TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 12, No. 8, August 201
4, pp. 6101 ~ 6110
DOI: 10.115
9
1
/telkomni
ka.
v
12i8.517
3
6101
Re
cei
v
ed
No
vem
ber 2
0
, 2013; Re
vi
sed
March 31, 20
14; Accepted
April 15, 201
4
Slip Enhancement in Continuously Variable
Transmission by Using Adaptive Fuzzy Logic and LQR
Controller
Ma Shu
y
uan
1
, Sameh Bdran*
2
, Samo Saifullah
3
, Jie Huan
g
4
Mechatro
nics
Centre, Scho
ol
of Mechanic
a
l
E
ngi
neer
in
g, Beiji
ng Instit
ute of
T
e
chnol
og
y,
Beiji
ng, 10
00
8
1
, P.R.China
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: bitms@bit.ed
u
.cn
1
, sameh_
badr
an2
00
8@
ya
ho
o.com
2
,
sf.samo@gmai
l.com
3
, bit_hua
ngji
e
@
b
it.ed
u
.cn
4
A
b
st
r
a
ct
Enha
nce
m
ent
of fuel
co
nsu
m
pti
o
n
a
nd tr
ans
missi
on
eff
i
ciency
n
e
e
d
s
a c
onti
nuo
us
i
m
pr
oved
variator
perfor
m
a
n
ce
in co
nti
nuo
usly var
i
a
b
l
e
trans
missi
on
(CVT
). T
h
is pa
per focus
e
s o
n
the i
m
pr
ove
m
ent
of asli
p contro
l
l
er for a hy
dra
u
lica
lly
actuate
d
metal
push-
belt co
ntinu
o
u
s
ly varia
b
l
e
tra
n
smissio
n
(CV
T
),
usin
g
mo
de
l fo
r vari
ator dy
na
mic
in
the
CV
T
.
T
he sli
p
c
o
ntrol
purp
o
se
i
s
to i
m
prove
t
he
perfor
m
a
n
c
e
of
variator an
dto incre
a
se
th
e e
fficiency of
CV
T
by
d
e
ter
m
in
ation
the
lin
e
pressur
e
w
h
ic
h g
ener
ates th
e
cla
m
pi
ng force.
T
he selecti
on
of slip refere
nc
e-po
int
is taken
at the transitio
n
regi
on b
e
tw
een the
micro
a
n
d
macr
o sli
p
regi
on to gu
ara
n
te
e the maxi
mu
m vari
ator effici
ency. T
he ad
a
p
tive fu
zz
y
lo
gi
c control (F
LC)
and
Lin
ear Qua
d
ra
tic Regu
lator (
L
QR) control
l
e
r
s are app
lie
d
to control the
clampin
g
forc
e. T
he propos
ed
control systems are
des
igned to
ensur
e the exist
ence of
a slip v
a
lues
within
the region, which has
the
traction co
effici
ent
maxi
mu
m
valu
e, w
h
ile th
e lo
ad
disturb
a
n
ces ca
use
d
b
y
sudd
en
ly ch
ang
ed tor
q
u
e
s
in
the drive l
i
nes.
T
hese ap
pro
a
c
hes hav
e pote
n
tial for
the CV
T
efficiency improve
m
ent, as compar
ed to PID
control
l
er. T
h
e
ad
aptiv
e fu
zzy lo
gic c
ontro
l
techni
qu
e
us
e
s
a
s
i
mpl
e
gro
up of me
mber
ship
functi
ons an
d
rules to ac
hiev
e the des
ired c
ontrol re
qu
ire
m
ents of slip i
n
CVT. Simu
lati
on results show that satisfactory
slip i
m
prov
e
m
ent is ac
hiev
e
d
togeth
e
r w
i
th go
od
r
obust
ness a
gai
nst
sudd
enly c
h
a
n
ged tor
q
u
e
s. It is
further reve
ale
d
that all
ad
a
p
tive fu
zz
y
lo
gic cont
ro
l a
n
d
LQR contro
l
l
er hav
e a va
lua
b
le
effect on
mi
ni
mi
z
i
ng th
e slip a
m
ount a
n
d
max
i
mi
z
e
th
e
variator efficie
n
cy.
Ke
y
w
ords
: Pu
sh-Belt CVT
,
slip contro
l, ada
ptive fu
zz
y
log
i
c control, LQR control,
CVT
variator effici
enc
y
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
The research for imp
r
ov
ed fuel e
c
o
nom
y, red
u
ced emi
ssi
on
and imp
r
ov
ing the
efficien
cy of transmissio
n
are e
s
sential
chall
enge
s f
o
r the a
u
tom
o
tive indu
stry. Continu
o
u
s
ly
variable tra
n
smissi
on (CVT
) is one of the promi
s
ing
solutio
n
s for i
m
prove
d
fuel econo
my and
redu
ction
em
issi
on i
s
sue
s
due to
its
ability to
achieve mo
re
efficient op
erating level
s
for
comb
ustio
n
e
ngine
than
st
eppe
r tran
smi
ssi
on
s. Th
e transmi
ssion
ef
ficien
cy play
s a m
a
jo
r
role
i
n
the improve
m
ent of auto
m
otive powe
r
train effici
en
cy. Due to th
e con
s
tructio
n
of a CVT the
probl
em of sli
p
is always e
x
istent but the incr
ea
sing
slip ab
ove ce
rtain amo
unt has a
signifi
cant
unfavora
b
le e
ffect on the CVT performan
ce compa
r
e
d
to manual tra
n
smi
ssi
on [1, 2].
To p
r
event th
e belt
slip, th
e hig
h
cl
ampi
ng force
is a
pplied t
hat lo
wers th
e efficiency
of
CVT. In ad
dition, the hi
ghe
r cl
ampin
g
force n
eed
high
er hyd
r
auli
c
p
r
essu
re
s, the
r
eby le
ading
t
o
increa
se
pum
ping l
o
sse
s
.
So the
usi
ng
slip
co
ntrol
wi
ll lead
to o
p
e
r
ate th
e
CVT
in its
maximu
m
efficacy poi
nts and rest
rai
n
exce
ssive
belt slip am
o
unt [3]. By replaci
ng hydra
u
lic pa
rts by pur
electri
c
cont
rol system, the relia
bility and perf
o
rmance of CVT v
ehi
cular can hi
ghly be improv
ed,
while the fu
el
con
s
um
ption
and
cost
will
be furt
he
r d
e
crea
sed. T
h
e CVT mo
dul
ates velo
city by
the elect
r
oni
cally controlle
d mechani
cal
spee
d re
gula
t
or me
chani
sm co
st, fuel consumption
a
nd
failure rate c
a
n s
i
gnific
antly
be decreas
ed [4-6].
In this p
ape
r, adaptive fu
zzy l
ogic con
t
rol an
d line
a
r
Qu
ad
ratic
Reg
u
lator (L
QR) a
r
e
pre
s
ente
d
an
d comp
are
d
to the slip con
t
rol in Jatc
o-CK2 CVT. Th
e aim of the slip controller i
s
to
alleviate
the load
di
sturba
nce whi
c
h caused
by
su
ddenly to
rqu
e
ch
ang
e in
driveline
s
.
The
prop
osed
sli
p
co
ntroll
er f
o
r b
o
th me
n
t
ioned a
bove
are
sim
u
lat
ed in MAT
L
AB / SIMULINK
environ
ment.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 8, August 2014: 610
1 –
6110
6102
2. Working P
r
inciple of th
e Push-Belt
CVT
The Jat
c
o – CK2 pre
s
e
n
ted here is CVT wo
rked with a Van Doorne’
s Tran
smissi
o
n
metal pus
h
belt. In a metal V-belt CVT,
the torque
is tran
sferred f
r
om the
drive
r
to the d
r
ive
n
pulley by the belt eleme
n
ts pu
shing
action. Due
to the existence of frictio
n
betwee
n
b
e
lt
element
s an
d
band
s, the b
and, like flat rubb
er
belts
,
is al
so pa
rtici
pated in th
e tran
smi
ssi
on
of
torque.
The
r
e
f
ore, the
com
b
ined
of p
u
sh–pull
a
c
tion
in the
belt e
n
able
s
the
torque to
tra
n
smit
through the metal V-belt CVT s
y
s
t
em [7].
In Van Doo
r
n
e
’s metal pu
sh belt, the belt compo
s
ed o
f
a lot number (aroun
d 350
) of V-
sha
ped steel
block eleme
n
t
s,
detai
n
ed t
ogethe
r by
u
s
ing
a n
u
mb
er
(bet
ween
9 an
d 12
) of
thin
steel ten
s
ion
ring
s. The bel
t is hold by two pull
e
ys; at the engine
si
de namely th
e prima
r
y pull
e
y
and
at the
wheel
sid
e
n
a
m
ely the
se
conda
ry p
u
lle
y. The two
p
u
lleys
are
co
mposed
of o
ne
axially fixed
she
a
ve a
nd t
he oth
e
r mov
eable
s
heave
,
whi
c
h
a
c
tua
t
ed byme
an
s of a
hyd
r
auli
c
cylinde
r. Th
e
hydra
u
lic cy
l
i
nders
ca
n b
e
pressu
ri
zed
to cre
a
te
axi
a
l force
s
(thrusts o
r
clamp
i
ng
force
s
) on th
e belt and th
ey are e
s
sen
t
ial for tran
smissi
on the t
o
rqu
e
an
d ch
ange
of the ratio.
Shifting of the sh
eave
s
in
axial dire
ctio
ns vari
es th
e
runni
ng
radii
of the belt a
nd, hen
ce, th
e
transmissio
n ratio [7]. Figure 1 depi
cts a
sch
ematic
of
a metal V-be
lt CVT workin
g prin
ciple.
Figure 1. The
Continu
o
u
s
ly Variable T
r
a
n
sm
i
ssi
on Me
tal Push-belt Wo
rkin
g Prin
ciple
3. Slip Control Strategy
In Figure 2, the rel
a
tion b
e
twee
n tra
c
tion coefficie
n
t
µ
ef
f
, and vari
ator effici
ency again
s
t
the slip
ν
a
r
e
sho
w
n. The
signifi
can
c
e
of slip co
ntro
l is evident b
e
ca
use an in
cre
a
si
ng of slip
amount
will d
r
ive to an in
crea
sing i
n
tra
c
tion
coeffi
ci
ent in the mi
cro
-
slip regio
n
; therefo
r
e t
h
e
torque
will tra
n
smit efficie
n
t
l
y. In the cont
rary, in
the
m
a
cro-slip
regi
on, the sli
p
in
cre
a
se lea
d
s
to
destructive ef
fects in the
case of no a
c
ti
on take
n
to maintain the a
m
ount of slip
at the maximum
trac
tion c
oeffic
i
ent.
The maximu
m effective traction
coeffici
ent oc
cu
rs at
the turning p
o
int betwe
en
the two
regio
n
s. Th
e improvem
ent
of variator e
fficienc
y can
be achieved
near to this t
u
rnin
g point, as
s
h
ow
n
in
F
i
gu
r
e
2
.
The mo
st of clampi
ng fo
rce
strategie
s
are mainta
ined the
slip
amount
s where th
e
traction
co
efficient i
s
max
i
mum that le
ads to m
a
ximize the
efficien
cy of th
e variato
r
. Any
increa
se in th
e torque
will cau
s
e ex
cessive slip
but by modifying the clam
ping f
o
rce the slip
will
back to acce
ptable level so the damag
e
can be avoi
d
ed [8, 9].
The slip dyn
a
mic mo
del is req
u
ire
d
to des
ign the
slip controll
er which base
d
on the
dynamic m
o
d
ling by Bonsen et al. [3].
Figure 3
re
prese
n
ts the d
y
namic mo
de
l of CVT. The
relative slip
ν
betwe
en the
belt and pull
e
y is describ
ed
as:
.
1
g
s
r
r
(
1
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Slip Enhan
ce
m
ent in Continuou
sl
y Varia
b
le
Tra
n
sm
ission b
y
Usin
g
Adaptive…
(Ma Shuyuan
)
6103
Whe
r
e
r
s
de
n
o
te the spe
e
d
ratio and
r
g
i
s
geom
etri
c ratio.
Figure 2. Effective Frictio
n
Coeffici
ent an
d Efficiency versus the Slip
in Variator
Figure 3. Model of CVT on
Drive Train
The sp
eed
ra
tio
r
s
is described a
s
:
p
s
s
r
(
2
)
Whe
r
e
ω
p
an
d
ω
s
denote the prima
r
y and se
con
d
a
r
y angula
r
velocity,respectiv
e
ly. The
geomet
ric ratio
r
g
is assum
ed qua
si-stati
onary an
d ca
n be de
scribe
d by:
s
p
g
R
R
r
(
3
)
With
r
g
is qu
a
s
i –statio
nary
,
the dynamics of s
lip can b
e
derived u
s
i
ng (1
) and (3), lead to:
g
s
r
r
(
4
)
2
p
p
s
p
s
s
r
(
5
)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 8, August 2014: 610
1 –
6110
6104
As
sho
w
n
in
Figu
re
3 fro
m
the
engi
ne
sid
e
T
e
and
J
e
den
ote th
e en
gine
torque
and
corre
s
p
ondin
g
engine an
d
CVT inertia on the prima
r
y shaft resp
ectively, at wheel sid
e
T
d
and
J
s
denote the road torque
and iner
tia on
the secondary
shaft.
Th
e dynami
c
of
the primary
and
se
con
dary sh
aft of CVT va
riator
can b
e
rep
r
e
s
ente
d
by:
e
p
cvt
e
p
J
T
T
,
(
6
)
s
s
cvt
d
s
J
T
T
,
(
7
)
The
T
cvt,p
a
nd
T
cvt,s
denote
the torq
ue o
n
the p
r
ima
r
y and
se
con
d
a
r
y shaft respe
c
tively.
That torqu
e
s i
s
tran
smitted
via belt can b
e
cal
c
ulate
d
according to:
cos
)
,
(
2
,
,
,
g
eff
s
p
s
s
p
cvt
r
R
F
T
(
8
)
Substituting
Equation (1), (2), (3), (4
), and (5
) lead
s to:
e
e
e
g
eff
s
p
s
p
g
d
d
g
d
g
eff
s
p
s
p
J
T
J
r
R
F
r
J
T
r
J
r
R
F
)
cos(
)
,
(
2
)
1
(
)
cos(
)
,
(
2
1
,
,
(9)
For
control design, the sli
p
dynami
c
syste
m will be linearized,
so the
state space
illustratio
n
of
the
slip
dyn
a
mic mod
e
lin
g will
be
d
e
fined. T
he t
r
a
c
tion
co
effici
ent bet
wee
n
the
pulley and b
e
l
t is associ
ate
d
to the slip a
nd t
he ratio; therefo
r
e, it ca
n be de
scribe
d by:
)
(
)
(
)
,
(
2
1
g
i
g
i
g
eff
r
d
r
d
r
(
1
0
)
Whi
c
h is a pi
ece
w
i
s
e linea
r, the micro
-
sl
ip regime
rep
r
esents by
i
=1 and
i
=2 rep
r
esent
s
the macro-sli
p
regime. Th
e ratio relati
o
n
is taken a
c
count by choi
ce of
d
1i
and
d
2i
. Describin
g
the
state vector a
s
x
=
ν
, input vector
u
= [
F
s
T
e
T
d
] an
d output vect
or
y
=
x
, the d
y
namics ca
n be
pre
s
ente
d
, when the linea
rzed a
r
ou
nd a
definite working point
x
=
v
0
,with the linearized mod
e
l a
state sp
ace illustratio
n
of the slip dy
nami
c
model
will b
e
descri
bed a
s
follow:
u
B
x
A
x
s
s
(
1
1
)
x
C
y
s
(
1
2
)
e
p
cvt
e
p
J
T
T
,
(
1
3
)
e
g
i
e
e
p
J
i
k
r
F
k
F
J
T
A
2
0
0
0
1
0
0
0
0
1
(
1
4
)
T
g
d
e
e
i
g
i
i
p
r
J
J
J
k
r
k
k
B
0
0
,
2
0
0
0
,
1
0
0
0
,
2
0
0
0
1
)
1
(
1
(
1
5
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Slip Enhan
ce
m
ent in Continuou
sl
y Varia
b
le
Tra
n
sm
ission b
y
Usin
g
Adaptive…
(Ma Shuyuan
)
6105
1
C
(
1
6
)
cos
2
0
0
s
R
(
1
7
)
0
0
0
1
g
d
e
g
r
j
j
r
(
1
8
)
The derived slip dynami
c
will
be
used for
th
e controller design i
n
next
section. The
model ha
s
3
inputs
F
s
,
T
e
and
T
d
, but
o
n
ly
the se
con
dary clam
pin
g
force
F
s
ca
n be
controll
ed.
The inp
u
t torque
T
e
i
s
com
m
ande
d by th
e drive
r
via th
e throttle p
e
d
a
l and th
e out
put torq
ue
T
d
is
spe
c
ified
by the ro
ad
con
d
itions. T
herefore th
ey can be
co
nsi
d
ered
a
s
di
stu
r
ban
ce
s o
n
t
he
sy
st
em.
The second
a
r
y pre
s
sure cylinder i
s
co
n
necte
d to the
line pre
s
sure
and is
con
c
e
r
ned to
the se
con
d
a
r
y clampin
g
force
F
s
. The PWM si
gnal d
u
ty cycle is
a
c
counta
b
le of
the se
con
d
a
r
y
pre
s
sure dete
r
minatio
n whi
c
h
sets u
p
th
e clam
ping fo
rce, o
r
the lin
e pre
s
sure. T
he line p
r
e
s
sure
is b
oun
ded
b
e
twee
n 6.6
a
nd 4.2[b
a
r]
a
nd
can
be
co
nsid
ere
d
lin
e
a
r va
riation
to the
duty
cycle.
Due
to the
complex a
nd t
i
me wastin
g
dynamic mod
e
ling of li
ne
pre
s
sure
syst
em, Fre
que
n
c
y
Re
spo
n
se F
unctio
n
(F
RF) me
asure
m
ent is
pre
c
ed
ed by B
onson et
al. [3]. The sy
stem
estimation
fro
m
duty cy
cle
as in
put a
nd t
he line
pressure
as the o
u
tput ca
n p
r
e
s
ented by
a thi
r
d
orde
r lo
w pa
ss filter with
cu
t – off
freque
ncy of 6 [Hz].
4. The Con
t
r
o
l Design of
Slip D
y
namic Model
Due to
the
slip dyn
a
mi
cs relays on
the many va
riable
s
, it is difficult to
desi
gn
a
controlle
r whi
c
h is
cap
able
of achievem
ent t
he dema
nded p
e
rfo
r
m
ance. So, the lineari
z
in
g of
the slip dyna
mics i
s
do
ne
in different workin
g
point
s. The controller of sli
p
can be de
sig
n
ed
according
to
the line
a
ri
zed
slip
dynami
c
model
an
d t
he eval
uated
the
clampi
n
g
force
syste
m
transfe
r fu
nct
i
on. In this section, Ad
apt
ive Fuzz
y log
i
c
control an
d Line
ar
Qua
d
ratic Regul
a
t
or
(LQ
R
) a
r
e p
r
opo
sed a
nd explained. Th
e referen
c
e p
o
int of the slip controlle
r is ch
osen at the
turning
poi
nt betwe
en
th
e micro-a
nd m
a
cro slip
re
gi
on, ba
se
d o
n
the Fi
gure 2
.
Slip refe
ren
c
e
point is taken
depen
d on th
e ratio as
sho
w
n in Tabl
e 1
.
Table 1. Refe
ren
c
e Point o
f
Slip Depend
ent on the Ra
tio
Ratio [-]
v
re
f
[%]
0.43 2.5
1 1.5
2.25 1.5
4.1. Fuz
z
y
PI
D (FPID) Control
Fuzzy control
offers a fo
rm
al metho
dolo
g
y to charact
e
rize, ma
nipu
late, and i
m
pl
ement a
human’
s heu
ristic kn
owle
d
ge
ab
out cont
rol
a
sy
stem
.
As sho
w
n in
Figure 4, the
fuzzy
co
ntroll
er
block dia
g
ra
m is present
ed, whi
c
h ill
ustrate
s
a fu
zzy
controller embed
ded i
n
a clo
s
e
d
-lo
o
p
control syste
m
. The plant
outputs a
r
e repre
s
e
n
ted b
y
y(t
)
,
its inpu
ts are represented by
u(t)
, a
nd
r(t)
is
the referenc
e input to the fuz
z
y
controller.
The fuzzy co
ntrolle
r co
nsi
s
ts of four m
a
in com
pon
e
n
ts: (1)
“the
rule
-ba
s
e
”
ho
lds the
kno
w
le
dge, i
n
the sha
pe
of a grou
p of rule
s,
to ach
i
eve the best
control the system. (2) T
h
e
inferen
c
e m
e
cha
n
ism
esti
mates
whi
c
h
control rul
e
s
are a
ppli
c
abl
e at the cu
rre
nt time and then
make
s a d
e
cision
what th
e plant input
shoul
d be.
(3) The fu
zzification inte
rfa
c
e ad
apts th
e
inputs to ca
n be co
nst
r
ued an
d co
mpared to the rule
s in
the rule-ba
s
e an
d (4
) the
defuzzificatio
n
interfa
c
e
ch
ange
s the
co
nclu
sio
n
s
rea
c
he
d by the i
n
feren
c
e
me
chani
sm into t
h
e
inputs to the
plant [10].
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 8, August 2014: 610
1 –
6110
6106
Figure 4. Fuzzy Logi
c Co
ntrol Architectu
re
Table 2. Ada
p
tive Fuzzy Logic
Control Rule Ba
se
e
∆
e
N P
N LN
Z
P
Z LP
Figure 5(a
)
. Simulink m
odel
of the propo
sed A
daptive Fuzzy logic
control for
slip
model
Figure 5(b
)
Figure 5(c)
Figure 5(d
)
Figure 5(e
)
. T
he rule
s, me
mber fun
c
tion
s and
surfa
c
e of the
propo
se
d ad
aptive Fuzzy logic
control
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Slip Enhan
ce
m
ent in Continuou
sl
y Varia
b
le Tr
a
n
sm
ission b
y
Usin
g
Adaptive…
(Ma Shuyuan
)
6107
4.2. LQR Co
ntroller
The de
sign o
f
optimal con
t
rol system
s is an
impo
rta
n
t function of control e
ngi
neeri
ng.
The inte
ntion
of de
sign
is to re
co
gni
ze
a
syste
m
with reali
s
tic co
m
pone
nts
th
at will
p
r
ovide
s
t
he
desi
r
ed o
p
e
r
a
t
ing perfo
rma
n
ce.
The
de
sign
o
f
a
system
m
u
st b
e
ba
sed
on
minimizin
g
a perform
a
n
ce
s index. Systems
that are
adju
s
ted to provide
mi
nimum
performan
ce i
n
d
e
x are
often called o
p
timal
control
syste
m
sho
w
n in Fi
g
u
re 6. Lin
e
a
r
-qua
dratic-re
gulator
(L
Q
R
) is a
n
eleme
n
t of optimal control st
rate
gy
whi
c
h
ha
s b
e
en
widely
de
veloped
an
d
use
d
in
different ap
plicatio
ns. L
Q
R de
si
gn i
s
b
a
sed
on
the sel
e
ctio
n
of feedba
ck
gain
s
K
su
ch
that the co
st
function
J
i
s
minimized. T
h
is e
n
sure
s t
h
a
t
the gain sele
ction is o
p
timal for the co
st
function spe
c
ified [11].
Figure 6. Line
ar Qua
d
ratic
Reg
u
lator
(L
QR) with
State Feedba
ck
Table 3. Vari
ation of the feedba
ck gain
K
for variou
s op
erating p
o
int
Ratio
r
g
[-]
K
0.43 -0.5424
1
-0.74246
2.25
-0.912
For de
sig
n
L
Q
R
controller, the Matlab lqr fun
c
tion ca
n be u
s
ed to
cal
c
ulate the
value of
the vector of
feedba
ck g
a
i
n
K
which re
pre
s
ente
d
the feedba
ck
control la
w. That achi
eved
by
sele
cting o
u
tput and inp
u
t weight matri
c
e
s
Q
and
R
, as
R
=
1
and
Q
=
ρ
*C
T
*
C
wh
ere
CT
is
th
e
matrix tran
sp
ose of
C
from
state Equatio
n (14
)
. The control
si
gnal
can b
e
adju
st
ed by regul
a
ting
the value of
ρ
in Q matrix which i
s
done in m-file code.
R = 1;
Q =
ρ
* [1];
[K, S
,
e] =
lqr [A, B
,
Q, R];
So, by adjust
ed the value
of
ρ
= 80
0, the followin
g
value
s
of matri
x
K
are obtai
ned. If
ρ
is incre
a
se
d
even high
er,
the re
spo
n
se
of sy
stem wil
l
be improve
d
,
but the values of
ρ
= 800
is
cho
s
e
n
due t
o
it is achi
eved the de
sire
d req
u
irem
en
ts. The value
of feedba
ck
matrix
K
are
vary
with different
operating poi
nt of ratio
r
g
, that demon
strated by the Table 3.
Figure 7. Line
ar Qua
d
ratic
Reg
u
lator
(L
QR) with State Feedb
ack
In ord
e
r to
di
minish
the
st
eady state
error
of the
system
output a
nd tra
c
king
referen
c
e
inputs the int
e
rnal m
odel d
e
sig
n
techni
q
ue used, t
he basi
c pri
n
ci
pl
e of design th
e internal mo
del
is to in
se
rt an
integrator in
the feedfo
r
wa
rd
p
a
th bet
we
en the e
r
ror
compa
r
ator an
t the plant a
n
d
a value of
co
nstant g
a
in
k
I
sho
u
ld b
e
put
after the i
n
te
grato
r
. With
a
full-state fe
e
dba
ck
co
ntrol
l
er
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 8, August 2014: 610
1 –
6110
6108
all the state
s
are fee
dba
ck. The ste
ady-state va
lue
of the state
s
should
be calculated, multip
ly
that by the chosen g
a
in
K
, and ad
ded t
o
the e
rro
r
compa
r
ator
an
d
k
I
us
ed
a
ne
w
va
lue
as
th
e
control
sign
al
for
comp
utin
g the in
put. T
he integ
r
al
ga
in
k
I
ca
n b
e
o
b
tained
by u
s
ing m
-
file cod
e
.
The
block di
agra
m
of th
e
method
u
s
e
d
in
simul
a
tio
n
mod
e
l i
s
d
one
by expo
rted both
valu
e of
feedba
ck gai
n matrix
K
an
d integral gai
n
k
I
, as
sho
w
n in Fig
u
re 7. The Simuli
nk mo
del of
slip
dynamic
with LQR
cont
rol and u
s
ing int
e
rnal m
odel d
e
sig
n
tech
niq
ue is sho
w
n i
n
Figure 8.
Figure 8. Simulink Mo
del o
f
Slip Dynamic with
LQ
R Control an
d usi
ng Internal M
odel Desi
gn
Tech
niqu
e
5. Simulation Resul
t
s an
d Discus
s
io
n
In ord
e
r to
in
vestigate
and
evaluation
th
e pe
rform
a
n
c
e of the
prop
ose
d
two con
t
rollers
(LQ
R
, a
dapti
v
e fuzzy lo
gic
cont
rol) an
d comp
are
d
with PI to
va
lidate the
ro
bustn
ess
of t
he
prop
osed con
t
rol strate
gy, the simul
a
tion
test is con
d
u
c
ted.
As sh
own in Figure 7, the re
spon
se of
prop
osed con
t
rol sy
stem
wi
th PI, Adaptive fuzzy
logic an
d the linear q
uad
ra
tic regul
ator (LQR) ar
e de
monst
r
ated.
Acco
rdi
ng to the Figure 7, the
results obvio
usly app
eare
d
that
Adaptive Fuzzy logic co
ntrol ha
s the fastest resp
on
se with
the
risin
g
time
of
0.116 [
s
] an
d
settling
time
of 2.
14 [
s
]. For the
pe
rcen
t of oversho
o
t
(%), LQ
R
h
a
s
the minim val
ue 0.5%
whi
c
h a
c
hieve
s
th
e de
sired
req
u
irem
ent of
controlle
r d
e
si
gn. In a
dditio
n
the LQ
R
con
t
roller te
nd
s
to gene
rate
very small
st
eady state
e
rro
r (E
ss), i
s
within th
e li
mit
0.01%. This can be
signif
y
ing that LQR cont
rolle
r has the abilit
y to attenuate the effect of
disturban
ce
s
in the system
.
Figure 9. Step Re
spo
n
se of Slip Model wi
th LQ
R, Adaptive Fuzzy Logi
c and PI Control
0
1
2
3
4
5
6
7
8
0
0.5
1
1.5
T
i
m
e
[s
]
A
m
pl
i
t
u
de
St
e
p
PI
AF
L
C
LQ
R
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Slip Enhan
ce
m
ent in Continuou
sl
y Varia
b
le
Tra
n
sm
ission b
y
Usin
g
Adaptive…
(Ma Shuyuan
)
6109
Figure 10(a).
The engi
ne to
rque
T
e
disturban
ce
Figure 10(b).
Re
spo
n
se of the slip mod
e
l with
PI control an
d AFLC
Figure 10(c). Re
spo
n
se of the Clampi
ng
Force
F
S
with PI Control a
nd AFLC
Figure 11(a).
Re
spo
n
se of the slip mod
e
l
with
PI control, AFLC an
d LQ
R control
Figure 11(b).
Re
spo
n
se of clampi
ng force
F
S
with PI contro
l, AFLC and L
Q
R control
From Fi
gu
re
9, 10, and
11
, it can b
e
rea
lized th
at Ada
p
tive Fuzzy l
ogic
co
ntrol
a
nd LQ
R
controlle
rs a
r
e adequ
ate to utilize in the contro
l of slip dynami
c
model sy
ste
m
and gave
the
simila
r results a
s
th
e PI
controlle
r. Whatever,
due
to the
high
e
r
g
a
in
of LQ
R
cont
rolle
r,
the
clampi
ng fo
rces a
r
e
rea
c
h
ed to a mu
ch
highe
r level
durin
g sudd
e
n
ly cha
nging
torque
co
mpa
r
e
d
to the clampi
ng forces b
a
sed on PI cont
ro
lle
r and Ad
aptive Fuzzy logic
controlle
r.
Ho
wever, th
e
re
sults prove
n
that Ada
p
tive
Fuzzy lo
gi
c
controller a
nd L
Q
R
co
ntroller
are
give signifi
ca
nt performan
ce bette
r than
PI control in slip control of
CVT.
6. Conclusio
n
In this pape
r,
two cont
rolle
rs, Adaptive
Fuzzy logic
a
nd LQ
R for sl
ip cont
rol in
CVT are
prop
osed a
n
d
tested
by si
mulation u
s
in
g MATLAB.
Also the t
w
o
controlle
rs
are co
mpa
r
ed
wit
h
PI cont
rol.
Base
d o
n
th
e
result
and
the
analysi
s
, the control of approa
ch fu
zzy PID control a
n
d
LQR i
s
cap
a
b
l
e on
controlli
ng the
slip o
n
the variat
o
r
of CVT. The
slip
controlle
r based o
n
fuzzy
PID and
LQ
R
cont
rol d
e
sig
n
afford
better di
st
urb
ance alleviati
on, whi
c
h
ca
use
d
by to
rq
ue
pea
ks,
com
p
ared
to PI
controlle
r. Simulation
an
d an
alysi
s
result
s
sho
w
that a
better
15
20
25
30
35
40
40
60
80
100
120
Ti
m
e
[
s
]
T
e
[s
]
15
20
25
30
35
40
-5
0
5
10
Ti
m
e
[
s
]
Sl
i
p
[
%
]
P
I
c
ont
r
o
l
Ad
a
p
t
i
v
e
F
L
C
Re
f
e
rn
ce
15
20
25
30
35
40
10
12
14
16
18
Ti
m
e
[
s
]
F
s
[k
N
]
P
I
cont
r
o
l
A
d
a
p
tiv
e
F
L
C
15
20
25
30
35
-8
-6
-4
-2
0
2
4
6
8
10
Ti
m
e
[
s
]
Sl
i
p
[
%
]
LQR
Re
f
e
r
n
c
e
PI
AF
L
C
28
30
32
34
36
38
40
42
44
8
10
12
14
8
10
Ti
m
e
[s
]
Fs
[
K
N
]
PI
AF
L
C
LQ
R
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 8, August 2014: 610
1 –
6110
6110
perfo
rman
ce i
s
achieved
b
y
the LQR wit
h
re
spe
c
t
to b
o
th adaptive f
u
zzy cont
rol
strategy a
nd
PI
controlle
rs i
n
cont
rolling
the sli
p
, so
the imp
r
oved
efficien
cy fo
r CVT i
s
m
a
i
n
tained
and
the
improve
d
efficien
cy for CV
T is achieved
by
controlli
n
g
the clam
pin
g
force and p
r
events
exce
ss
slip.
Referen
ces
[1]
T
i
an Jinsi, Su
Jiam.
Hydr
aul
i
c
System S
i
mulati
on
of CVT
w
i
th F
u
zz
y
L
ogic
Contr
o
ll
er
s
. Shan
gha
i.
200
5; 136-
141.
[2]
Lei Z
h
a
ng, Xia
o
mei Co
ng, Hu
jian Pa
n, Z
uge
Cai,
Xiumi
n
Ya
ng. T
he Control S
y
stem Mod
e
ling a
nd the
Mechanical S
t
ructure Analysis for EMCVT.
T
E
LKOMNIKA Indon
es
ian Jo
urn
a
l
of Electrical
Engi
neer
in
g
. 2013; 11(
7): 415
9-41
67.
[3]
Bonse
n
B, K
l
a
a
ssen
T
W
GL,
Pull
es RJ, S
i
m
ons
SW
H, St
ei
nbuc
h M, V
e
e
nhu
izen
PA. P
e
rformanc
e
optimiz
ation
of the pus
h-be
lt CVT
b
y
vari
ator
slip co
ntrol.
Internati
ona
l Jour
nal of ve
hicl
e d
e
sig
n
. 200
5;
39(3): 23
2-2
5
6
.
[4]
Guang
qi
ang
W
u
,
Xia
n
a
n
S
un.
Dev
e
lo
pme
n
t
surve
y
of a
u
to
mobil
e
c
onti
n
u
ousl
y
vari
abl
e
transmissio
n
techni
qu
e and
appl
icatio
n a
s
pects.
Journ
a
l
of T
ongji U
n
iversity (Natur
al Scie
nce)
.
2
009;
37(
12):
164
2-16
47.
[5]
SONG Yunpu.
Stud
y
on the
Energ
y
-R
eg
e
nerati
on-
b
a
se
d
Velocit
y
C
ont
rol of the H
y
d
r
aulic-
H
ybri
d
Vehic
l
e.
T
E
LKOMNIKA Indon
esia
n Journ
a
l o
f
Electrical Eng
i
ne
erin
g
. 201
2; 10(7): 170
0-1
707
[6]
Naish
i
C
hen
g,
W
e
ihu
a
Zha
n
g
, Hui
lin Y
a
n
g
. T
he princi
ple
and
des
ign
of
metal b
e
lt CV
T. Beijin
g of
Chin
a: Mecha
n
i
cal Ind
u
str
y
Pr
ess. 2010.
[7]
Lei Z
h
ang, H
u
i
w
u
Li, Gu
oni
ng
Li.
T
he Co
ntro
l T
heory a
nd I
m
p
l
e
m
e
n
tatio
n
Method
of Cla
mp
in
g F
o
rce.
Shan
gh
ai. 200
9; 422-4
26.
[8]
Simons SW
H. Shift d
y
n
a
mic
s
model
ing for
optimiz
ati
on
of variator sli
p
control in a
contin
ous
al
y
varia
b
le trans
mission. Maste
r
s
T
hesis. Eind
hove
n
: Eind
ho
ven Un
iversit
y
of
T
e
chnol
og
y.
2006.
[9]
SONG Yunpu.
Parameter D
e
sig
n
for the Energ
y
-
R
eg
en
eratio
n S
y
ste
m
of Series H
y
dr
au
lic-H
ybr
i
d
Bus.
T
E
LKOMNIKA Indon
esi
an Jour
nal
of Electrical E
ngi
ne
erin
g.
201
2; 10
(6): 1293-
13
00
.
[10]
Kevin M. Passi
no, Stephe
n Y
u
rkov
ic. F
u
zz
y
Contro
l in cont
rol. Californ
i
a:
Addiso
n W
e
sl
e
y
L
o
n
g
man,
Inc. 1998.
[11]
Katsuhik
o
Oga
t
a. Modern Co
ntro
l Eng
i
n
eeri
ng in co
ntrol 3r
d ed. Ne
w
Jers
e
y
: Pre
n
tice H
a
ll Inc.19
97.
Evaluation Warning : The document was created with Spire.PDF for Python.