TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol.12, No.5, May 2014, pp
. 3873 ~ 38
8
6
DOI: http://dx.doi.org/10.11591/telkomni
ka.v12i5.4218
3873
Re
cei
v
ed Au
gust 25, 20
13
; Revi
sed
No
vem
ber 1
0
, 2013; Accepte
d
De
cem
ber
5, 2013
The Position Control of the Hydraulic Cylinder
Controlled by the High-Speed On-Off Valve
Zhi-hao Liu*,
Qin-he Ga
o
Xi
’a
n Rese
arch
Institution of High T
e
chnol
og
y, Xi’a
n, Peo
p
le
’
s
Repu
blic of C
h
in
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: liuzhi
h
a
oai
na
na@
126.com
A
b
st
r
a
ct
T
h
is pap
er ai
ms to research t
he pos
itio
n co
ntro
l of the hy
drau
lic cyli
nder
controll
ed by t
he hi
gh-
spee
d o
n
-off v
a
lve to
re
ali
z
e
the pr
ecise
po
sition c
ontro
l.
A co
mpo
u
n
d
a
l
gorith
m
of pr
o
portio
nal
inte
gr
al &
spee
d fe
edfor
w
a
rd- dis
p
lac
e
me
nt fee
d
b
a
ck
is
put forw
ar
d b
a
se
d o
n
th
e res
earch
o
n
the r
e
l
a
tions
h
i
p
betw
een the
duty ratio an
d
the past flow
for the
high
spee
d on-
off valve. T
he sp
eed fee
d
forw
a
r
d-
displ
a
ce
ment f
eed
back
al
gori
t
hm c
a
n
solv
e
the
hyster
esi
s
resulti
n
g
fro
m
th
e fe
ed
bac
k contro
l a
n
d
be
consi
dere
d
as
the certain d
u
ty ratio is giv
en ah
ea
d,
an
d the pro
porti
ona
l inte
gral c
ontrol a
l
gor
ith
m
is
desi
gne
d
an
d
utili
z
e
d
to
d
e
a
l
w
i
th the
pos
iti
on
error
by
ad
justin
g to
the
duty rati
o
of h
i
gh s
p
e
ed
on-
o
ff
valve. T
he
ge
n
e
tic al
gorith
m
i
s
used to tu
ne
the c
ontro
l p
a
r
ameters, inc
l
u
d
in
g spe
ed for
w
ard coefficie
n
t
,
prop
ortion
al
an
d int
egra
l
co
efficient
of d
i
spl
a
cement
fe
ed
ba
ck, output c
oef
ficient to
ga
in t
he
opti
m
i
z
at
io
n
result, on t
e
rms of the perfor
m
a
nce
of the position
contr
o
l
system
. The
mathem
atics modeling is
analy
z
e
d
and s
i
mul
a
ted
w
i
th Simuli
nk
usin
g the
bu
lk-
c
avity-no
de
method,
and
the
hydra
u
lic
lo
op
is esta
blis
he
d
to
verify the si
mu
l
a
tion r
e
sult w
i
th the o
p
timi
z
e
d results on the FESTO platform
.
T
he r
e
se
arch fin
d
s that the
prop
osed
posi
t
ion sche
m
e i
s
effective to increas
e
the
positio
n prec
ision, a
nd the
positio
n cont
ro
l
para
m
eters opt
imi
z
e
d
by gen
e
t
ic algor
ith
m
ca
n decre
ase the
positio
n error
mor
e
.
Ke
y
w
ords
:
positi
on c
ontrol
,
spee
d fee
d
for
w
ard-disp
lace
me
nt fee
dback
alg
o
rith
m, h
i
g
h
-spe
ed
on-off
valve
,
gen
etic al
gorith
m
, PI control
Copy
right
©
2014 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
High
-spee
d o
n
-off valve
(HSV) is a
ne
w
kind
of
ele
c
tro-hyd
r
auli
c
di
gital valve, which ha
s
the advantag
e of fast switching
capabilit
y, anti-pollution and hig
h
repeata
b
ility,
and the pa
st flow
can be ch
ang
ed
by
a
d
ju
sting
the ope
n and clo
s
ed
ti
me of the val
v
e [1]. HSV can be
controll
ed
by the pulse
signal, and
combi
n
e
s
effectively
the compute
r
co
ntrol tech
nolo
g
y and hydrau
lic
fluid techn
o
lo
gy, and it is widely u
s
ed i
n
the spe
ed, positio
n, force cont
rol an
d
other o
c
casi
ons
[2].
In orde
r to solve the inhe
rent no
nlinea
r dy
nami
c
an
d uncertai
n
ties asso
ciate
d
with the
hydrauli
c
system, many re
sea
r
che
r
s h
a
v
e done
so
m
e
wo
rks. Ma
ny woks a
r
e
focu
sed o
n
the
feedba
ck con
t
rol, whi
c
h
compen
sate
s
the effect
brought by the
interferen
ce
to deal with
the
error
betwee
n
the
reali
s
tic value a
nd th
e refe
ren
c
e
value afte
r d
e
tected. PID control i
s
wide
ly
utilized
for its
agility, effect,
and
wid
e
a
p
p
lication.
Wei.
W
[3] utilized
the
PID co
ntrol
to cont
rol
t
he
automotive susp
en
sion sy
stem
to de
cre
a
se
t
he
b
ody vibration and improve
th
e
v
ehicl
e
comfort.
Wan
g
.L.B [4] used th
e PI
to
control th
e spee
d of t
he b
r
u
s
hle
ss DC m
o
tor b
y
com
pari
ng
the
target sp
eed
and the act
u
al spe
ed for i
nput to im
pro
v
e the stability of control system, while
the
control pa
ram
e
ters a
r
e initi
a
l artificially a
nd
ca
n’t be a
d
juste
d
with the ch
angi
ng
con
d
ition.
These contro
l
algo
rithms mentione
d
a
bov
e bel
ong
to the feed
back
cont
rol,
and th
e
control sig
nal
lags behin
d
the interfere
n
ce, and
it may leads to
the steady error du
ring t
h
e
regul
ation proce
s
s,
while the
feedf
o
r
ward
control
st
rategy i
s
wid
e
ly used i
n
t
he o
c
ca
sion
s with
much
inte
rference, which
ke
ep
s the
controlle
d
pa
ramete
rs cha
ngele
s
s to
compen
sate
the
influen
ce of t
he inte
rfere
n
c
e, a
nd it i
s
t
i
mely
com
p
a
r
ed with
the f
eedb
ack
cont
rol
strate
gy, but
with we
ak a
n
t
i-interferen
ce
ability. The feedfor
wa
rd
control
strateg
y
should b
e
combi
ned
with
the feedb
ack
control st
rate
gy, which ca
n be co
n
s
ide
r
ed a
s
the co
mpoun
d com
pen
sated
con
t
rol
method
ba
se
d on th
e inp
u
t. Dong.M.H [
5
] pro
p
o
s
e
s
the comp
ound
algo
rithm PD & feedforwa
rd-
feedba
ck, an
d utilize
d
to t
he spee
d
con
t
rol for
el
evat
or. Bao.L.Z [6
] used
the fe
edba
ck term
and
feedforwa
rd term to co
ntrol
the optimize
d
trac
king p
e
rforman
c
e of vehicl
e active
su
spe
n
si
on.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3873 – 38
86
3874
The i
n
itiation
para
m
eters o
f
the PID
co
n
t
rol
alg
o
rithm
influen
ce
largely the
pe
rforma
nce
of the control
system, many rese
arches
are done to optimize t
he param
e
ters, some intelligence
algorith
m
s a
r
e used to adj
usting the pa
ramete
rs o
n
-l
ine or off-line
.
Jiefan.C [7] design
s
a PID
controlle
r tun
ed by fuzzy
algorith
m
whi
c
h mad
e
use
of the PID control an
d the intelligen
ce
of
fuzzy
cont
rol
to achi
eve th
e po
sition
co
ntrol of
d
oubl
e cylind
e
rs b
y
comp
en
sati
ng the va
rian
ts
and no
-line
a
r cha
r
a
c
teri
stic. Xiaodon
g.
C [8] prop
oses a ne
w PID tuning al
g
o
rithm ba
se
d
on
Widrow-Hoff neural netwo
rk lea
r
nin
g
a
l
gorit
hm, whi
c
h can de
crease the respond time a
nd
improve the
control p
r
e
c
isi
on. While the
on-line t
uni
n
g
algorith
m
in
cre
a
se the co
mplexity of th
e
system. Th
e geneti
c
algo
ri
thm (GA) i
s
use
d
to
tune
the cont
rol p
a
ram
e
ters off-line, Karam
M.
Elbayomy [9] used the
ge
netic
algo
rith
m to adj
us
t t
he PID
para
m
eters to
co
ntrol the
mov
able
surfa
c
e
of
sp
ace
vehi
cle
s
.
This pa
per a
nalyse
s
and
utilize
s
the
g
enetic alg
o
rit
h
m to
optimize the
control para
m
eters to gain the
optimization
result, including
speed fo
rward coefficie
n
t,
prop
ortio
nal a
nd integral co
efficient of displacement fe
edba
ck and o
u
tput coeffici
ent.
The main
con
t
ribution
s
of this pa
per a
r
e:
(1)
The flo
w
cha
r
acte
ri
stic i
s
studie
d
with
the mathe
m
a
t
ical mo
del
o
f
HSV, and t
he
pulse width m
odulatio
n(P
W
M) method i
s
prop
osed to control the pa
st flow of HSV;
(2) The
comp
ou
nd alg
o
rithm
of PI & sp
eed fee
d
forward
-
di
spla
ce
ment feedb
a
ck i
s
resea
r
ched a
nd ada
pted to
preci
s
e p
o
siti
on co
ntrol;
(3)
GA is an
alyzed a
nd
used
to optimi
z
e
the control
a
l
gorithm
s
an
d the
optimi
z
ed
results a
r
e ve
rified by simul
a
tion and exp
e
rime
nt.
The re
st of the pap
er is
o
r
gani
ze
d as f
o
llows: the e
x
perime
n
tal system of the positio
n
control for th
e hydra
u
lic
cylinder
cont
ro
lled by HSV i
s
p
r
esented i
n
“
Design o
f
H
y
draulic Loop
”
Section. The
mathematical model of f
l
ow cha
r
a
c
teristic for
HSV
and the hyd
r
auli
c
sy
stem
is
analyzed an
d
presented
in
“
T
h
e M
a
thematics
M
odeling o
f
the Position
Con
t
rol Sy
s
t
em
”
se
ction. Th
e
positio
n con
t
rol strategy
is p
r
e
s
ented
in “
Th
e Co
mpound
Algorithm of PI
&
Speed F
eed
for
w
a
r
d
-
Displacemen
t Fe
e
dback
”
se
ctio
n. The o
p
timi
zation
and
si
mulation fo
r t
h
e
control para
m
eters
o
p
timized
by G
A
is present
ed in “
The
Position Co
ntrol Algorithm
Parameters
Optimized
b
y
GA
” se
ction. The
“
Ex
perimental Studies
”
section ver
i
fies
the
prop
osed
co
ntrol st
rategy
and the
opti
m
ized
re
su
lt
by experim
en
t. Finally, the study is
end
ed
with seve
ral concl
uding
re
marks of the rese
arch work.
2. Design of
H
y
draulic Loop
Hydra
u
lic po
sition
co
ntrol
loop
de
sig
n
ed in
th
is pa
per a
s
sho
w
n in
Figure 1 is mainly
comp
osed of
the Lenovo
comp
uter
with the data a
c
quisitio
n
ca
rd
PCI-62
21, the displ
a
ceme
nt
sen
s
o
r
for the
hydrauli
c
cyli
nder, HSV, a
nd thr
ee p
o
sit
i
on four-way reversi
ng valve and so on.
The di
spl
a
ce
ment sen
s
o
r
is in
stalled
to
mea
s
u
r
e th
e hydrauli
c
cylinder
displa
ceme
nt
whi
c
h i
s
gath
e
red
by the A
I
port of PCI
-
6221
data a
c
quisitio
n
card
(DAQ
). HSV
controls the fl
ow
to the hydra
u
l
i
c cylin
der
no
-stem
cavity by adju
s
ting the op
en time
of HSV. HSV is controll
ed
by
the amplified
PWM sig
nal, whi
c
h was p
r
odu
ced by th
e AO port of data acqui
sition ca
rd.
Figure 1. The
Experimental
Hydrauli
c
Lo
op
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
The Positio
n
Control of the Hyd
r
auli
c
Cylinder
Cont
rol
l
ed by the
Hig
h
-Spe
ed… (Z
hi-ha
o
Liu
)
3875
3 The Mathe
m
atics Mod
e
ling of the Position Co
ntrol Sy
stem
3.1. The Mathematics M
o
deling of Flow
Char
acteri
s
tic for
HSV
The rel
a
tion
ship between t
he sp
ool di
sp
lacem
ent an
d
PWM sig
nal
is dete
r
mine
d
by th
e
swit
chin
g cha
r
acte
ri
stic of HSV,
and
the
sp
ool di
s
p
la
ceme
nt
d
e
cid
e
s
dire
ct
ly
t
h
e pa
st
f
l
o
w
[
1
0]
.
The eq
uation
s
of the valve openi
ng a
r
ea and th
e p
a
st flow a
r
e
need
ed to a
nalyze the fl
ow
c
h
ar
ac
te
r
i
s
t
ic.
The area of valve openin
g
is sh
own in Equation (1):
sin(
2
)
/
2
hs
v
h
s
v
AD
x
(
1
)
Whe
r
e,
hs
v
A
is the
avera
ge val
v
e-po
rt area
of HSV;
D
is th
e
diamete
r
of
valve ball;
is the half-
angle
of valv
e seat;
hs
v
x
is
the averag
e displ
a
cem
ent of
valv
e ball,
whi
c
h
ca
n
be
ref
e
rred i
n
[1
1-
12].
The pa
st flow of HSV is sh
own in Equ
a
tion (2
):
2(
)
/
hs
v
d
hs
v
y
w
QC
A
p
p
(
2
)
Whe
r
e,
hsv
Q
is the past flow of HSV;
d
C
is the flow co
efficient
;
y
p
is the pre
s
su
re of rod cavity;
w
p
is
the pre
s
sure of no-st
em ca
vity;
is the oil den
sity.
Table 1. List
of Compo
nen
t Parameters
Component
Parameters
Value
HSV
Max
i
mum s
p
ool dis
p
lacement
0.0013
[
m
]
Diameter of valve ball
0
.005
[
m
]
Half-an
g
le of valve seat
20
[
de
g]
Relief valve
Adjustment pressure
3.5[MPa]
(a)
(b)
Figure 2. The
Mathematics Modeling
Re
sult
of the Flow Charac
teristic
for HSV
The m
a
them
atics mod
e
li
ng i
s
simula
ted with
the
sam
e
swit
ching
parame
t
ers and
different freq
uen
cy firstly,
the initial parameters a
r
e
12
3
4
[
,
,
,
]
[
2.5
,
1
,
2.5
,
1
]
tt
t
t
,
30
/
6
0
/
10
0
f
and the
simulatio
n
re
sult is shown
in Figu
ra
2(a
)
. The m
a
the
m
atics mod
e
l
i
ng is
also
si
mulated
with
the
same f
r
equ
ency an
d
different
swit
chin
g
p
a
ram
e
ters, the initial para
m
eters are
1
234
[
,
,
,
,
]
[
2
.5
,
0
.5
,
2
.5
,1
,
3
0
]
tt
t
t
f
12
3
4
[
,
,
,
,
]
[
1
.5
,
0
.
5
,
2
.5
,
1
,
3
0
]
tt
t
t
f
,
12
3
4
[
,
,
,
,
]
[
2
.5
,1
,
2
.5
,
1
,
3
0
]
tt
t
t
f
12
3
4
[
,
,
,
,
]
[
2
.
5
,1
,1
.
5
,1
,
3
0
]
tt
t
t
f
,
12
3
4
[
,
,
,
,
]
[
2
.5
,
1
,
2
.5
,
0
.5
,
3
0
]
tt
t
t
f
respe
c
tively and the
sim
u
lation result is sho
w
n in
Figure 2(b).
T
he
mathemati
c
analysi
s
resu
lt sho
w
s th
at: (1) T
he switching ch
ara
c
teri
st
ic
para
m
eters an
d the
freque
ncy
of pulse si
gnal
affect directly
the flow
ch
a
r
acte
ri
stic of
HSV; (2) On
occa
sion
of the
mech
ani
sm
and ele
c
tro
m
agneti
s
m chara
c
te
risti
c
s
of the spo
o
l, the past
flow ha
s linear
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3873 – 38
86
3876
relation
shi
p
with the
duty
ratio of
the
control
sig
nal
within th
e d
u
ty ratio
ran
g
e
[0
.
1
~
0
.
9
]
; (3
) T
h
e
dead
and
sut
u
red
zo
ne ex
ists a
s
the f
r
eque
ncy of t
he pul
se
sig
n
a
l be
come
s b
i
gger, the
lin
ear
zon
e
de
creases, and th
e flow of HSV m
agnifie
s
,
whe
n
the duty rat
i
o of pulse
si
gnal be
com
e
s
bigge
r. Dea
d
zone, saturated
zon
e
, and none
-li
near
zon
e
becom
e bigg
er as
switch
ing
cha
r
a
c
teri
stic paramete
r
s of HSV getting bigge
r
with frequ
en
cy of pulse sign
al fixed.
The
minimal
re
sp
ondin
g
d
u
ty ratio of
HSV i
s
12
mi
n
tt
dc
T
and the
ma
ximal re
sp
on
ding
duty ratio is
34
ma
x
1
tt
dc
T
.
Based o
n
the analysi
s
o
f
the flow chara
c
te
risti
c
, the pulse sign
al with the fixed
freque
ncy 60
HZ is utilized
and the pa
st flow is a
d
ju
ste
d
with the dut
y ratio by PWM method.
3.2. The Mathematics M
o
deling of the
H
y
draulic Sy
stem
The flow
equ
ation can illu
minate the
relation
ship b
e
twee
n the p
r
essu
re
and
the total
flow with the
bulk-cavity-n
ode metho
d
[13], which is
sho
w
n in Equ
a
tion (3
).
0
i
E
PQ
d
t
V
(
3
)
Whe
r
e,
i
Q
is
the total flow of the c
a
vity
;
P
is the pressu
re of the ca
vity
;
0
E
is the
elasti
c modul
us of oil
;
V
is the volume of rodle
ss
cavity;
The hydrauli
c
system mo
d
e
l is e
s
tabli
s
h
ed
with the
b
u
lk-cavity-n
o
de metho
d
a
c
cordi
n
g
to hydra
u
lic
system sch
e
m
a
tic dia
g
ra
m, and the
appli
c
ation i
s
m
o
d
e
ling respe
c
tively, as sh
own
in Equation (4-7
);
The flow eq
u
a
tion of rodle
ss
cavity:
0
()
we
hs
v
x
w
ww
dp
QQ
A
x
dt
V
A
x
(
4
)
Whe
r
e,
e
is ela
s
tic m
odulu
s
of oil,
hsv
Q
is the fl
ow that flo
w
i
n
to the rod
cavity;
x
Q
is the fl
ow that
leakage from
rod
cavity to rodle
s
s cavity;
0
w
V
is the initi
a
l volume of
rodless cavity;
w
A
is the
effective stre
ss a
r
ea of pi
ston in rodle
ss cavity;
The equ
ation
of rod cavity:
0
()
y
e
x
y
out
yy
dp
QA
x
Q
dt
V
A
x
(
5
)
Whe
r
e,
y
A
is the effective stre
ss a
r
ea
of piston in rod cav
i
ty;
0
y
V
is the initial volume of rod cavity,
out
Q
is
the flow that flow out of rodles
s
c
a
vity;
The equ
ation
of force b
a
lan
c
e:
yy
w
w
c
mx
p
A
p
A
mg
x
(
6
)
Whe
r
e,
m
is the
equivalent m
a
ss of pi
ston;
c
is viscou
s d
a
m
ping
coeffici
ent of oil;
g
is th
e gravity
accele
ration;
The equ
ation
of the leaka
g
e
flow:
()
x
cy
w
Qk
p
p
(
7
)
Whe
r
e,
c
k
is the leakage flo
w
coefficient;
p
p
is the pre
s
sure of system.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
The Positio
n
Control of the Hyd
r
auli
c
Cylinder
Cont
rol
l
ed by the
Hig
h
-Spe
ed… (Z
hi-ha
o
Liu
)
3877
Table 2. List
of Compo
nen
t Parameters
of the Hydrau
lic System
Serial Number
Parameters
Value
Unit
1
elastic modulus
of oil
e
750
MP
a
2
the pressure of s
y
st
em
p
p
3.5
MP
a
3
The densit
y
of
oil
850
3
kg/
m
4 maximal
displacement
x
c
x
200
mm
5
the area of
the ro
d cavity
y
A
120
2
mm
6
the area of
the ro
dless cavity
w
A
200
2
mm
7
The mass of piston
m
9
kg
8
viscous damping
c
oefficient of oil
c
250
Ns
/
m
9
the initial volume
of r
od cavity
0
y
V
2827.5
3
mm
10
the initial volume
of rodless cavity
0
w
V
5696.5
3
mm
11
the leakage flo
w
coefficient
c
k
0.0033
3
m/
(
Pa
s
)
The
open
loo
p
expe
rime
nt is
don
e to v
a
li
date th
e m
a
thematics m
odelin
g by m
easure
the displa
ce
ment of the hydrauli
c
cylin
der with
the
control sig
nal
for HSV given ahead an
d the
open lo
op co
ntrol si
gnal a
nd the di
spla
ceme
nt are
s
hown in Figu
re 2-3. Th
e di
spla
cem
ent e
rro
r
of the mathe
m
atics
mod
e
l
i
ng and
expe
riment result
exists for th
e
control de
ad
zon
e
of HS
V
durin
g the e
x
perime
n
t proce
s
s, the chang
eable p
r
essure lead
s to the instability switchi
n
g
para
m
eters, while the
swit
chin
g parame
t
er
s of sim
u
la
ting environ
m
ent is fixed.
Figure 2. The
Open Lo
op
Duty Ratio Te
st
Signal
Figure 3. The
Result of the Open Lo
op
Experiment
4. The Comp
ound Algori
t
hm of PI & Speed Fe
edfo
r
w
a
r
d
-
Dis
p
la
cement F
eed
back
This
pape
r p
r
opo
se
s the
compou
nd al
g
o
rithm of PI
& spe
ed fee
d
f
orwa
rd
-di
s
pl
acem
ent
feedba
ck, whi
c
h is
sho
w
n i
n
Figure 4.
Figure 4. The
Compo
und A
l
gorithm of PI &
Speed Fee
d
forward-di
sp
lacem
ent Fee
dba
ck
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3873 – 38
86
3878
The fe
edforward
value
is referred
as th
e control
sig
nal, an
d the
f
eedb
ack
algo
rithm i
s
utilized to co
mpen
sate the
erro
r betwee
n
the referen
c
e inp
u
t and the reali
s
tic o
u
tput to deal with
the un
ce
rtaint
ies
asso
ciate
d
with
the
hydrauli
c
syste
m
. The fe
edf
orward in
put
sign
al i
s
multi
p
lied
by the PI fee
dba
ck
control
so that the final co
nt
rol du
ty ratio can b
e
expre
s
sed i
n
Equation (8
):
21
()
[
*
()
()
]
id
e
a
l
P
I
ut
k
k
v
t
u
t
(8)
Whi
c
h:
2
k
is the output co
ntrol
coefficie
n
t;
1
k
is the spe
ed fee
d
forward co
efficient;
()
PI
ut
is
the displ
a
ce
ment feedba
ck o
u
tput
ba
sed on the fee
dba
ck al
gorit
hm
;
The PI
control is
used
a
s
the
feedb
a
c
k alg
o
rithm
with it
s agil
i
ty, effect, and
wide
appli
c
ation
to
co
mpe
n
sate
the
ch
angin
g
loa
d
a
n
d
di
sturb
a
n
c
e. T
he p
o
sitio
n
e
rro
r
between
the
referen
c
e an
d reality is:
12
ex
x
(
9
)
Whe
r
e,
1
x
is the refere
nce displacement;
2
x
is the reali
s
tic di
spla
cem
ent o
f
hydraulic
cylinder.
The output of
the PI contro
l is:
11
()
()
(
)
()
(
)
kk
pi
a
p
i
ii
uk
K
e
k
K
e
j
T
K
e
k
K
e
j
(
1
0
)
Whe
r
e,
ii
a
K
KT
;
p
K
is the pro
p
o
r
tion
al co
efficient
;
i
K
is the integ
r
al coefficie
n
t
;
T
is sam
p
ling
perio
d;
k
is sam
p
ling serial n
u
mbe
r
.
1) T
he p
r
o
p
o
rtional
coefficient
p
K
is u
s
e
d
to control
the sy
stem
error
s
e
, while the
incresce
nt proportio
nal co
efficient may decrea
s
e the
stability or ca
use the
syste
m
instable;
2) T
he i
n
tegral coefficie
n
t
i
K
is
used to
decrea
s
e
the
stea
dy e
rro
r and
in
cre
a
se th
e
steady preci
s
i
on.
The spe
ed
fe
edforwa
rd ref
e
ren
c
e
curve and
th
e
di
spl
a
cem
ent fee
d
back
referen
c
e curve
are sho
w
n in
Figure 5.
(a) T
he feedb
ack displa
ce
ment refe
ren
c
e
curv
e
(b) T
he feedf
orward sp
eed
refere
nce cu
rve
Figure 5. The
Speed Fee
d
forward-di
sp
l
a
ceme
nt Feed
back Refere
n
c
e Curve
The p
a
ra
met
e
rs are initi
a
lize
d
a
s
60
p
K
、
2
i
K
、
1
4
k
、
2
1
k
to verify the spee
d
forwa
r
d
-
di
spl
a
cem
ent feed
back al
gorith
m
, and the
re
sults are
sho
w
n
in Figu
re 6.
The simula
tion
result
sho
w
s that: (1
) th
e spee
d forward-disp
lacement fe
edb
ack al
gorith
m
is effectiv
e t
o
decrea
s
e th
e tracking
error, and the disp
lacem
ent e
rror of po
sition
control is limit
ed in 2mm; (2)
the tra
c
king
e
rro
r
of the
sta
r
t an
d e
nd
proce
s
s i
s
la
rg
er th
an th
e ot
her process f
o
r th
e
cycle
ratio
0
0.
5
1
1.
5
2
2.
5
3
0
0.
0
2
0.
0
4
0.
0
6
0.
0
8
0.
1
0.
1
2
0.
1
4
0.
1
6
0.
1
8
0.
2
Ti
m
e
/
s
The
di
spl
a
cem
ent
f
e
e
dback
r
e
f
e
r
enc
e/
(
m
)
0
0.
5
1
1.
5
2
2.
5
3
0
0.
01
0.
02
0.
03
0.
04
0.
05
0.
06
0.
07
0.
08
0.
09
0.
1
Ti
m
e
/
s
T
h
e
s
p
e
e
d
f
e
ed
f
o
r
w
ar
d r
e
f
e
r
e
nc
e/
(
m
/
s
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
The Positio
n
Control of the Hyd
r
auli
c
Cylinder
Cont
rol
l
ed by the
Hig
h
-Spe
ed… (Z
hi-ha
o
Liu
)
3879
belon
gs to th
e de
ad
zo
ne
of HSV; (3) th
e contro
l p
a
rameters
of th
e spee
d fo
rward
-
di
spla
ce
ment
feedba
ck al
g
o
rithm i
s
initialize
d
p
e
rson
ality, and the
optimize
d
m
e
thod
can
be
use
d
to
adju
s
ting
the cont
rol pa
ramete
rs to d
e
crea
se the p
o
sition e
r
ror
more.
(a) T
he displa
ceme
nt error
(b) T
he duty ratio
Figure 6. The
Performa
nce
of the Position C
ontrol with
out Paramete
rs O
p
timize
d by GA
5 The Positi
on Con
t
rol Algorithm Parameter
s
Opti
mized b
y
GA
The
fou
r
control para
m
eters
p
K
、
i
K
、
1
k
and
2
k
are
initiale
d ba
sed
on
the expe
rt
intelligen
ce
a
nd diffe
rent
control
pa
ram
e
ters infl
u
ence the
dynami
c
p
e
rfo
r
ma
nce of th
e p
o
sit
i
on
control. In
order to d
e
cre
a
se
the
po
si
tion erro
r, thi
s
p
ape
r
utilizes th
e g
enet
ic al
gorith
m
t
o
optimize the
para
m
eters o
ff-line.
GA is based
on natural
sele
ction an
d geneti
c
theory, and it is the efficie
n
t global
optimizatio
n search alg
o
rith
m whi
c
h is
co
mbination
of t
he su
rvival of fitte
st rules a
nd the ra
ndo
m
informatio
n e
x
chan
ge m
e
chani
sm of
ch
romosome
s
within the g
r
o
u
p
[14]. The
p
o
ssible
sol
u
tion
of the proble
m
domain is
see
n
as an i
ndividual o
r
chromo
som
e
, and the gro
up is re
peate
d
ly
operated
with
geneti
c
op
eration a
s
sele
ction, cro
s
so
ver, and m
u
tation. In orde
r to a
c
hieve t
he
optimal
soluti
on, GA u
s
e
s
the fitness fu
nction
to eval
uate ea
ch
in
dividual, an
d
gain
s
the
bet
ter
grou
p ba
sed
on evolution
a
r
y rule
s, the opt
imization p
r
oce
s
s sh
own in Figure 7.
Figure 7. The
Optimization
Process
of the Position Co
ntrol System
0
0.
5
1
1.
5
2
2.
5
3
-0.
5
0
0.
5
1
1.
5
2
2.
5
x 1
0
-3
Ti
m
e
/
(
s
)
T
h
e di
s
p
l
a
cem
ent
er
r
o
r
of
pos
i
t
i
on c
ont
r
o
l
/
(
m
)
0
0.
5
1
1.
5
2
2.
5
3
0
0.
05
0.
1
0.
15
0.
2
0.
25
0.
3
0.
35
0.
4
0.
45
0.
5
Ti
m
e
/
(
s
)
The dut
y r
a
t
i
o/
(
%
)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3873 – 38
86
3880
The
spe
ed fe
edforwa
rd-displacement fe
edba
ck
alg
o
ri
thm paramet
ers are optim
ized i
n
the followin
g
step
s:
Step 1: Initial
i
ze th
e first g
eneration
wit
h
ra
ndo
mly p
opulatio
n, an
d then
evalu
a
te ea
ch
popul
ation;
Step 2:
(a) Sele
ct chil
dren from th
e
set with the better pe
rformance;
(b) Apply cro
s
sover with a
given rate;
(c) Apply mutation with a gi
ven rate;
(d)
Rep
eat (a
) to (d) g
ene
rate the new g
eneration;
Step 3: Repe
at steps 2 u
n
til as toppin
g
criterio
n is satisfied.
The gen
etic a
l
gorithm p
a
ra
meters are
sh
own in Ta
ble
3.
Table 3. The
Initial Parame
ters of GA
Parameters
Val
ue
Parameters
Val
ue
Population size
20
Crossover rate
0.7
Number of
gener
ations
200
Mutation rate
0.1
Gene
ration gap
0.9
In order to ob
tain the satisf
ied dynami
c
pro
c
e
ss
cha
r
acteri
stic, we use Equ
a
tion
(11) a
s
the obje
c
t function:
2
1
1
()
N
i
ye
i
N
()
Ns
i
z
e
e
(
1
1
)
The fitness is evaluated by
the object fu
ncti
on
with the linear
so
rtin
g and the differen
c
e
is equ
al to 2.
The ra
nge of
the variable
s
whe
n
optimized is sho
w
n i
n
Table 4.
Table 4. The
Ran
ge of the Optimize
d Va
riable
s
Parameters
Range
1
k
(The speed
feedf
or
w
a
r
d
coefficient)
[
0
~
10]
p
K
(The displaceme
nt feedback prop
ortional coefficie
n
t)
[
0
~
100
]
i
K
(The displaceme
nt feedback integ
r
al coefficient)
[
0
~
10]
2
k
(the output
coefficient)
[
0
~
10]
This p
ape
r propo
se
s thre
e
optimizatio
n pro
c
e
s
ses fo
r the positio
n control sy
ste
m
. The
first is that the param
eters
p
K
,
i
K
,
2
k
is initialed and
1
k
need
s to be optimize
d
by GA; The secon
d
is that th
e pa
rameters
2
k
is ini
t
ialed and
p
K
,
i
K
and
1
k
nee
d to b
e
optimize
d
by
GA; the third i
s
th
e
para
m
eters
2
k
,
p
K
,
i
K
and
1
k
need to b
e
optimize
d
b
y
GA.
The pa
ram
e
ter
1
k
is optimi
z
e
d
by GA with
60
p
K
,
2
i
K
and
2
1
k
,and the value of obj
ect
function i
s
sh
own in
Figu
re 8(a
)
. The
simulation i
s
d
one with th
e
optimal soluti
on
1
5.45
k
,
a
nd
the displ
a
ce
ment error a
nd the cycl
e
ratio is s
h
o
w
n in Figu
re
8(b, c). Th
e
simulation
result
sho
w
s that: (1) the optimi
z
ed po
sition control alg
o
rith
m para
m
eters ca
n de
crea
se effectively the
positio
n error from[0-2]mm
to [0.9-1.5]mm; (2)
the large
r
tra
cki
n
g
error of the start and e
n
d
pro
c
e
ss exi
s
ts whi
c
h results from the de
ad zo
ne of HSV.
The paramet
ers
p
K
,
i
K
and
1
k
a
r
e op
timiz
e
d b
y
G
A
w
i
th
2
1
k
,and
the valu
e
o
f
obje
c
t
function is
sh
own in Figu
re 9(a). The
simulation is d
one with the optimal sol
u
tions
96.2
4
p
K
,
5.76
i
K
,
1
5.47
k
, and
the
di
spla
cem
ent e
rro
r
and
the
cycle
ratio i
s
sho
w
n
in
Fig
u
re
9
(b,
c).
The
simul
a
tio
n
result sho
w
s that:
(1) the
optimi
z
ed
po
sition
co
ntrol
algorith
m
s ca
n de
crea
se
the
positio
n erro
r from [0.9-1.5
] mm to [-0.5-1] mm; (2
) th
e larg
er tra
c
ki
ng erro
r of the start a
nd e
nd
pro
c
e
ss al
so
exists which result
s from th
e dead
zon
e
of HSV.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
The Positio
n
Control of the Hyd
r
auli
c
Cylinder
Cont
rol
l
ed by the
Hig
h
-Spe
ed… (Z
hi-ha
o
Liu
)
3881
(a)
Comp
utational optimi
z
e
d
curve of GA
(b) T
he displa
ceme
nt error
(c
) The d
u
ty
ratio
Figure 8. The
Performa
nce
of the Position Control with
1
k
Optimize
d by GA
(a)
Comp
utational optimi
z
e
d
curve of GA
(b) T
he displa
ceme
nt error
(c
) The d
u
ty
ratio
Figure 9. The
Performa
nce
of the Position Control with
p
K
,
i
K
and
1
k
Optimize
d by GA
0
5
10
15
20
25
30
35
40
45
50
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
4.
5
5
x 1
0
-6
Th
e ge
n
e
r
a
t
i
on
of
po
pu
l
a
t
i
o
n
s
T
he
v
a
l
u
e of
obj
ec
t
B
e
s
t
f
i
tn
e
s
s
Av
e
r
a
g
e f
i
t
n
es
s
0
0.
5
1
1.
5
2
2.
5
3
-1
-0
.
5
0
0.
5
1
1.
5
2
2.
5
x 1
0
-3
Ti
m
e
/
(
s
)
T
he di
spl
a
c
e
m
ent
er
r
o
r
of
pos
i
t
i
on c
ont
r
o
l
/
(
m
)
T
he pos
i
t
i
on c
ont
r
o
l
w
i
t
hout
G
enet
i
c
A
l
gor
i
t
hm
opt
i
m
i
z
i
n
g
T
he pos
i
t
i
on c
ont
r
o
l
w
i
t
h
k
1
opt
i
m
i
z
ed by
G
enet
i
c
A
l
gor
i
t
hm
0
0.
5
1
1.
5
2
2.
5
3
0
0.
0
5
0.
1
0.
1
5
0.
2
0.
2
5
0.
3
0.
3
5
0.
4
0.
4
5
0.
5
Ti
m
e
/
(
s
)
The dut
y
r
a
t
i
o/
(
%
)
T
he p
o
s
i
t
i
on c
ont
ro
l
w
i
t
h
out
G
ene
t
i
c
A
l
g
o
ri
t
h
m
op
t
i
m
i
z
i
n
g
T
he p
o
s
i
t
i
on c
ont
ro
l
w
i
t
h
k
1
o
p
t
i
m
i
z
e
d
by
G
e
n
e
t
i
c
A
l
gori
t
hm
0
5
10
15
20
25
30
35
40
45
50
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
4.
5
5
x 1
0
-6
T
h
e
ge
n
e
r
a
t
i
on
of
p
opu
l
a
t
i
on
s
T
h
e
va
l
u
e
of
o
b
j
e
ct
B
e
s
t
fi
t
n
e
s
s
A
v
er
ag
e f
i
t
n
e
s
s
0
0.
5
1
1.
5
2
2.
5
3
-1
-0
.
5
0
0.
5
1
1.
5
2
2.
5
x 1
0
-3
Ti
m
e
/
(
s
)
T
h
e di
s
p
l
a
c
e
m
ent
er
r
o
r
of
pos
i
t
i
o
n c
o
n
t
r
o
l
/
(
m
)
T
he
pos
i
t
i
o
n c
ont
r
o
l
w
i
t
hout
G
enet
i
c
A
l
gor
i
t
hm
opt
i
m
i
z
i
n
g
T
he
pos
i
t
i
o
n c
ont
r
o
l
w
i
t
h
k
1
opt
i
m
i
z
ed by
G
enet
i
c
A
l
gor
i
t
hm
T
he
pos
i
t
i
o
n c
ont
r
o
l
w
i
t
h
k
i
,
k
d,
k
1
opt
i
m
i
z
ed by
G
enet
i
c
A
l
gor
i
t
hm
0
0.
5
1
1.
5
2
2.
5
3
0
0.
05
0.
1
0.
15
0.
2
0.
25
0.
3
0.
35
0.
4
0.
45
0.
5
Ti
m
e
/
(
s
)
Th
e du
t
y
r
a
t
i
o
/
(
%
)
T
h
e pos
i
t
i
o
n
c
o
n
t
r
o
l
w
i
t
h
out
Gen
e
t
i
c
A
l
go
r
i
t
h
m
o
p
t
i
m
i
z
i
n
g
T
h
e pos
i
t
i
o
n
c
o
n
t
r
o
l
w
i
t
h
k
1
op
t
i
m
i
z
e
d by
Ge
ne
t
i
c
A
l
gor
i
t
hm
T
h
e pos
i
t
i
o
n
c
o
n
t
r
o
l
w
i
t
h
k
i
,
k
d,
k
1
o
p
t
i
m
i
z
ed b
y
Gen
e
t
i
c
A
l
go
r
i
t
h
m
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3873 – 38
86
3882
The paramet
ers
p
K
,
i
K
,
2
k
and
1
k
are
o
p
timized
by
GA, and
the
value of
obje
c
t fun
c
tion i
s
sho
w
n
in
Fig
u
re
10
(a
). T
he
simul
a
tion
is do
ne
with
the
optimal
solutio
n
s
68.68
p
K
,
9.66
i
K
,
1
2.78
k
,
2
1
.
9657
k
, and the
displacement
error and
the
cycle ratio is shown in Fi
gu
re 1
0
(b,
c).
The si
mulatio
n
re
sult sho
w
s th
at: (1) t
he optimi
z
ed
positio
n co
n
t
rol algo
rithm
para
m
eters
can
decrea
s
e
the
positio
n e
rro
r from [
-
0.5-1
]
mm to [-0.
4-0.8]mm; (2) the la
rge
r
tra
c
king
erro
r of the
start an
d end
pro
c
e
ss a
nd the light turbul
ence exis
ts
which results from the dead
zon
e
of HSV.
(a)
Comp
utational optimi
z
e
d
curve of GA
(b) T
he displa
ceme
nt error
(c
) The d
u
ty
ratio
Figure 10. Th
e Perform
a
n
c
e of the Position Co
ntrol wit
h
p
K
i
K
1
k
and
2
k
Optimized by GA
The optimi
z
at
ion re
sults of
the four optim
ization p
r
o
c
e
ss a
r
e
sho
w
n
in Table 6.
Table 6. The
Optimizatio
n
Re
sults for th
e Optimize
d Variabl
es
The position control algorithm
1
k
p
K
i
K
2
k
The position control w
i
tho
u
t being
optimized by GA
4
60
2
1
The position control w
i
th
1
k
optimized b
y
GA
5.45
60 2 1
The position control w
i
th
p
K
,
i
K
and
1
k
optimized by
GA
5.47 96.24
5.76
1
The position control w
i
th
p
K
,
i
K
,
1
k
and
2
k
optimized b
y
GA
2.78 68.68
9.66 1.96
6. Experimental Studies
The po
sition
control expe
riment is d
o
n
e
to
verify the simul
a
tion
result ba
sed
on the
FESTO platfo
rm. The hyd
r
aulic lo
op is
set up in
Figu
re 1 and the
system pre
s
su
re is 6
M
Pa. The
sampl
e
rate
of the displa
cement si
gnal
is 1KHZ,
and
the rate
of control al
go
rith
m is 10
0HZ.T
h
e
positio
n co
ntrol expe
rime
nts are don
e with di
fferent spe
ed fo
rwa
r
d
-
di
spla
cement feed
b
a
ck
algorith
m
parameters to verify the effective
of the optimization
re
su
lt optimized b
y
GA.
0
5
10
15
20
25
30
35
40
45
50
0
0.
5
1
1.
5
2
2.
5
3
3.
5
4
4.
5
5
x 1
0
-6
T
h
e g
e
n
e
r
a
t
i
on
of
p
o
p
u
l
a
t
i
on
s
T
h
e v
a
l
u
e
of
obj
e
c
t
B
e
s
t
f
i
t
nes
s
A
v
er
ag
e f
i
t
nes
s
0
0.
5
1
1.
5
2
2.
5
3
-1
-0
.
5
0
0.
5
1
1.
5
2
2.
5
x 1
0
-3
Ti
m
e
/
(
s
)
T
he di
s
p
l
a
c
e
m
e
nt
error of
po
s
i
ti
on c
o
n
t
rol
/
(
m
)
T
he po
s
i
t
i
on c
o
n
t
r
o
l
w
i
t
hout
G
enet
i
c
A
l
gor
i
t
hm
o
p
t
i
m
i
z
i
ng
T
he po
s
i
t
i
on c
o
n
t
r
o
l
w
i
t
h
k
1
opt
i
m
i
z
ed
by
G
enet
i
c
A
l
gor
i
t
hm
T
he po
s
i
t
i
on c
o
n
t
r
o
l
w
i
t
h
k
i
,
k
d,
k
1
opt
i
m
i
z
ed by
G
enet
i
c
A
l
g
o
r
i
t
h
m
T
he po
s
i
t
i
on c
o
n
t
r
o
l
w
i
t
h
k
i
,
k
d,
k
1
,
k
2 o
p
t
i
m
i
z
ed b
y
G
enet
i
c
A
l
gor
i
t
hm
0
0.
5
1
1.
5
2
2.
5
3
0
0.
05
0.
1
0.
15
0.
2
0.
25
0.
3
0.
35
0.
4
0.
45
0.
5
Ti
m
e
/
(
s
)
T
h
e du
t
y
rat
i
o/
(
%
)
T
he po
s
i
t
i
o
n
c
o
n
t
r
o
l
wi
t
hou
t
G
e
net
i
c
A
l
go
r
i
t
h
m
opt
i
m
i
z
i
n
g
T
he po
s
i
t
i
o
n
c
o
n
t
r
o
l
wi
t
h
k
1
opt
i
m
i
z
e
d
by
G
ene
t
i
c
A
l
gor
i
t
hm
T
he po
s
i
t
i
o
n
c
o
n
t
r
o
l
wi
t
h
k
i
,
k
d,
k
1
op
t
i
m
i
z
ed by
G
e
n
e
t
i
c
A
l
gor
i
t
hm
T
he po
s
i
t
i
o
n
c
o
n
t
r
o
l
wi
t
h
k
i
,
k
d,
k
1
,
k
2
opt
i
m
i
z
ed
by
G
enet
i
c
A
l
g
o
r
i
t
h
m
Evaluation Warning : The document was created with Spire.PDF for Python.