
4
5
6
7
8
-0
.0
8
-0
.0
6
-0
.0
4
-0
.0
2
0
0.02
0.04
0.06
Ti
m
e
,
s
Er
ro
r,
m
m
0
0.
1
0.
2
0.
3
-0
.
0
8
-0
.
0
7
-0
.
0
6
-0
.
0
5
-0
.
0
4
-0
.
0
3
-0
.
0
2
-0
.
0
1
0
0.
0
1
F-
N
C
T
F
-
e
x
s
t
F
-
N
C
T
F
-p
ro
p
Zoo
m
v
i
ew
0
1
2
3
4
5
6
7
8
-0
.
2
-0
.
1
5
-0
.
1
-0
.
0
5
0
0.05
0.1
0.15
Ti
m
e
,
s
E
rro
r,
m
m
0
0.
1
0.
2
0.
3
-0
.1
4
-0
.1
2
-0
.
1
-0
.0
8
-0
.0
6
-0
.0
4
-0
.0
2
0
F-
N
C
T
F
-
e
x
s
t
F
-
N
C
T
F
-p
ro
p
Zoo
m
v
i
ew
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Im
provem
ent of Fuzzy Ba
sed Pra
c
tical
Contro
lle
r for Contin
uou
s
Motion Co
ntrol (Purtojo
)
123
Referen
ces
[
1
]
Liu
X,
Wu Y,
L
i
u B.
T
he Rese
arch of
A
dapt
iv
e Sli
d
in
g Mo
de
Cont
ro
ller f
o
r
Mot
o
r Servo
S
y
st
em Us
in
g
F
u
zz
y
U
p
p
e
r
Boun
d o
n
Dist
urba
nces.
Inte
rnatio
nal J
our
nal
of Co
ntrol,
Auto
matio
n
,
and Syste
m
s
.
201
2;
10(5):
10
64-1
069.
[
2
]
Ang KH,
Cho
n
g
G.
PI
D Cont
rol S
y
st
em An
al
ysis,
Desi
gn,
and T
e
chnol
o
g
y
.
IEEE Transactions on
Contro
l Systems T
e
chn
o
l
ogy
.
2005;
13(
4):
559-5
76.
[
3
]
T
s
eng YT
,
Liu JH.
High-s
p
e
e
d
an
d precis
e
posit
i
oni
ng of
a
n
X–Y t
abl
e.
C
ontrol E
ngi
ne
e
r
ing Practic
e
.
200
3;
11(4):
35
7–3
65.
[
4
]
Don
ga
X,
Jian
-qu Z
,
F
eng W.
F
u
zzy
PI
D
Cont
rol t
o
F
e
ed Servo S
y
st
em of
CNC Machi
ne T
ool.
Proced
ia En
gin
eeri
n
g
.
20
12;
2
9
:
2853
–2
85
8.
[
5
]
Hua
ng SJ,
S
h
i
eh MH.
A
p
p
lica
t
ion
of
DSP
Co
nt
roll
er o
n
X–Y
T
able Serv
o C
ont
rol.
Inter
nati
ona
l Jo
urn
a
l
of Advance
d
M
anufactur
i
n
g
T
e
chn
o
lo
gy
.
200
0;
16(3):
20
5–2
11.
[
6
]
Quach D
C
,
Hu
ang S,
Y
i
n Q,
Z
hou C.
A
n
I
m
prove
d
Dir
e
ct
Adapt
iv
e F
u
zz
y C
ont
ro
ller f
o
r
an U
n
cert
a
i
n
DC Mot
o
r Spe
ed Co
nt
rol S
y
s
t
em.
T
e
lkomnik
a
.
2013;
1
1
(2):
108
3-10
92.
[
7
]
Wah
y
ud
i,
Sat
o
K,
Shim
oko
hbe
A.
Ch
ara
c
t
e
rist
ics of
p
r
act
i
cal c
ont
ro
l f
o
r p
o
int
-
t
o
-
poi
nt
(PT
P
)
posit
i
oni
ng s
y
s
t
ems Ef
f
e
ct
of
desi
gn p
a
rame
t
e
rs and act
u
at
or sat
u
rat
i
o
n
o
n
posit
i
o
n
i
ng
p
e
rf
ormanc
e.
Precisio
n Eng
i
neer
ing
.
2
003;
27(2):
15
7-1
6
9
.
[
8
]
Maed
a GJ,
S
a
t
o
K.
Pr
act
i
c
a
l c
ont
rol
met
hod
f
o
r
ult
r
a-
precisi
o
n
p
o
sit
i
oni
ng
us
ing
a b
a
llscr
e
w
mechanism.
Pre
c
i
s
i
o
n
En
gi
nee
ri
ng
.
200
8;
32
(4):
309-3
18.
[
9
]
Wah
y
ud
i,
I
b
rah
i
m T
F
,
Muhida
R,
Salami MJE.
Robust
f
u
zzy-b
ase
d
NCT
F
cont
roll
er f
o
r poi
nt
-t
o-poi
n
t
(PT
P
) posit
ion
i
ng s
y
st
ems.
Internati
ona
l Jour
nal of Syste
m
s
Simu
lati
on
.
20
07;
1(1):
11-2
9
.
[
10]
Sat
o
K,
Maed
a GJ.
A pract
i
cal cont
ro
l met
hod f
o
r prec
isi
on mot
i
o
n
- I
m
provem
ent
of
NCT
F
cont
rol
met
hod f
o
r co
n
t
inuo
us mot
i
o
n
cont
rol.
Precis
i
on Eng
i
n
eeri
n
g
.
2009;
33(
2):
175–
18
6.
Evaluation Warning : The document was created with Spire.PDF for Python.