TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol.12, No.4, April 201
4, pp. 3093 ~ 3
0
9
9
DOI: http://dx.doi.org/10.11591/telkomni
ka.v12i4.4779
3093
Re
cei
v
ed Se
ptem
ber 27, 2013; Revi
se
d No
vem
ber
12, 2013; Accepted Decem
ber 6, 201
3
Study on Measuring and Forecasting of Fully
Mechan
ized Working Face Roof Pressure System
Yong Zhang
*
1,a
,
XueQiang Yang
2,b
,
ZengXin Wang
2,
c
1
Dept of Radi
o
a
ctivit
y
,
T
a
i Shan Med
i
cal C
o
l
l
eg
e, Chin
a
2
Xin C
ha Z
h
u
a
ng Min
g
Co.Ltd
, Shan Don
g
P
r
ovinc
e
, Chin
a
*Corres
p
o
ndi
n
g
author, Ch
an
gche
ng R
oad 6
19,T
a
ian, (+
86)053
86
222
17
4
Corresp
on
din
g
author, e-mai
l
: gczk
yanch
e
n
@
12
6.com
*a
, xq
w
a
ng@
16
3.com
b
,
zx
w
a
n
g
@
163.com
c
A
b
st
r
a
ct
Chin
a is
on
e o
f
the lar
gest co
al pr
od
ucer a
n
d
c
ons
u
m
er c
o
untries
in th
e
w
o
rld. How
e
ve
r, due t
o
the compl
e
xity of coal reso
urc
e
s, storage co
nditi
ons, ge
ol
o
g
ical
disaster-
p
rone co
al
min
e
s
, it is also a coal
mi
ne
acci
dents
multip
le c
o
u
n
try, coal
min
e
a
ccide
nts an
d d
eaths of
Ch
in
a accou
n
ted
t
o
a
bout 80%
total
of
the w
o
rld. I
n
t
he c
oal
mi
ne
a
ccide
nt occurr
ed, ro
of
acc
i
de
nt has
acc
ount
ed for
aro
u
n
d
40%, s
u
ch
as
th
e
roof co
lla
pse
d
, sli
ppe
d, d
e
for
m
ati
on,
obstru
c
tion
an
d so
o
n
. So t
he
moni
tor an
d
early
w
a
rning
of r
o
o
f
i
s
particu
larly
i
m
portant. State
of motio
n
is
cl
osely
rel
a
ted
to
mi
ne r
oof
pr
essure.
Roof
s
upp
ort press
u
r
e
or
resistanc
e
ca
n be measur
ed
b
y
the pressur
e
sensor.
T
h
e
da
ta sen
d
to
the
ring
Ethern
e
t u
nder
grou
nd
a
n
d
transmit to th
e
mo
nitori
ng
cen
t
er of gr
oun
d.
T
h
roug
h
i
n
for
m
ation
an
alysis
process
i
ng,
it c
oul
d pr
ovid
e re
al-
time data an
d early
w
a
rn
ing, alar
m
i
n
for
m
ati
on. App
l
i
ed ti
me seri
es the
o
ry
ana
lysis a
nd f
o
recasti
ng futu
re
pressur
e
chan
ges, can master the
roof mo
vement trends
and regu
lar
i
ty, guide safe pr
oducti
on. So t
h
e
decisi
ons h
a
ve
some
practica
l
signific
ance.
Ke
y
w
ords
:
mi
ne roof, press
u
re me
asur
ing,
dyna
mic m
onit
o
ring, safety fore
cast, tim
e
series
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
Chin
a is on
e
of the la
rge
s
t
coal
produ
ce
r and
co
nsum
er
cou
n
trie
s i
n
the
wo
rld [1
], [7-8].
With the re
ce
nt incre
a
se in
capa
city, increa
sing
the in
tensity of exploitation and t
r
an
sfer to de
ep
mining, coal
mine ro
of safety issue
s
increa
singly
appa
rent.Co
a
l-related
se
curity in
cide
nts
freque
ntly occur, an
d the se
curity
situa
t
ion remain
s
grim. As we
kno
w
, due to the compl
e
xity of
coal re
so
urce
s,
sto
r
ag
e co
ndition
s,
and
geolo
g
ical
di
saster-p
ron
e
coal
min
e
s, China coal
mine
accide
nts an
d death
s
a
c
counted to a
b
out 80% total of
the worl
d. In the acci
de
nt occurred, roof
accide
nt ha
s acco
unted f
o
r a
r
ou
nd
4
0
%,Such
as
the roof
col
l
apsed, sli
p
p
ed, defo
r
mati
on,
obstructio
n
a
nd so o
n
. So the roof monit
o
ring
a
nd ea
rl
y warnin
g is p
a
rticul
arly im
portant.
State of motion is
clo
s
ely
related to mi
ne ro
of pre
ssure. Roof su
pport p
r
e
s
sure ca
n be
measured by the strain g
a
u
ges
sen
s
o
r
lo
cating o
n
bra
c
ket. The motion of the roof
is non-li
nea
r.
2. Fully
Mec
h
anize
d
Wor
k
ing Face
Roof Pres
sure
Monitoring
Sy
stem
The above m
onitorin
g
syst
em has follo
wing main feat
ure
s
and te
ch
nical p
a
ra
met
e
rs:
(1)
System Monit
o
ring Poi
n
ts: 1-64 m
onitori
ng station (up
to 192 points);
(2)
Measurement
range: 0 - 6
0
MPA;
(3)
Comp
re
hen
si
v
e
erro
r <2.5
%
;
(4)
Display outpu
t: LCD, 20 ×
4 (LED b
a
ckli
ght)
(5)
Bus interfa
c
e:
RS-48
5
or C
A
N Bus;
(6)
Enterpri
se
standa
rd comm
unication protocol
sho
r
t fra
m
es;
(7)
Comm
uni
cati
on fault che
c
k (CRC);
(8)
Comm
uni
cati
on rate: 240
0
bps;
(9)
Supply voltage: 18V, intrinsically safe p
o
w
er
sup
p
ly;
(10
)
Ambient temperatu
r
e: 0 -
40; Relative
Humidity: 0 - 95% RH;
(11
)
Form of intrin
sic
safety explosio
n-p
r
oof:
Exib1.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 4, April 2014: 3093 – 3
099
3094
3. The Worki
ng Principle of Remo
te M
onitoring Sy
stem
There are four pa
rts in th
e monitori
ng
system: (1
)
Do
wn hol
e monitori
ng e
quipme
n
t
lay
e
r; (2) d
a
ta colle
ction a
nd monito
ring
lay
e
r; (3
) ap
plicatio
n se
rver layer; (4
) the rem
o
te cli
ent.
Figure 1 is th
e long-ra
nge
comm
uni
cati
ons
system t
o
monitor
sch
e
matic.
Figure 1. Lon
g-rang
e Com
m
unication
s Systems to M
onitor Schem
atic
3.1. Do
w
n
Hole Monitori
ng Equipment La
y
e
r
Colle
cting
d
a
ta from
b
r
a
c
ket sen
s
o
r
s of
workin
g f
a
ce
un
dergro
und, eve
r
y p
r
essu
re
comm
uni
cati
on su
b-statio
n has RS
-48
5
or CAN in
t
e
rface, it can
conn
ect with
undergro
und
ring
Ethernet thro
ugh NPo
r
t interface.
U
n
de
r
g
r
o
u
n
d
s
e
c
t
io
n us
ed
tw
o iso
l
a
t
ed
R
S
-48
5
B
u
s.
Comm
uni
cation
ma
ste
r
station
con
n
e
c
ted to
the
su
rveyed a
r
ea
co
mmuni
cation
su
b-station.
It can
b
e
con
n
e
c
ted to
16
measuri
ng
st
ations. Su
rve
y
ed area
co
mmuni
cation
su
b-station take on different
m
onitori
ng
function
s,
g
e
nerally a com
m
unica
tion sub-station re
spo
n
si
ble
fo
r monitori
ng a mining
fa
ce
a
nd
road
way. Ea
ch
com
m
uni
cation
su
b-station c
an
conne
ct maxi
mum 6
4
m
onitorin
g
sites,
extensio
n or sen
s
or, whi
c
h can meet
the domes
ti
c larg
e-scale
mines an
d layout pre
s
sure
monitori
ng n
eed
s. Differe
nt types of
monito
ri
ng
si
tes u
s
ed
a
unified
codi
n
g
. Through
the
comm
uni
cati
on proto
c
ol id
entifier co
uld
dist
ing
u
ish different types o
f
paramete
r
s.
Comm
uni
cati
on maste
r
e
m
bedd
ed DE
311 commu
ni
cation inte
rfa
c
e (NPort int
e
rface).In
addition to
su
pport Ethe
rn
e
t
(TCP/IP pro
t
ocol) dat
a transmi
ssion, it
also
su
ppo
rt
telepho
ne lin
e
or a
sin
g
le m
ode o
p
tical fi
ber
data tra
n
s
missio
n. Un
derg
r
o
und
co
mmuni
cation
maste
r
statio
n via
a telepho
ne
line, single
-
mode fibe
r or Ether
n
e
t ring net
wo
rk could
con
n
e
ct with g
r
o
und
monitori
ng se
rver conn
ecti
on.
3.2. Data
Col
l
ection and
Monitoring L
a
y
e
r
The laye
r
ca
n re
ceive
dat
a that came
from
the und
erg
r
ou
nd ring
Ethernet rea
l
-time,
store
d
in the monitor serve
r
/databa
se
se
rver. Th
rough
information
analysi
s
processing, it cou
l
d
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Study on Me
asu
r
ing a
nd F
o
re
ca
sting of Fully Me
ch
an
ized
Workin
g Face
Roof…
(Yong Zh
ang
)
3095
provide
re
al-t
ime data a
n
d
early warnin
g, alar
m i
n
formation. The
databa
se
co
uld up
date a
nd
ma
in
te
n
a
n
c
e in
time
; it als
o
r
e
s
p
on
s
i
ble
fo
r
s
e
n
d
i
ng
r
e
a
l
-
t
ime da
ta
to
th
e we
b
s
e
r
v
er
. Fo
r
different h
a
rd
ware a
r
chite
c
ture
s
and
o
peratin
g
sy
st
ems on
a ne
twork
of
com
puters,
TCP/IP
network
prot
ocol has the
ability to com
m
unicate
each other, so the la
yer and the underground
field control layer co
mmun
i
cation
s u
s
ing
TCP/IP protocol.
3.3. The Ap
plication Ser
v
er
La
y
e
r
This laye
r
ca
n be u
s
e
d
to manag
e web
page
s, an
d
make th
ese
page
s throug
h a lo
cal
netwo
rk o
r
th
e Internet for
cu
stome
r
bro
w
ser. Re
al-ti
m
e data of web se
rver
get
information f
r
om
the data
coll
ection
an
d
monitori
ng l
a
yer mo
nitori
n
g
serve
r
a
n
d
availabl
e.
Web
serve
r
coul
d
acce
ss mo
nitoring
se
rver d
a
taba
se u
s
in
g ADO.NET t
e
ch
nolo
g
y.
3.4. The Remote Client
Accept
clien
t
browse
r re
que
sts, via the In
tern
et a
c
cess data
o
f
applicatio
n
servi
c
e
layer [1].
4. Time Series Ba
sic Alg
o
rithm
Time se
rie
s
has the abilit
y to express
nonlin
ear
cha
r
acte
ri
stics [2]. Applied tim
e
seri
es
theory analy
s
is and fo
reca
sting future p
r
essu
re
ch
a
n
ges, can ma
ster the roof m
o
vement tren
ds
and
reg
u
larit
y
, could
guid
e
safe p
r
od
u
c
tion. So th
e
deci
s
io
ns
ha
ve som
e
p
r
a
c
tical
si
gnificance
[3-4].Time se
ries m
odel
s h
a
ve four ba
si
c form
s:
(1)
AR model: au
toreg
r
e
ssive
model;
(2)
(2)MA m
odel:
moving-ave
r
age;
(3)
ARMA model
: Auto-reg
r
e
s
sive Moving
-Average
；
(4)
(4)A
RIMA mo
del: Autoreg
r
essive Integrated Moving
Average Mo
d
e
l
；
Next, let’s ta
ke A
R
IMA m
odel fo
r an
e
x
ample, to b
e
discu
s
sed.
Assuming
a
ran
dom
pro
c
e
ss
with
d of unit root
, after it passes th
rough d
times the differential can
be tran
sform
e
d
into a stationary autoreg
ressive movi
ng avera
ge
pro
c
e
ss. The
stocha
sti
c
process is
cal
l
ed
singl
e wh
ole autore
g
ressiv
e moving ave
r
age p
r
o
c
e
s
s [2].
Consider the
following model:
t
t
d
u
L
y
L
)
(
)
(
(1)
)
(
L
,
whi
c
h is
a stationary auto
r
eg
re
ssive o
p
e
rato
r. The ro
ot of
0
)
(
L
is greater
than 1.
)
(
L
, it represents
reversible movin
g
average operator. If we take:
t
d
t
y
x
(2)
Equation (1)
can b
e
expre
s
sed a
s
:
t
t
u
L
x
L
)
(
)
(
(3)
It means tha
t
t
y
after a
d
time differen
c
e, it could b
e
rep
r
e
s
ente
d
by a stabl
e,
rev
e
r
s
ibl
e
ARMA
proce
s
s
t
x
.
After
d
time differen
c
e, random
process
t
y
can b
e
tran
sform
e
d
into a smo
o
th,
rev
e
r
s
ibl
e
st
o
c
ha
st
ic
pr
oce
ss.
)
(
L
, is a p
-
o
r
der
autoregre
ssive
ope
rato
r,
)
(
L
, is the m
o
ving
averag
e ope
rator of q-o
r
d
e
r.
t
y
, is also
calle
d (p, d, q)-o
rde
r
si
n
g
le wh
ole au
toreg
r
e
ssive
moving ave
r
age mo
del,
denote
d
b
y
)
,
,
(
q
d
p
ARIMA
.
A
R
I
M
A
pro
c
e
ss i
s
also
call
ed
autore
g
ressiv
e integ
r
ated
moving
averag
e p
r
o
c
e
ss.
d
L
、
)
(
, call
the g
ene
rali
zed
autore
g
ressiv
e operator.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 4, April 2014: 3093 – 3
099
3096
Whe
n
p
0, d = 0, q
0,
)
,
,
(
q
d
p
ARIMA
become
s
the pro
c
e
s
s
)
,
(
q
p
ARIMA
;
Whe
n
d = 0, p = 0, q
0,
)
,
,
(
q
d
p
ARIMA
become
s
the pro
c
e
s
s
)
(
q
MA
;
Whe
n
d = q = 0,
)
,
,
(
q
d
p
ARIMA
become
s
the pro
c
e
s
s
)
(
p
AR
;
And whe
n
p = d = q = 0, t
he
)
,
,
(
q
d
p
ARIMA
pro
c
e
ssi
n
g
become
s
white noise pro
c
e
ss;
5. Roof Inte
grate
d
Sy
st
em D
y
namic Monitorin
g
Time Series Modeling
Method
s a
n
d
Procedur
es
The ba
si
c ide
a
of the ARI
M
A model i
s
to predi
ct the
cha
nging
of roof p
r
e
s
sure
as the
data seq
uen
ce
fo
rme
d
by
a ran
dom
seque
nce,
u
s
i
ng a
math
e
m
atical
mod
e
l
to d
e
scri
be
the
approximate
seq
uen
ce. T
h
is mo
del is
use
d
to app
roximate de
scription of this sequ
en
ce. O
n
ce
the model is i
dentified, the time
seri
es can use the past value
s
an
d the pre
s
ent
value to pre
d
ict
future value
s
. Next, let us illustrate
s the
time seri
es m
odelin
g and fore
ca
sting p
r
oce
dures. Thi
s
method con
s
i
s
ts of the followin
g
five steps:
(1) Data
Inspec
tion
For fully mecha
n
ized co
a
l
mining face
working resi
stan
ce of su
pport, it first shoul
d
inspect
real
-time data for data validation, testing ti
me
seri
es sampl
e
stabilit
y, normality, periodic,
zero-m
ean, make
the
ne
ce
ssary data pro
c
e
ssi
ng.
If
the
time
seri
es no
rmality or smo
o
th
i
s
not
enou
gh well, it need for data co
nve
r
sio
n
. Often
there a
r
e
two metho
d
s: a differen
c
e
transfo
rmatio
n (u
sing the
transfo
rm-Create Tim
e
Serie
s
) an
d
logarithmi
c
transfo
rmati
ons
pro
c
ee
d (u
sin
g
the Tran
sfo
r
m-Com
pute). Generally
it sho
u
ld be re
peating tra
n
sf
orm, com
p
a
r
e,
until the data sequence norma
lity, stability, to achieve a relatively optimal.
Figure
2.
Tim
e
Serie
s
Mod
e
ling Steps
(2) Pattern Rec
o
gnition
Usi
ng a
u
toco
rrel
a
tion a
nal
ysis a
nd p
a
rti
a
l co
rrelation
analysi
s
, di
scrimi
nate mo
del form
and o
r
de
r, by analyzin
g th
e time se
rie
s
sampl
e
,
co
m
parin
g them i
ndividually an
d cal
c
ulatin
g
AIC
Call SPSS,
dat
a
validation,
hi
stogram
,
correl
a
tion
char
t
Pattern
recogn
i
tio
n
,
au
to
correl
a
tio
n
,
p
a
rtial
co
rrelatio
n
di
a
g
ram
,
p
ara
m
eter estim
a
tion
Mo
d
e
ling
AR
MA, calcu
late AIC, SBC an
d
ot
he
r param
e
t
e
rs
M
odel
C
h
ecki
n
g
M
odel
o
p
ti
m
i
z
a
tio
n
,
rec
o
n
f
i
g
uri
n
g
p,
d
,
q
va
l
u
es
Pre
d
iction m
echanize
d
m
i
ning
face
resistance bra
c
k
et
future wor
k
Call
m
echanized m
i
ning
face
real-tim
e data
support resista
n
ce
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Study on Me
asu
r
ing a
nd F
o
re
ca
sting of Fully Me
ch
an
ized
Workin
g Face
Roof…
(Yong Zh
ang
)
3097
values
(or S
B
C value
)
,
whe
n
AIC is the mi
nimu
m value
, th
e mo
del i
s
determi
ned.
Then
determi
ne the
type form of the model, to determi
ne p, d, q of order.
(3) Pa
ramete
rs e
s
timation
Usi
ng maxim
u
m likelih
ood
estimation o
r
least sq
ua
re
s estim
a
tion
method to e
s
timate
φ
,
θ
para
m
eter
values
and si
gnifica
nce tests perfo
rmed.
(4) Mo
del test
ing
Test n
e
w m
o
del is
wh
ethe
r re
asona
ble
or not. If
the test is
not pa
ssed, the
n
adj
ust (p,
q)
values, re-est
imate pa
ram
e
ters
and te
st repeate
d
u
n
til get acce
ptance date.
Ho
wever, a
b
o
ve
three
pro
c
e
s
se
s of m
odel
identificatio
n
,
para
m
et
er
estimation, te
sting
co
rre
cti
on can i
n
flue
nce
each othe
r, sometime
s ne
ed cro
ss,
rep
eated exp
e
ri
ments, in o
r
d
e
r to ultimate
ly determin
e
the
model form.
(5) Mo
del p
r
e
d
iction
Predi
ction
ro
of moving
st
ate at
some
time in
th
e fut
u
re, b
a
sed
o
n
predi
ctive
model to
cal
c
ulate the
predi
cted val
ue [5-6].
6. Roof Time
Series Prediction Engine
ering Practic
e
Figure 3 is #
2 stent
s average workin
g resi
st
an
ce ra
w data ma
p, from fully mechani
zed
coal fa
ce in
a mine in Inn
e
r Mon
golia,
Chin
a. A tota
l of 44 of the data set, the data colle
cte
d
once every five minutes. For the co
nvenien
ce of
d
a
ta pro
c
e
ssi
n
g
, the paper
sele
ct each d
a
ta
colle
ction tim
e
interval of
1 hou
r, unit is M
PA. First,
the ra
w data
applied SPS
S14 software
for
autocorrelatio
n
and pa
rtial correl
ation an
alysis. Ge
t Fi
gure 4 a
nd Fi
gure 5. See
n
from the Figu
re
4 and 5, the sam
p
le
sequence data f
r
om the
correlation coeffi
cient oscillat
e
s around
a fixed
hori
z
ontal,
a
nd at
pe
riod
ic g
r
a
dually
decrea
s
e
s
,
so it
can
kn
ow th
at the
time
se
rie
s
is
sub
s
tantially
stationa
ry. So it can be use
ti
me serie
s
to predi
ct the future data.
Figure 3. Average
Wo
rki
n
g
Resi
stan
ce
Ra
w
Data Figure
Figure 4. Autoco
rrelation
Figure 5. Partial Correl
ation Diag
ram
Figure 6. Foreca
st Re
sult
s
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 4, April 2014: 3093 – 3
099
3098
Figure 7. Foreca
st Re
sult
s
Figure 8. Foreca
st Re
sult
s
From SPSS table can be found in the autocorr
elation coefficients
and partial correl
a
tion
coeffici
ents.
Acco
rdi
ng to Coeffici
ent de
termine
d
the model pa
ram
e
ter value
s
. Here
，
take p
= 8,
q = 1, usin
g ARIMA (8,0,1
) model. Figu
re 6, 7, 8 sho
w
s the p
r
edi
cted results.
Figure 6 fore
ca
st the date from 45 ho
urs to 51 hou
rs,
Predictin
g da
ta and actu
al values
match
,
good
ne
ss of fit is better.
Figure 7 pre
d
i
cts value afte
r 55 hou
r, go
odne
ss of fit i
s
better
still good.
Figure 8 is foreca
st value a
fter 59 hou
r,
the pre
c
i
s
ion
of predi
ction redu
ced.
Seen to be o
v
er time, it m
u
st allow the
syst
em to re
-lea
rn, its accuracy will i
m
prove.
Table 1 give
s the compa
r
i
s
on between p
r
edi
cting valu
e and the a
c
tual value.
Table 1. Fo
re
ca
st Re
sults
No.
Actual values
predicting
value |difference
|
error
percenta
g
e
45 72.81
73.81
2.43 3.3375
46 70.47
70.38
1.77
2.5117
47 69.24
68.7
0.76
1.098
48 71.45
70
4.69
6.564
49 68.32
66.76
2.31
3.3811
50 65.31
66.01
4.79
7.334
51 72.47
70.1
0.86
1.1867
52 73.84
71.61
0.27
0.366
53 76.01
74.11
1.01
1.329
54 76.19
77.02
1.41
1.851
55 81.2
77.6
3.48
4.2857
56 82.4
77.72
3.74
4.5388
57 84.7
78.66
6.48
7.6505
58 85.6
78.22
8.76
10.2336
59 79.3
76.84
2.46
3.1021
7. Conclusio
n
The ap
plication of integ
r
at
ed sy
stem of
C/S+B/S of the net
work t
opolo
g
y enha
nce th
e
spe
ed comu
nicatio
n
of mine
net
wo
rks,a
s
we
ll
as imp
r
ove
the entire
cont
rol net
work
comm
uni
cati
ons reli
ability. The
data
se
nd to th
e ri
ng
Ethern
e
t un
derg
r
o
und
an
d tran
smit to
the
monitori
ng ce
nter of grou
n
d
. Throu
gh in
formation
an
alysis p
r
o
c
e
s
sing, it could
provide real
-time
data and e
a
rly warnin
g, a
l
arm info
rmat
ion.Using
tim
e
se
rie
s
method
s, we ca
n better pred
ict
mine
coal
ro
of re
sista
n
ce, maste
r
the
roof movi
ng
a
nd chan
ging
law. To
stren
g
thenin
g
of roof
sup
port and maintena
nce,
red
u
ce se
cu
rity
incid
ents,
guidin
g
for
safe
pro
d
u
c
tion of
coal
mine
enterp
r
i
s
e, ha
s impo
rtant practical sig
n
ifican
ce.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Study on Me
asu
r
ing a
nd F
o
re
ca
sting of Fully Me
ch
an
ized
Workin
g Face
Roof…
(Yong Zh
ang
)
3099
Ackn
o
w
l
e
dg
ements
This
work is
suppo
rted by the Natu
ral Sc
ience Foun
da
tion of Shand
ong Provin
ce,
China
(Nu
m
be
r: ZR2011EL
019
).
Referen
ces
[1]
Z
hang Y
ong.
Stud
y
on i
n
teg
r
ated tech
nol
o
g
y
bas
ed
o
n
r
oof d
y
n
a
mic
monitori
ng. Qi
ng Da
o: Sha
n
Don
g
Univ
ersit
y
of scie
n
ce a
n
d
techno
lo
g
y
. 2
009.
[2]
Chatfie
l
d C. T
h
e ana
l
y
sis of time series. Ne
w
York: Ch
ap
man an
d Hal
l
. 199
6: 91-1
10.
[3]
JIA Rui-sh
eng,
SUN H
ong-m
e
i, YAN
Xi
an
g-
hon
g. Mod
e
l fo
r safet
y
eval
ua
tion of c
oal m
i
ne ro
of bas
e
d
on evi
d
e
n
ce fu
sion the
o
r
y
.
Jo
urna
l of Chin
a
Coal S
o
ciety
. 2
010; 35(
9): 149
6-15
00.
[4]
Li L
i
, Ch
eng
Ji
ulo
ng. F
l
o
o
r
w
a
ter irru
ption
p
r
edicti
on
base
d
on
inform
ati
on fusi
on.
Jo
u
r
nal of
Ch
in
a
Coal S
o
ciety
. 2
006; 31(
5): 623
-626.
[5]
Z
hou H
ao, L
i
Shao
ho
ng. C
o
mbin
ation
of s
upp
ort
vector
machi
ne
and
e
v
ide
n
ce th
eor
y in i
n
formati
on
fusion.
Ch
in
ese Journ
a
l of Se
nsors an
d Actu
ators
. 2008; 2
1
(
9): 1566-
15
70
.
[6]
Han J, Kamber M. Data Mining.
C
once
p
t
s
an
d T
e
ch
ni
ques
. S
an F
r
ansisc
o
: Morg
an K
aufin
an
n
Publ
ishers. 20
01: 30-4
2
.
[7]
Zhang
Yong, YAN X
i
ang-
hong, Song
y
ang.
T
he co
mp
ute
r
bracket
pres
sure
mo
nitor s
ystem
of full
y
m
e
ch
an
i
z
ed
co
a
l
fa
ce
i
n
co
al m
i
ne
b
a
s
ed
on
C
A
N
Bu
s
. Procee
din
g
s of the SNPD 20
0
7
. QingDa
o
.
200
7: 317-
322.
[8]
Z
hangY
on
g, Yan
Xi
ang
Ho
ng
, Z
hu Hon
g
M
e
i, Son
g
Ya
ng.
Ethernet-b
ase
d
Co
mp
uter M
onitor
i
ng th
e
Roof Absciss
io
n Layer W
i
th E
x
perts F
o
recas
t
ing Syste
m
.
T
he 5th Intern
ati
ona
l Co
nferen
ce on F
u
zz
y
S
y
stems a
nd K
n
o
w
l
e
d
ge Disc
o
ver
y
. Ji
Na
n. 2008: 62
2-6
26.
Evaluation Warning : The document was created with Spire.PDF for Python.