TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 15, No. 2, August 201
5, pp. 284 ~
293
DOI: 10.115
9
1
/telkomni
ka.
v
15i2.815
8
284
Re
cei
v
ed Ma
y 19, 201
5; Revi
sed
Jun
e
17, 2015; Accepted July 1
0
,
2015
Resear
ch on Roll Variable Speed Grinding Based on
Adaptive Fuzzy Control
Bi Junxi*
1
, Gao Qing
2
, W
a
ng Shu
w
e
i
3
Mecha
n
ica
l
En
gin
eeri
ng Instit
ute of Inner
Mo
ngo
lia U
n
ivers
i
t
y
of
T
e
chnol
og
y
,
Chi
na
*Corres
p
o
ndi
n
g
author
, e-ma
i
l
: jun
x
i
b
i@im
ut.edu.cn
1
, ga
o15
034
98@s
i
n
a
.com
2
,
ws
w1
1
0
1
@
q
q
.
c
o
m
3
A
b
st
r
a
ct
T
he traditio
n
a
l
meth
od
of imp
r
ovin
g the
mac
h
ini
ng
precis
io
n is main
ly on
accou
n
t of cha
nges i
n
the w
o
rk piece
and the gri
ndi
ng w
heel i
n
gri
ndi
ng proc
e
ss, includ
in
g, con
s
ider
ed
the w
eakest of the wor
k
piec
e fro
m
th
e
axia
l stiffness
,
usin
g
ce
nter
frame
to i
n
cre
a
se th
e stiffne
ss of the
w
o
rk piec
e, take
er
ro
r
compe
n
satio
n
,
and an
aly
z
e
t
he mec
h
a
n
is
m of gear gr
in
din
g
and the s
uppr
essio
n
me
asures of varia
b
le
spee
d gr
in
din
g
an
d gr
ind
i
n
g
chatter. Bas
e
d o
n
the
a
nal
ysis of th
e re
l
a
tions
hip
a
m
o
ng
gear
gri
n
d
i
ng
mec
h
a
n
is
m, va
riabl
e sp
ee
d gr
indi
ng
an
d gri
n
din
g
chatte
r s
u
ppress
i
on
an
d
grin
din
g
prec
isi
on, the var
i
a
b
l
e
spee
d
micr
o-fe
ed
grin
di
ng
pro
c
ess strategy
b
a
sed
o
n
a
d
a
p
ti
ve fu
zz
y
c
ontro
l is
put
up,
that
is
accord
in
g t
o
the roll
er mate
rial a
nd structure to reaso
n
a
b
ly det
er
mine
the character
i
s
t
ics of grindi
ng
w
heel an
d th
e
parameters of
grinding pr
oc
ess. And with full consideration of roll
gr
inders, changes
of roller system along
the axia
l stiffness and the d
e
formatio
n
contr
o
l in the ro
ll gri
ndi
ng proc
ess,
a new
strategy of variab
le spe
e
d
grin
din
g
micro-
feed
a
d
a
p
tive cont
rol
opti
m
i
z
ation
bas
ed o
n
ada
ptive fu
zz
y
control
is pro
p
o
sed, w
h
ich
C
a
n
effectively i
m
pr
ove the rol
l
gri
ndi
ng prec
isi
o
n
.
Ke
y
w
ords
:
ad
aptive fu
zz
y
co
ntrol, trans
miss
ion, rol
l
grin
din
g
, micro fe
ed
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
There a
r
e
th
ree
main
tra
d
itional m
e
th
ods
to im
prove the
ma
chining
preci
s
ion roll
grindi
ng. The
first is to co
nsid
er the
rol
l
axial
wea
k
stiffness, to select the rota
tional sp
eed
of
work pi
ece
w
V
and the longitu
dinal feed
sp
eed
f
V
con
s
tant
spe
ed g
r
indin
g
, at the sam
e
time, the
grindi
ng
dept
h as small a
s
possi
ble. Th
e param
eters
w
V
and
f
V
relative to the larger sti
ffness of
the point of t
he value i
s
not likely to be the mo
st advantages, like
this
will i
nevitably bring about
the gri
ndin
g
co
st and
waste of time. Seco
nd, by
u
s
i
ng the
metho
d
of cente
r
frame to i
n
cre
a
se
the ri
gidity of
the work
pie
c
e, the
pro
c
e
s
sing
effi
cie
n
cy is l
o
w, th
e
a
d
justme
nt i
s
complex ,b
ut the
highe
r re
quirement on the
technol
ogy worke
r
s, impr
ove the machining p
r
e
c
isi
on of the rol
e
is
limited. Third
is the two
strategi
es
ab
out
the error comp
en
sati
on mea
s
u
r
e
s
, the process
comp
en
satio
n
strategy of
adju
s
t or
rep
a
ir an
d t
he group strate
gy
of
sele
ct
ion,
cal
c
ulate
d
by
the
formula
of gri
nding
qua
ntity for ea
ch p
r
oce
s
sing
poi
nt, for make t
he work pie
c
e’s a
c
tual
am
ount
of grindin
g
to achieve
cal
c
ulation in ma
chini
ng,
in order to re
du
ce the influen
ce of the ela
s
tic
deform
a
tion.
But it is ve
ry
difficult that
calcul
at
ed
a ti
me in
gri
ndin
g
p
r
o
c
ess of
grindi
ng
and
the
elasti
c deformation.
In this p
ape
r,
ba
sed
on th
e stu
d
y of the
mechani
sm
of variabl
e
sp
eed
grin
ding,
variable
spe
ed
grin
din
g
an
d g
r
indi
n
g
trem
or
su
p
p
re
ssi
on
grin
ding
pre
c
i
s
io
n on
the
relati
onship An
d in
full
con
s
id
eratio
n
,
the roll g
r
in
der
and th
e
cha
nge
alon
g the roll axi
a
l stiffness
o
f
roll sy
stem
roll
deform
a
tion i
n
grindi
ng, b
a
se
d on cont
rol the roll
d
e
f
ormation in
grindi
ng, thro
ugh the control of
grindi
ng force, finished th
e grindi
ng sp
eed, pro
p
o
s
e
d
The optimi
z
ation
strateg
y
of roll grind
i
ng
spe
ed intellig
ent adaptive
control ba
se
d
on fuzzy di
re
ct adaptive control.
2.
The Roll of V
a
riable Spee
d Grinding
The esse
nce
of variable speed g
r
indi
n
g
is
co
ntinuo
usly varying the sp
eed of
grindi
ng
whe
e
l, the lo
ngitudin
a
l fee
d
speed,
roll
spe
ed, n
o
t to let the g
r
in
ding flutter vib
r
ation al
ways
in
the maximum
flutter gro
w
t
h
rate of
co
rresp
ondi
ng
freque
ncy flutter, but the fl
utter growth
rate
increa
se in t
he nea
r maxi
mum flutter rate co
ntinuou
s chang
e, so
as to en
su
re
in the variab
le
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Re
sea
r
ch on
Roll Vari
able
Speed G
r
indi
ng Base
d on
Adaptive F
u
zzy
Cont
rol (Bi
Jun
x
i)
285
spe
ed du
ring
grindin
g
the
averag
e flutter gro
w
th
rate less than ma
ximum flutter gro
w
th rate,
in
orde
r to inhibi
t or delay the purp
o
se of the chatte
r gro
w
th, to impro
v
e the grindin
g
accuracy [1
].
Whe
n
the
co
ntinuou
s
ch
a
nge
of the
work pie
c
e
sp
eed
and
gri
n
ding
wh
eel
speed, the
ratio will
cha
nge continu
o
u
sly, at the same time
, the radi
al force
also i
s
co
ntinuou
sly ch
an
ged
[2]. Optimizat
i
on of th
e
sy
stem to
ma
ke an
alysi
s
a
nd p
r
o
c
e
ssi
n
g
on
the
nu
meri
cal va
riat
ion,
adju
s
ting th
e
wo
rk pie
c
e
spe
ed
and
the o
u
tput of
wh
eel
sp
ee
d, cycl
e to
b
egin, e
n
sure
to
sup
p
re
ss
tre
m
or and al
ways
re
ach the
optimal
grindi
ng. Th
e gri
ndin
g
p
r
ocess
of e
a
ch
comp
one
nt in the directio
n of velocity is sho
w
n in Fig
u
re 1.
w
V
f
V
g
V
mr
V
Figure 1. Schematic di
ag
ram of roll grin
ding work
2.1. Variable Speed Grind
i
ng and the
Amoun
t of F
eed
The
grin
ding
rolle
r i
s
a
kind of
sp
eci
a
l gri
nding
p
r
oce
s
s fo
rme
d
bet
wee
n
CNC roll
grind
e
r, g
r
ind
i
ng wh
eel
s, and the roll
er [
3
]. Theref
o
r
e,
in the study of roll grin
din
g
mechani
sm
,
contem
plated
unde
r the g
r
inding g
eom
e
t
ry relation
s,
con
s
id
erin
g the influen
ce
of roll g
r
indin
g
motion pa
ram
e
ters on the
contact a
r
c len
g
th, derivat
io
n of mathem
atical mo
del f
o
r the m
o
tion
of
grindi
ng roll contact len
g
th (as
sho
w
n in
Figure 2 belo
w
).
Figure 2. Roll
er and
whe
e
l conta
c
t arc le
ngth
2
/
1
2
/
1
2
2
]
[
]
)
60
(
)
60
1
[(
w
s
p
w
s
s
w
s
w
r
k
d
d
a
d
d
V
V
f
V
V
K
l
(1)
In formula,
k
l
-
The motio
n
o
f
grindin
g
roll
er conta
c
t arc length corre
c
tion value
(
mm
);
w
V
-
Roll
er line speed
(
m/
s
);
s
V
-
The sp
eed of the
grindi
ng whe
e
l
(
m/s
);
s
d
-
Grindi
ng
whe
e
l dia
m
eter
(
mm
);
w
d
-
Roll di
ameter
(
mm
);
p
a
-
Grin
ding
depth
(
mm
);
f
-
Lon
gitudin
a
l
feed
(
mm
/m
i
n
);
r
K
-
The
re
vision
coeffici
ent rel
a
ted to
roll
sh
ape
;
“
+
”
-
M
ean
s
"counte
r
clo
c
kwise gri
ndin
g
"
, The grin
din
g
wh
eel an
d the roll
er
spe
e
d
in the op
po
site directio
n
of
the grin
ding
zone
;
“
-
”
-
M
e
ans "
c
lo
ckwise grindi
ng", T
he grin
ding
wheel an
d the rolle
r sp
eed i
n
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 2, August 2015 : 284 –
293
286
the sam
e
di
rection
of grin
ding
zon
e
.
s
w
V
V
/
is reflect that du
e to the roll line spee
d
w
V
c
a
us
ed
the conta
c
t a
r
c len
g
th ch
a
nge
s. Excludi
ng the
impa
ct of the linear spe
ed of the rolle
r(if
0
w
V
),the conta
c
t arc length
k
l
be
come
s the
ge
ometri
c len
g
th of co
ntact a
r
c
g
l
for the g
r
in
ding
whe
e
l
and the roll
er
[1].
2.2. Variable Speed Grind
i
ng and the
Grinding For
ce
From th
e re
gene
rative ef
fect theo
ry of se
lf-excite
d
vibration
grindi
ng, cha
nge
s of
grindi
ng
chatt
e
r is
mainly
caused by the
corru
gat
ed
surface of the
rolle
r or
grin
d
i
ng wheel [4].
For
the reg
e
n
e
ration effect of
roll,
the sp
ecific pr
o
c
e
ss is a
s
follo
ws:
In the roll g
r
i
nding
pro
c
e
s
s,
Whe
n
cutting
the roll su
rfa
c
e produ
ced
a
ri
pple
in
t
h
e
previou
s
rota
ry, the g
r
in
din
g
wheel
will
b
e
cutting in th
e
su
rface with
corru
gated i
n
the
behi
nd
rotary, and th
e ne
w ri
pple
formation
on
the
surfa
c
e
of the rolle
r. If the self
vibration freq
uen
cy
and velo
city of roll into i
n
teger
ratio,
the
grindi
ng dept
h is basi
c
ally
uncha
nge
d, the grindi
ng
force is ba
sically unchan
ged; if the self
vibration frequency and ve
locity of
roll
not into integer
ra
tio, it
will produce
dynamic gri
n
ding
force.
Ch
ang
e of rolling
speed
wh
en t
he vari
able
speed
gri
ndin
g
, it make th
e roll
er
su
rfa
c
e
corru
gation
p
hase
chan
gin
g
, it is p
o
ssibl
e
to ma
ke
th
e sy
stem
con
s
tantly leave
flutter excitati
on
regio
n
, thus suppress the flutter, improv
e
the surfa
c
e q
uality of grind
i
ng roll [5].
So as long a
s
the variable
amplitude, fre
quen
cy app
ro
priate when v
a
riabl
e grin
di
ng, the
grindi
ng
syst
em can in
a
perio
d of tim
e
to avoid
flutter exc
i
tation region, to get better
results
than the co
nstant spee
d gri
nding.
In rollers whi
c
h are the gri
nding ratio of length to diameter
d
l
/
is different, the syst
em
stiffness influ
ence on g
r
in
ding a
c
cura
cy will
be not the sam
e
, long si
ze big
g
e
r
impa
ct
[6]
.In this
pape
r, the st
rategy of ad
aptiv
e co
ntrol
of variable
spe
ed g
r
indi
ng which
ba
sed
on the f
u
zzy
dire
ct ada
ptive control consi
deri
ng th
e roll
sy
ste
m
of roll g
r
i
nder
ch
ang
e
s
alo
ng the
axial
stiffness of th
e roll ,it ca
n control the
def
ormatio
n
of
g
r
indin
g
roll
s. In the vertical
grindi
ng rolle
r,
the deform
a
tion of the rolle
rs in poi
nt
of grindi
ng force
can be exp
r
e
s
sed a
s
:
2
1
2
1
2
1
2
)
2
(
)
1
)(
(
)
2
(
)
1
)(
(
]
[
tan
8
rR
R
r
V
V
V
K
K
rR
R
r
V
V
V
rkK
F
K
Y
S
W
f
t
a
S
W
f
p
n
t
(
2
)
2
2
)
2
(
tan
8
rR
R
r
rkK
K
a
(3)
H
T
t
K
l
x
l
K
l
x
l
d
E
x
l
x
K
2
2
2
2
4
2
2
)
(
)
64
/
(
3
)
(
(4)
In formula,
Y
is t
he
roll d
e
formation e
r
ror
of dista
n
ce from the
ma
chi
ne's top
x;
k
is The
tangential coefficient;
r
is the Minimum
radiu
s
;
is th
e tangential
angle;
k
is deformation
coeffici
ent;
is angul
ar
velo
city;
is constant;
p
is depth
o
f
cut ;
S
V
is roll
speed;
R
is the
maximum radius of
roll
er;
a
K
is co
rrectio
n
co
efficient of lo
ngitudin
a
l;
n
F
is
norm
a
l g
r
indi
ng force;
E
is modul
us of
elasticity;
T
K
is the rigidity of the tail stock;
H
K
is frame
stiffness;
l
is roll length.
By the empirical mod
e
l by grindi
ng force
,
in t
he proce
ss of roll g
r
in
ding no
rmal
grindi
ng force
n
F
(as
sho
w
n in
Figure 3
)
ca
n be expre
ssed as:
2
1
2
1
1
)
(
1
e
s
w
N
s
w
n
D
a
V
V
C
a
A
P
a
V
V
K
F
(
5
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Re
sea
r
ch on
Roll Vari
able
Speed G
r
indi
ng Base
d on
Adaptive F
u
zzy
Cont
rol (Bi
Jun
x
i)
287
In formula,
a
is the flutter growth rate;
is mean valu
e
of index;
N
A
is
normal
flutte
r
amplitude;
1
C
is cutting co
efficient (al
s
o kn
own a
s
the consta
nt of proportio
nality);
is con
s
tant;
is grin
ding thi
c
kne
ss;
e
D
is grin
ding diam
eter.
Figure 3. Sch
e
matic dia
g
ra
m of the grind
i
ng force in grinding p
r
o
c
e
s
s of the roll
Acco
rdi
ng to
the longitu
dinal feed
speed i
s
f
V
, fro
m
the relatio
n
w
f
V
V
r
f
2
betwe
en lon
g
i
tudinal feed
per revolution
of roll
f
and th
e type (13),
whe
n
the
s
V
c
ons
tant, afte
r
grindi
ng thi
ckness
is determined.
it can
ch
ang
e
w
V
and
f
V
to co
ntrol th
e
roll
defo
r
mat
i
on. In
pra
c
tice, du
e
to the limita
t
ion of the grindin
g
system by many
conditions, for any roll axial
positio
n, there exists a be
st of
w
V
and
f
V
[7].
In summ
ary, the propo
se
d
control
strate
gy is
in the
grinding
pro
c
e
s
s,the relative
motion
of roller
and
wheel, by usi
ng the direct
adaptive fu
zzy control, real time optimization control
of
grindi
ng para
m
eters
w
V
and
f
V
,throug
h conti
nuou
s cha
n
g
e
s of
w
V
an
d
f
V
to c
o
ntrol the
grindi
ng force of points, t
o
co
ntrol the
deform
a
tion
of the wo
rk
p
i
ece, in o
r
d
e
r to limit erro
r in
rang
e of pro
c
ess paramete
r
s i
s
allowed.
3.
Direc
t
Ada
p
tiv
e
Fuzz
y
Control
Dire
ct a
daptive fuzzy control is fu
zzy log
i
c
sy
stem
s wi
th adaptive le
arnin
g
alg
o
rit
h
m, the
learni
ng
algo
rithm i
s
relying on
data
i
n
formatio
n to
adju
s
t the
p
a
ram
e
ters of
the fuzzy lo
gic
system, an
d i
t
can gu
arant
ee the st
a
b
ility of the control system [8].
The sy
stem
of adaptive is as
follows
.
3.1. Problem Descrip
tion
Con
s
id
erin
g the re
sea
r
ch o
b
ject de
scrib
ed by the followin
g
equati
o
n
:
bu
x
x
x
f
x
n
n
)
,
,
,
(
)
1
(
)
(
(6)
x
y
In formula,
f
is
function
s of f
o
rmul
a (5)
o
n
the
rotation
al sp
eed
of t
he rolls
S
V
,
b
is the
norm
a
l numb
e
r of un
kno
w
n.
Dire
ct ada
ptive fuzzy co
n
t
rol use the f
o
llo
win
g
IF-T
HEN fuzzy rules to de
scribe the
control kno
w
l
edge, Th
at is: if
1
x
is
r
P
1
and
n
x
is
r
n
P
, then
u
is
r
Q
, in formul
a:
r
P
i
,
r
Q
are f
u
zzy
set in the set
of real num
be
rs, and
u
L
,
2
,
1
r
.
Set the position com
m
an
d is
m
y
,
let,
T
n
m
m
e
e
e
x
y
y
y
e
)
,
,
,
(
,
)
1
(
e
(
7
)
Selec
t
T
n
k
k
)
,
(
1
K
,
It is a polynomial
n
n
n
k
s
k
s
)
1
(
1
)
(
all roots in th
e left half plane
of the comple
x plane. The control law fo
r the sele
ctio
n
:
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 2, August 2015 : 284 –
293
288
]
)
(
[
1
)
(
*
e
K
x
T
n
m
y
f
b
u
(8)
The form
ula (8) into formul
a (6).G
e
t t
he clo
s
ed lo
op control sy
stem
of equation:
0
)
1
(
1
)
(
e
k
e
k
e
n
n
n
(9)
By the selecti
on of
K
can be obtaine
d
0
)
(
t
e
w
hen
t
, that is the output of the
sy
st
em
y
grad
u
a
l conve
r
ge
n
c
e to the de
si
red outp
u
t
m
y
.
Dire
ct ad
apti
v
e fuzzy con
t
rol to
de
sig
n
a feedb
ack controll
er
)
|
(
x
u
u
and the
adaptive rule
is an
adju
s
t
m
ent pa
rame
ter vecto
r
θ
based
on fu
zzy
system,
so t
hat the outp
u
t
y
of the system
as mu
ch a
s
possi
bl
e to track the de
sired output
m
y
[10].
3.2. The Con
t
roller De
sig
n
Dire
ct ada
ptive fuzzy controller
:
)
|
(
θ
x
D
u
u
(10)
In formula,
D
u
is a f
u
zzy
sy
st
e
m
s
,
θ
is the adj
ustabl
e para
m
eters.
Fuzz
y s
y
s
t
ems
D
u
is compo
s
ed
of two
step
s to
co
nstru
c
t. Step
1: The va
riable
)
,
2
,
1
(
n
i
x
i
is d
e
fined
i
m
fuzz
y sets
)
,
2
,
1
(
i
i
l
I
m
l
A
i
.
S
t
ep 2:
S
t
ruct
u
r
e
o
f
f
u
zzy
sy
st
e
m
)
|
(
θ
x
D
u
with the follo
wing
n
i
i
m
1
rules, That is: if
1
x
is
i
l
I
A
and......and
n
x
is
n
l
n
A
,then
D
u
is
n
l
l
S
1
, in formula:
i
i
m
l
,
2
,
1
,
n
i
,
2
,
1
.
Usi
ng the p
r
odu
ct infere
n
c
e en
gine, th
e singl
e-val
u
ed fuzzy co
n
t
rol and th
e averag
e
s
o
lution c
e
nter fuzz
y c
o
ntroller
to des
i
gn fuz
z
y
controller [11],
)
)
(
(
)
)
(
(
)
|
(
1
1
1
1
1
1
1
1
1
1
1
n
i
i
L
A
m
l
m
l
n
i
i
L
A
m
l
l
l
u
m
l
D
x
u
x
u
y
u
I
I
n
n
I
I
n
n
n
θ
x
(
1
1
)
Let
n
l
l
u
y
1
is the free
param
eters
,
on the
n
i
i
m
R
1
θ
coll
ecti
on, the fuzzy
controlle
r is:
)
ξ
(
x
θ
θ
x
T
D
u
)
|
(
(12)
In formula,
)
ξ
(
x
is
n
i
i
m
1
d
i
me
ns
io
na
l ve
c
t
o
r
,
t
he
n
l
l
,
,
1
elements a
r
e:
)
)
(
(
)
(
)
(
1
1
1
1
,
,
1
1
1
n
i
i
L
A
m
l
m
l
n
i
i
L
A
l
l
x
u
x
u
I
I
n
n
I
I
n
x
(
1
3
)
Among them
, fuzzy control rule is em
bedd
ed into
the fuzzy co
ntrolle
r, by setting the initial
para
m
eters.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Re
sea
r
ch on
Roll Vari
able
Speed G
r
indi
ng Base
d on
Adaptive F
u
zzy
Cont
rol (Bi
Jun
x
i)
289
3.3. The Desi
gn of the
Ad
aptiv
e
La
w
By
formula (8
):
e
K
b
x
T
n
m
y
u
f
)
(
*
)
(
(14)
De
sign of ad
aptive law
:
)
(
)
x
Pb
e
ξ
T
(15)
In formula,
is positive con
s
tant,
P
is po
sitive definite matri
x
.
The fo
rmul
a
(6) into th
e f
o
rmul
a
(10
)
i
t
can
get
th
e follo
wing
close
d
-lo
op
d
y
namic
equatio
ns of fuzzy cont
rol system
:
e
K
θ
x
b
T
D
n
u
u
e
)]
|
(
[
*
)
(
(16)
Let
:
1
1
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
k
k
k
n
n
Λ
b
0
0
0
b
(
1
7
)
The clo
s
e
d
lo
op syste
m
dynamic e
quati
on (16
)
can b
e
written a
s
a
vector of the form:
)]
|
(
[
*
θ
x
b
Λ
e
e
D
u
u
(18)
Define
d the o
p
timal param
eter:
|
)
|
(
|
sup
min
arg
*
*
1
u
u
D
R
x
R
N
n
i
i
m
θ
x
θ
θ
(19)
Define
d the minimum ap
p
r
oximation e
r
ror:
*
)
|
(
u
u
D
θ
x
(20)
By the formula (18
)
:
*
*
)
|
(
)]
|
(
)
|
(
[
u
u
u
u
D
D
D
θ
x
θ
x
θ
x
b
Λ
e
e
(
2
1
)
By the formula (12
)
error e
quation
can b
e
rewritten a
s
(21):
]
)
(
)
[(
*
x
θ
θ
b
Λ
e
e
ξ
T
(22)
Analysis of stability of this cont
rol al
go
rithm is
a
s
fol
l
ows, the d
e
finition of Lya
punov fun
c
tio
n
[12]:
)
(
)
(
2
2
1
*
*
θ
θ
θ
θ
b
Pe
e
T
T
V
(
2
3
)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 2, August 2015 : 284 –
293
290
In formula,
is positive con
s
tant
,
P
is a p
o
sitive definite matrix and
conforms to
the
Lyapun
ov eq
uation:
Q
P
Λ
P
Λ
T
(24)
In formula,
Q
is positive defini
t
e matrix of an arbitrary
n
n
,The
Λ
formula (1
7) is given,
the real pa
rt of the given value is n
egati
v
e.
Selec
t
ed
Pe
e
T
V
2
1
1
,
)
(
)
(
2
*
*
2
θ
θ
θ
θ
b
T
V
令
]
)
(
)
[(
*
x
θ
θ
b
M
ξ
T
, formula
(22
)
be
come
s
:
M
Λ
e
e
PM
e
Qe
e
PM
e
Pe
M
Qe
e
PM
e
Pe
M
P
ΛΛ
P
Λ
M
e
Λ
P
e
Pe
M
Λ
e
e
P
e
Pe
e
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
V
2
1
2
1
2
1
2
1
2
1
2
1
)
(
2
1
)
(
2
1
)
(
2
1
2
1
2
1
1
e
(
2
5
)
That is
:
)
(
)
(
2
1
*
1
x
Pb
e
θ
θ
Pb
e
Qe
e
ξ
T
T
T
T
V
θ
θ
θ
T
V
)
(
1
*
2
(26)
The de
rivatives of
V
:
2
1
V
V
V
(27)
The ada
ptive law form
ula (15) into the fo
rmula (27
)
, then
:
b
e
Qe
e
n
T
T
p
V
2
1
(28)
Due to
0
Q
,
is t
he minim
u
m
approximatio
n erro
r,
throu
gh the d
e
si
g
n
of fuzzy sy
stem
s
enou
gh
rule
s ca
n ma
ke
i
s
suf
f
i
cie
n
t
l
y
small,
a
nd in
ac
co
rda
n
c
e
wit
h
Q
e
b
e
T
n
T
p
2
1
,s
o
that
0
V
. The structure of dire
ct
adaptive fu
zzy control sy
stem
is
sho
w
n in Figure 4.
Figure 4. Dire
ct adaptive
fu
zzy control sy
stem
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Re
sea
r
ch on
Roll Vari
able
Speed G
r
indi
ng Base
d on
Adaptive F
u
zzy
Cont
rol (Bi
Jun
x
i)
291
4. Sy
stem
Simulation
Controlled d
e
vice ori
g
in
al formula
2
1
2
1
1
)
(
1
e
s
w
N
s
w
n
D
a
V
V
C
a
A
P
a
V
V
K
F
,
Simplified as:
w
w
n
V
B
V
A
F
'
'
, thereinto
S
V
a
K
A
'
,
2
1
2
1
1
'
)
(
1
)
(
1
D
a
V
C
a
A
P
B
S
N
.
Position com
m
and is
)
sin(
t
.
Selected the f
o
llowin
g
six ki
nds of mem
b
ership fun
c
tio
n
;
)))
2
(
5
exp(
1
(
1
1
x
N
,
)
)
5
.
1
(
exp(
2
2
x
N
,
)
)
5
.
0
(
exp(
2
3
x
N
,
)
)
5
.
0
(
exp(
2
1
x
p
,
)
)
5
.
1
(
exp(
2
2
x
p
,
)))
2
(
5
exp(
1
(
1
3
x
p
.
The initial
sta
t
e of the
syst
em is [1,0], the in
itial valu
e
of ea
ch el
em
ent were
cho
s
en
0 in
θ
, using
co
ntrol law
(11
)
,
adaptive
sele
ction of fo
rm
ula (1
5). Sel
e
cted
50
0
0
50
Q
, k
1
=1,
k
2
=1
0, adapti
v
e param
eters sh
ould b
e
selecte
d
50
.
Acco
rdi
ng to the membe
r
ship functio
n
d
e
si
g
n
pro
g
ra
m, which
can
get the membershi
p
function i
s
sh
own in Fig
u
re
below.
Figure 5. The
membershi
p
function of rol
ling
s
p
ee
d
Figure 6. Con
t
rol input sig
n
a
l
Figure 7. The
position tra
c
king
Figure 8. Tra
cki
ng erro
r
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 2, August 2015 : 284 –
293
292
Flow chart of fuzzy direct a
daptive cont
rol ar
e as follo
ws, in the pre
m
ise of con
s
t
r
uctin
g
adaptive sy
st
em, initializati
on pa
ramete
rs, the ada
pt
ive co
ntrol
syst
em ope
ra
tion
can o
u
tput the
optimal
spe
e
d
of roller
w
V
, u
nder the
co
n
t
rol of mi
cro
feed fo
rmul
a
w
f
V
V
r
f
2
output the
optimal fee
d
i
ng spee
d
f
V
, so that controlling m
o
tor
and
wo
rki
n
g
machine, a
nd feed
ba
ck
grindi
ng
whe
e
l cutting fo
rce and
roll
rou
ghne
ss after
pro
c
e
ssi
ng, throu
gh the
a
daptive la
w ,it can
real tim
e
cont
rol of
the o
p
timal outp
u
t, in
ord
e
r
to a
c
hi
eve the
purp
o
se
of a
dapti
v
e tran
smi
ssi
on,
to improve th
e roll gri
nding
rough
ne
ss.
Ho
wever,
in
pra
c
tice
the
r
e
are ma
ny un
certai
n fa
ctors, for exam
ple, the g
r
in
din
g
whee
l
of in homo
g
e
neou
s mate
ri
als, roll
with
different
mat
e
rial
s, gri
ndin
g
machine
a
nd environm
e
n
tal
uncertainty and so on. B
o
th of them
will affe
ct t
he grinding
with different
degree, more
importa
ntly is for grindi
ng
rolle
r su
rface
rough
ne
ss
reached the i
deal ro
ugh
ne
ss i
s
difficult to
control, so th
e above
de
scribed
there a
r
e un
ce
rtai
ntie
s in th
e mo
d
e
l. So this
pa
per
ado
pted
an
adaptive con
t
rol system t
o
solve the
s
e pro
b
lem
s
. The varia
b
le
micro fee
d
optimizatio
n of
adaptive cont
rol sy
stem is sho
w
n in Fi
gure
9, in
the pro
c
e
s
s of grindi
ng vibra
t
ion tran
smitted
dire
ctly to th
e g
r
indin
g
fo
rce
n
F
, fr
o
m
F
i
gu
r
e
9
,
th
e s
y
s
t
e
m
me
as
ure
d
n
F
as a
mo
d
e
l of the
kno
w
n n
u
me
rical i
nput di
rect a
daptive
contro
l opti
m
ization m
o
del, optimiza
t
ion model
o
f
contin
uou
sly
according
to
the inp
u
t value to
cal
c
ula
t
e the o
p
timal work pi
ece spee
d
w
V
and
optimal verti
c
al feed
rate
f
V
, and with determi
ne th
e spe
ed of grindi
ng whe
e
l
s
V
into the
optimizatio
n model of surf
ace
roug
hne
ss, throu
gh th
e optimizatio
n model of surface ro
ugh
ness
to cal
c
ulate t
he final optim
al wo
rk
piece
spe
e
d
w
V
and th
e optimal vert
ical feed
rate
f
V
, and sent
to the control unit, ensu
r
e the optimum
speed
d
r
ive sp
indle an
d the feed com
pon
ents work.
Figure 9. Flow ch
art of ad
aptive control
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Re
sea
r
ch on
Roll Vari
able
Speed G
r
indi
ng Base
d on
Adaptive F
u
zzy
Cont
rol (Bi
Jun
x
i)
293
5. Conclu
sion
This
pap
er i
s
su
ppo
rted
b
y
Natural Sci
ence Fo
und
a
t
ion of Inne
r
Mongoli
a
Aut
onomo
u
s
Regi
on of china (ID: 2
0
12MS07
31).
This p
ape
r analyzed the
con
c
ept, the mechani
sm of
variable
sp
ee
d grin
ding, th
e relatio
n
ship
betwee
n
variable spee
d
grindi
ng a
nd
tangential fo
rce
and the am
o
unt of feed, also de
scrib
e
d
the dire
ct ad
aptive fuzzy
contro
l pri
n
ci
ple, advantag
es
and a
pplication ,and
on th
is ba
si
s a
c
cording to t
he te
chn
o
logi
cal
condition
s of roller m
e
chani
sm
prin
ciple, p
r
o
posed the st
rategy of the variabl
e mi
cro feed optimi
z
ation
of self adaptive co
n
t
rol
based on the
direct a
dapti
v
e fuzzy co
ntrol, throu
gh the simul
a
tion
can effective
l
y improve the
roll gri
nding e
fficiency imp
r
oved the roll
grindi
ng roug
hne
ss.
Referen
ces
[1]
BI Junxi. CNC
roll grin
di
ng
machi
ne inte
lli
gent c
ontro
l s
y
stem research
based o
n
kno
w
l
e
d
ge. 2
0
0
7
;
06.
[2]
YUAN Julo
ng.
Research on
the devel
opm
e
n
t of sci
ence and the fiel
d of ultra precisio
n machin
ing.
Chin
ese Jo
urn
a
l of Mecha
n
ic
al Eng
i
ne
eri
ng.
2010; 4
6
(15):
161-
177.
[3] Hu
Zhanqi
,
Xie M
i
ng
li
,
Op
enGL B
a
se
d
Real
T
i
me&Inl
ine
Simu
lati
on
of C
N
C
Ca
ms Grind
i
ng
.
T
E
LKOMNIKA Indon
esi
an Jou
r
nal of Electric
al Eng
i
ne
eri
n
g
.
201
3.
[4]
K W
egener, H
W
Hoffmeister, B Karpusche
w
s
k
i
. Con
d
itio
nin
g
and m
oni
toring of gri
n
d
i
ng
w
h
ee
ls
.
Manufactur
i
ng T
e
chno
logy.
2
011; 60: 7
57-7
77.
[5]
F
e
i Guo, D
i
n
g
fei Z
h
ang,
Xus
hen
g Ya
ng. Inf
l
ue
nce
of
rol
lin
g sp
eed
o
n
mi
crostructure
an
d mec
han
ic
a
l
prop
erties of A
Z
31 Mg all
o
y r
o
lle
d b
y
l
a
rge s
t
rain hot rol
l
i
n
g
.
Materials Sci
ence & En
gin
e
e
rin
g
A.
2014;
607.
[6]
Adv Mater
Res
.
Advanc
ed I
n
formatio
n
a
nd
Comp
uter T
e
chno
log
y
in
En
g
i
ne
erin
g a
n
d
M
anufactur
i
n
g
.
Enviro
n
m
enta
l
Engi
neer
in
g. Advanc
ed Mater
i
als Res
earc
h
. 201
3.
[7]
DUAN Bin
h
u
a
. Desig
n
of Ultra precisi
on C
NC roll gr
ind
i
n
g
machi
ne.
Precisio
n man
u
facturin
g a
n
d
auto
m
ati
on.
20
11; 03.
[8]
F
an Jianj
un, Qiu Guanz
hou
1, Jiang
T
ao. Mecha
n
is
m of hi
gh pressur
e
ro
ll grin
din
g
on compressi
o
n
strength of o
x
i
d
ize
d
hematit
e pell
e
ts.
Journ
a
l
of Central So
u
t
h Univers
i
ty
.
2012; 09.
[9]
Adv Mater Res
.
Advanced Inf
o
rmatio
n
an
d Comp
uter
T
e
chno
log
y
in En
g
i
ne
erin
g an
d M
anufactur
i
n
g
.
Enviro
n
m
enta
l
Engi
neer
in
g.
Advanc
ed Mater
i
als Res
earc
h
. 201
3.
[10] Hu
Z
h
a
nqi
,
Li
Yukun. 4
A
xis
CNC Mac
h
i
ne
T
ool for Reli
ef
Grindi
ng Sp
her
e Gear H
o
b
.
TE
L
K
O
M
N
I
K
A
Indon
esi
an Jou
r
nal of Electric
al Eng
i
ne
eri
n
g
.
201
3.
[11]
Z
H
AO Yong.
Anal
ys
is o
n
R
oun
dn
ess Ultr
a
-error
an
d S
u
rface D
e
fects
of Ro
ller
Grin
din
g
Mac
h
in
e.
T
e
chnic
a
l en
gi
neer
ing.
2
011.
[12]
ZHU De-qi
ng,
YU W
e
i, ZHOU Xi
an-
lin, P
A
N Jian.
Stren
g
then
ing
pel
letiz
a
tion of ma
nga
nese or
e fine
s
contai
nin
g
hi
g
h
combi
n
e
d
w
a
ter b
y
hi
gh pr
essure ro
ll gr
i
ndi
ng a
nd o
p
ti
mized temp
era
t
ure elev
atio
n
sy
s
t
e
m
.
Jour
na
l of Central So
uth Univ
ersity
. 201
4; 09.
Evaluation Warning : The document was created with Spire.PDF for Python.