TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 12, No. 10, Octobe
r 20
14, pp. 7274
~ 727
9
DOI: 10.115
9
1
/telkomni
ka.
v
12i8.583
5
7274
Re
cei
v
ed Fe
brua
ry 21, 20
14; Re
vised Ju
ly 3, 201
4; Accepted
Jul
y
28, 2014
Automated Data Processing for Monitoring Based on
Median Algorithm
Yan Du*
1
, Mo
w
e
n Xie
1
, Xiaoli Yang
2
, Qiuqiang
Liu
3
1
School of Civ
il
and Envir
onm
ental En
gi
neer
i
ng, Un
iv
ersit
y
of Science & T
e
chn
o
lo
g
y
Bei
j
i
ng
Beiji
ng 1
0
0
083
, China, Ph./F
ax: +
86-1
0
-62
3
3
409
8
2
Institute of Ge
oSpati
a
l Inform
ation fo
r GeoH
azard Ap
plic
ati
on, Univ
ersit
y
of Science & T
e
chn
o
lo
g
y
Bei
j
i
ng
Beiji
ng 1
0
0
083
, China, Ph./F
ax: +
86-1
0
-62
3
3
326
8
3
Chin
a Institute of Geo-Enviro
nm
ent Mon
i
tori
ng, Beij
ing
100
081, Ch
in
a, Ph./F
ax: +
86-10-6
219
28
56
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: mutulei@
16
3
.
com
A
b
st
r
a
ct
As introd
uctio
n
of the a
u
to
ma
tion e
qui
p
m
e
n
t, har
dw
are aut
omatio
n
lev
e
l on
res
e
rvoir ha
s
greatl
y
improve
d
. F
o
r
the restricti
on
of t
he s
o
ftw
are perfor
m
ance
and
tech
nica
l
perso
nne
l, d
a
ta re
orga
ni
z
a
t
i
o
n
f
a
r
faile
d to me
et the req
u
ire
m
ent
s of har
dw
are. Base on th
e re
search o
n
do
me
stic and fore
ig
n techni
qu
e, it is
concluded that a set
of
data t
r
ansfor
m
ation
m
e
thods
s
u
itable for
the
aut
omatio
n system
of the s
m
all and
me
di
um-s
i
z
e
d
da
ms. T
h
roug
h the me
di
an
deno
isin
g
pr
ocessi
ng an
d eig
enva
l
u
e
aut
omatic statistica
l
techni
qu
es, a
larg
e a
m
ou
nt of dat
a ca
n
be scre
e
n
ed
and fi
ltere
d
. T
he
met
hod
ca
n solv
e pr
acti
ca
l
eng
ine
e
ri
ng pr
obl
e
m
s and
meet the ne
ed of
auto
m
atio
n eq
uip
m
ent.
Ke
y
w
ords
: dat
a-proc
essin
g
, me
di
an al
gorit
hm, d
eno
isin
g
process
i
ng, saf
e
ty mo
nitori
ng
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
With the intro
ductio
n
of aut
omation e
qui
pm
ent, auto
m
atic data
acquisitio
n
equi
pment in
reservoi
r ma
nagem
ent ha
s a
c
hieve
d
a
rapid
develo
p
ment. Tho
u
gh the a
u
to
matic a
c
qui
si
tion
device
s
still have
some p
r
oble
m
s, su
ch
as
reli
abilit
y and
stabilit
y, it can n
o
t be de
nied
th
ese
equipm
ents
a
nd inst
rume
nts have g
r
e
a
tly redu
ced t
h
e
worklo
ad an
d wo
rk i
n
ten
s
ity of monitoring
s
t
aff. Over y
ears
c
o
ns
iderable progress
on
hardware makes the automatic
data proc
essin
g
become a m
a
jor challe
ng
e. Compa
r
ed
with the tr
aditional data co
mpilation, at pre
s
ent the
main
probl
em
s of automated dat
a compil
ation
to be solved
are:
1) Th
e large
amount of
data. Du
e to autom
ation
equipm
ent is very conve
n
ient in
acq
u
isitio
n a
nd the ne
ed i
n
engin
e
e
r
ing
,
each p
o
int i
s
mea
s
u
r
ed
more th
an a
dozen time
s
daily
and
coll
ecte
d
every
2-4 h
ours,
whi
c
h i
s
fa
r g
r
e
a
ter than
the t
r
a
d
itional
data
re
org
ani
zati
on
acq
u
isitio
n d
ensity, re
sulti
ng in
a g
e
o
m
etric in
cre
a
s
e i
n
the
am
ount of
data.
Moreover,
static
deform
a
tion
data is ri
ch,
but informat
ion is poo
r [1]. It is an urge
nt probl
em that how to
system
aticall
y
identify mo
nitoring d
a
ta analysi
s
an
d analyze dam
geote
c
hni
cal
con
d
ition
s
saf
e
ty.
2) System error. Reliabilit
y of the auto
m
ati
on of data is questioned
by many. In recent
years the
rel
i
ability and stability of m
onitorin
g
inst
rume
nt has
been g
r
eatly
improved, the
sampli
ng a
ccura
cy ha
s rea
c
he
d mo
re th
an 80%, an
d
even that of some eq
uipm
e
n
t has
re
ache
d
90%, howev
er, duri
ng sampling, the
r
e are still
extreme value cau
s
ed
by equipm
ent,
environ
menta
l
and
othe
r
reason
s [2].
No m
a
tter i
n
the d
a
ta a
n
a
lysis warni
n
g or in l
a
ter
data
reo
r
ga
nization [3], they
will dire
ctly lead to
a de
cre
a
se in rel
i
ability and credibility of data
analysi
s
[4, 5].
Based
on
th
e ab
ove con
s
ide
r
ation,
accordin
g
to ex
perie
nce of
a
u
tomation i
n
tegratio
n
and mathem
atical statisti
cs method
s
[6-9]
, this paper focu
se
s on an
alyzing reorg
aniztio
n meth
ods
of dam
s a
n
d
finding
a
set
of scientific
data inte
g
r
ati
on
an
alysi
s
method whi
c
h
verified by th
e
engin
eeri
ng e
x
amples.
2. Curren
t
Situa
t
ion and
Metho
d
s
By investigating and
comp
aring the d
o
m
esti
c a
nd f
o
reig
n data
pro
c
e
ssi
ng, a
nalyzin
g
dome
s
tic d
e
velopin
g
state
and future
developm
ent
mode, this
pape
r is to fi
nd a si
mple
and
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Autom
a
ted Data Processin
g
for Monitori
ng Base
d on
Median Alg
o
ri
thm
(Yan Du)
7275
effective dat
a p
r
o
c
e
ssin
g
metho
d
of
autom
atic
reco
gnition
suitable fo
r
e
ngine
erin
g a
nd
comp
uter a
p
p
lication, meeti
ng the pr
a
c
tical need
s of the reservoi
r da
ms.
2.1. Data
Re
organization
Situation
Acco
rdi
ng to
their o
w
n
nati
onal
con
d
itio
ns, di
fferent
countrie
s
va
ry
in deg
re
e in t
he u
s
e
of automation
monitorin
g
system, but mainly us
e the
integrate
d
m
anag
em
ent that have a great
influen
ce in
the
worl
d. Th
e dam
monit
o
ring
sy
stem
invented
by the
F
r
en
ch
power co
mp
any
monitored m
o
re th
an 1
5
0
dam
s an
d a
nalyze
d
with
a sim
p
le
st
at
ist
i
cal
mod
e
l
of
t
he sy
st
e
m
in
orde
r to find out the stru
ct
ural ab
no
rma
lities.
The MIDAS system i
n
Italy has be
en su
cce
ssful
ly
use
d
for ne
a
r
ly thirty years, an
d i
s
kno
w
n fo
r the
u
s
e
of the hy
brid
model
a
nd d
e
termi
n
istic
model fo
r onl
ine monito
rin
g
. Cent
ralize
d
mana
gem
e
n
t by profe
s
si
onal
s is
a co
mmon featu
r
e in
these
sy
stem
s
whi
c
h i
s
a
very go
od
referen
c
e
for our co
untry
t
o
develo
pe d
a
ta
a
c
qui
sitio
n
pro
c
e
ssi
ng m
ode.
In recent yea
r
s, while appl
icating MIDA
S
system, Italy also devel
oped the d
a
m
safety
asse
ssm
ent
deci
s
io
n su
p
port sy
stem
(DAMSAFE).
North Am
eri
c
a an
d othe
r count
rie
s
al
so
bega
n to
use net
work
a
nd the
integ
r
ation in
auto
m
ated d
a
ta
pro
c
e
ssi
ng
manag
eme
n
t, like
CANA
RY. in
US. Th
rou
gh t
he Inte
rnet, CANA
RY mana
ge t
he d
a
m p
o
wer statio
n
data
reo
r
ga
nization and dam
safety monitoring system,
whi
c
h re
alize
d
the win-
win
situation of the
manag
eme
n
t, supe
rvisio
n and ho
ster a
n
d
achi
eved g
ood soci
al be
nefits.
In China, th
e deci
s
io
n
sup
port
syst
em and the
expert sy
stem req
u
ire compl
e
x
kno
w
le
dge
e
ngine
erin
g, a
nd a lot
of m
anpo
we
r an
d
material re
source
s.
So
in
the overall p
o
int,
this
study i
s
still in th
e i
n
itial sta
g
. Th
e
dam
s safety automation
monitori
ng
l
a
rgely depe
nd
s
o
n
the spe
c
ific e
x
perien
c
e a
n
d
techni
cal le
vel of the operators.
In early 1
993
, Hohai
Univ
ersity a
nd El
ectri
c
Po
we
r
Burea
u
in F
u
jian Provin
ce
jointly
develop
ed t
he "Fujia
n
Province ex
pert d
e
ci
sio
n
sy
stem o
n
hydropo
wer d
a
m
saf
e
ty
manag
eme
n
t" which ma
de
great pro
g
re
ss in re
mo
te
monitori
ng, real-time an
al
ysis and n
e
twork
desi
gn, and then in 199
4 Hoh
a
i Unive
r
sity dev
elope
d the "Longy
angxia dam
safety asse
ssment
expert sy
ste
m
" and in 2
002 Nanjin
g
Water
Co
n
s
erva
ncy
Hydrop
ower S
c
ience Re
se
a
r
ch
Institute deve
l
oped the "da
m
safety monitoring
d
a
ta analysi
s
and
evaluation sy
stem".The two
have
b
een
su
ccessfully applie
d
in p
r
acti
ca
l
engi
neeri
ng. In
2007, th
e Isafety autom
atic
monitori
ng a
nd man
age
m
ent platform
usin
g the
Int
e
rnet a
nd p
r
ofession
ally managi
ng
with the
cloud platform fills the domestic
blank in this field.
With the d
e
velopme
n
t of the data m
oni
tori
ng a
nd th
e comi
ng of
big-d
a
ta ,increasi
ng
spe
c
iali
zation
, integration
and net
work
will com
e
to data reo
r
g
ani
zation a
nd profession
al da
ta
hosting
servi
c
es will
be a mainst
ream. So
efficient
and reliabl
e al
gorithm f
o
r
collaborative
work
will ensure th
e coo
r
din
a
tio
n
among sub
-
sy
stems a
n
d
the high efficiency in every single syste
m
,
and i
m
prove
the
overall
functio
n
of
au
tomati
c m
oni
toring
sy
ste
m
[9]. A ne
w d
e
si
gn
me
thod
need
s
cle
a
r thinki
ng
and
consi
ders a
wi
de
ran
ge
of d
e
sig
n
p
r
o
c
e
s
s. F
r
om th
e v
i
ew
of the
wh
ole
system, it
ca
n coo
r
dinate
the relation
sh
ip bet
wee
n
th
e whole
an
d
parts a
nd
am
ong
pa
rts. T
h
e
pro
c
e
ssi
ng
m
e
thod sho
u
ld be
mo
re scie
ntific
an
d
effe
ctive, so
that
data p
r
o
c
e
ssi
ng
which b
a
sed
on experi
e
n
c
e algorith
m
in
the long-te
rm beco
m
e
s
more the
o
reti
cal, scientific
and ratio
nal.
2.2. Denoisi
ng Metho
d
s
Bas
e
d on th
e Median Th
eor
y
The den
oisi
n
g
method
s of small and m
ediam
-s
i
z
e
d
dams d
a
ta are artificial de
noisi
ng
and auto
m
ati
c
den
oisi
ng.
Artificial den
oisin
g
, namel
y manually compa
r
e an
d
review th
e d
a
ta
error, is wi
del
y used in trad
itional monito
ring,
but few in the automa
t
ic denoi
sing.
For req
u
irin
g
a
lot of time, it
is
only suita
b
le for small
dams with
le
ss mea
s
u
r
ing
point
s an
d l
o
we
r m
onitori
ng
freque
ncy. Automatic de
n
o
isin
g based
on automat
ion pro
g
ram
m
ing lang
ua
ge increa
sin
g
ly
become
s
the
mainstream
method, such as
th
reshol
d denoi
sin
g
and the
wav
e
let denoi
sin
g
[3]
whi
c
h are very common at
pre
s
ent.
Ho
wever, the
r
e are still so
me pro
b
lem
s
in automatic
denoi
sing m
e
thod:
1) Th
e alg
o
rithm are
not
unified. it is diffi
cult to find certain
al
gorithm to
m
eet all
engin
eeri
n
g
s
’
requi
reme
nts.
2) The inte
rf
eren
ce of h
u
m
an facto
r
s.
Beca
u
s
e it is base
d
on th
e experie
nce
of th
e
operator to d
e
termin
e the algorith
m
and
thresh
old, it maybe not ob
jective.
3) It is
difficul
t
to apply. Be
cau
s
e
so
me
automatic de
noisi
ng al
gori
t
hm is ve
ry complex
and con
s
ume
more time [10], the pre-p
r
oce
s
si
ng ta
sk becom
es the
most criti
c
al part of
the anlysi
s
[11].
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 10, Octobe
r 2014: 727
4
– 7279
7276
The me
dian
denoi
sing
alg
o
rithm
(MDA
) can
effective
l
y solve the
a
bove p
r
oble
m
s. In the
automatic m
onitorin
g
, wh
en the data
acqui
siti
on
is not no
rm
al and eve
n
extreme val
ues
occurre
d
for system e
rro
rs, MDA ca
n filter them
an
d get real re
sults a
s
med
i
an is taken
as
gene
ral rep
r
ese
n
tative. Beca
use medi
an is in th
e
middle p
o
si
tion in se
qu
ence amo
n
g
the
variable
s
, it is not affe
cted by extre
m
e va
lue
s
(maxima an
d
minima). A
nd be
ca
use
the
measured m
edian value i
s
an a
c
tual value, comp
aring
with the averag
e value it has
more
pra
c
tical
and
statistical
si
gnifica
nce. So con
s
id
erin
g the gre
a
t differen
c
e in
measurin
g the
variable valu
e and even t
he hug
e differen
c
e
amon
g automati
c
data, MDA i
s
app
ro
priate
to
cal
c
ulate
for it
not
o
n
ly re
duci
ng the influence of
system error, b
u
t also gre
a
tly reduci
ng the
amount of historical d
a
ta, in orde
r to ma
ke
the su
bse
quent tedio
u
s data more ef
fectively.
The p
r
oje
c
t requires fa
st
analysi
s
of th
e safety of d
a
ms, an
d the
cal
c
ulation
method i
s
too com
p
lex
to apply. Co
mpared with
wavelet f
ilteri
ng [3, 12] an
d other
auto
m
atic de
noi
si
ng
method, MDA
has several
advantag
es a
s
followed:
1)
The sim
p
le al
gorithm. The
comp
uter d
e
noisi
ng algo
ri
thm is more
efficient, and
it is
widely u
s
ed e
s
pe
cially in th
e age of big d
a
ta.
2)
MDA d
o
e
s
n
o
t ch
ang
e th
e mea
s
u
r
e
d
value. The
a
c
tual val
ue
acq
u
ire
d
m
a
ke
s
sub
s
e
que
nt data analysi
s
more valu
abl
e.
3)
The u
n
ified
al
gorithm. All
o
f
these
term
s are
am
bigu
o
u
s
so
someth
ing mo
re
pre
c
ise
is nee
ded
be
fore attempti
ng to qua
ntify features
[7]
. The alg
o
rith
m need
s not
to set
certai
n threshold a
nd
parameter whi
c
h can
avoi
d
the inte
rfere
n
ce
of subje
c
tive
factors and ,t
herefo
r
e, re
al
iz
e the autom
atic den
oisi
ng
.
4)
Advance
d
in
denoi
sing. A
c
cording to th
e formul
a (1
),
denoi
sing
rat
e
ca
n be a
s
high
as
99.99%
(con
sid
e
rin
g
t
he eq
uipm
en
t corre
c
t rate
as
90% a
n
d
the n
u
mbe
r
of
sampl
e
as 9
)
.
()
[]
n
ni
ii
n
i
FC
P
P
(1)
Whe
r
e:
F
is
noise reduction rate,
P
i
s
th
e eq
uipme
n
t
corre
c
t rate,
n
is th
e n
u
mb
er of th
e
sampl
e
,
i
is the corre
c
t num
ber in the sa
mples.
Becau
s
e
the
data a
c
qui
siti
on is ea
sy a
n
d
conveni
ent, it is
feas
ible to reduc
e
the
number
of sampl
e
s f
o
r sample
accuracy [13]. In the ne
w p
e
riod
of data
integratio
n, the mo
st criti
c
al
factor in a
nal
yzed stati
s
tics is not the
small num
b
e
r
of sampl
e
s traditionally, bu
t the erro
rs in
the
large am
ount
of data. How
to eliminate abnormal
data
from abno
rm
al measure
m
ent or indu
ction
failure in auto
m
ation equi
p
m
ent is very critical to
the accuracy an
d
the result
s o
f
error a
nalysis.
MDA
ca
n sol
v
e it
.
Ho
wever, M
D
A eliminate
s
half of the
sampl
e
data
.
Superficially
these d
a
ta may not
used in the l
a
ter analysis,
but the data are
still us
eful actually.
Automated
system can offer m
o
re
data than a
c
tural requi
re
ment [14], so the num
be
r of the sam
p
les
pro
c
e
ssed still me
et the
analysi
s
re
qui
reme
nts.
2.3. Technic
a
l Scheme
Firstly, in ord
e
r to elimin
ate the abn
orm
a
l
data in a
u
tomatic d
a
ta
and
system e
rro
r, we
pro
c
e
ss
auto
m
atically data
by using M
D
A. Secondl
y, according to t
he statisti
c m
e
thod ba
se
d on
c
h
arac
teris
t
ic value, we get the s
o
lution to
pro
c
e
ss the automat
ic data. In th
is way we
can
provide ta
rget
ed pro
c
e
s
sin
g
method for
small an
d me
dium dam
s.
3. Applicatio
n Case
s
3.1. Automa
tic Data Inv
estigation
In rece
nt years the relia
bility and stability
of monitoring in
stru
ment has b
e
en gre
a
tly
improve
d
an
d the
sampli
ng a
c
curacy
has
re
ac
hed more
th
an 80
%,
but
there are still
extre
m
e
value ca
used
by equipmen
t, environmen
tal and other
rea
s
on
s. Figu
re 1 sh
ows th
e history line
of
the autom
atic data. Thi
s
i
s
the auto
m
ati
c
d
a
ta
hi
story curve
of a
reinfo
rced m
e
ter in
practi
cal
engin
eeri
ng.
Abnorm
a
l d
a
ta o
c
cur
be
ca
use
of the
system o
r
sen
s
or p
r
obl
ems. The
s
e p
e
a
ks
belon
g to the
invalid data
and the
erro
rs ma
ke th
e a
nalysi
s
confu
s
ed,
whi
c
h m
a
y lead to
wrong
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Autom
a
ted Data Processin
g
for Monitori
ng Base
d on
Median Alg
o
ri
thm
(Yan Du)
7277
con
c
lu
sio
n
. All of these
term
s a
r
e a
m
biguo
us
so
something
mo
re p
r
e
c
ise i
s
need
ed
bef
ore
attempting to quantify features. So it
is n
e
ce
ssary to p
r
ocess the da
ta.
Figure 1. The
History Lin
e
of the Automatic Data
with
out Data Pro
c
essing
3.2. The App
lication of M
D
A
After using
MDA, pea
ks obviously re
mov
ed, and
the denoi
sin
g
rate wa
s 1
00%. As
sho
w
n in fig. 3, the denoi
si
ng effect is o
b
vious.
The algo
rithm
is as follo
ws:
n=1;
while
(n
<=fix(l
ength(a)/m
))
for i=
1:m
b(i)=a(i
+(n-1)*m);
c(n
)
=medi
an
(b);
end
n=n
+
1;
end
In the
contex
t of Gig
Data
, efficien
cy is impotent. A
c
cording
to
so
me p
r
oje
c
ts,
as
we
tested, the
re
spo
n
se time
of MDA i
s
wit
h
in 6
0
se
con
d
s,
whi
c
h i
s
f
a
r le
ss th
an
other algo
rith
ms.
Re
sults a
r
e shown in Tabl
e 1.
Figure 2. The
History Lin
e
of the Automatic Data afte
r MDA
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 10, Octobe
r 2014: 727
4
– 7279
7278
Table 1. The
Perform
a
n
c
e
of MDA
Program
The numbe
r of
points
The numbe
r of
masured values
Denoising time /
s
Small dam
112 327,040
1.63
H
y
dr
opo
w
e
r
Station
1014
3,701,100
30.8
3.3. The App
lication of Automa
tion Al
gorithm Ba
s
e
d on Cha
r
a
c
teris
t
ic Value
Usi
ng
comp
uter auto
m
at
ically stati
s
tica
l fun
c
tion
s [15] can
quickly anal
yze the
monitori
ng
d
a
ta an
d g
e
t
the po
sition
who
s
e
amplit
ude m
e
a
s
u
r
e
s
la
rge
r
. Th
u
s
it
will g
r
ea
tl
y
redu
ce the
search time fo
r us, provide
prelimin
a
r
y re
fe
r
e
nc
e
an
d fo
c
u
s
on
the target for dat
a
analysi
s
. As
sho
w
n i
n
Tab
l
e 2, R2
05 m
easurin
g poi
n
t
of the cross se
ction vari
e
s
greatly, whi
c
h
in Octob
e
r o
c
curre
d
in the
cha
nge of 70
.63mm.
So it requi
re
s furth
e
r attention a
nd analy
s
is f
o
r
the engin
eeri
ng staff in the practi
cal an
a
l
ysis an
d mo
nitoring
wo
rk.
The com
pari
s
on
can p
r
ovi
de
new id
ea
s a
nd refe
ren
c
e
s
. We
can
al
so u
s
e the
compute
r
information proce
ssi
ng fun
c
tio
n
to
reali
z
e two
-
di
mensi
onal im
age di
splay.
Table 2. Strai
n
Gaug
e Ob
servation Results in Octo
be
r
Unit
：
mm
Measurin
g
p
o
in
t
name
The maxi
mu
m
va
l
u
e
The maxi
mu
m
occurre
nce ti
m
e
The mini
mu
m
va
l
u
e
The mini
mu
m
occurre
nce ti
m
e
A
m
pl
it
ud
e
R201
9.34
2013/10/19 0
8
:0
0
4.39
2013/09/20 0
0
:0
0
4.96
R202
21.72
2013/10/19 0
8
:0
0
14.78
2013/09/20 0
0
:0
0
6.94
R203
141.61
2013/10/19 0
4
:0
0
118.53
2013/09/22 2
2
:0
0
23.07
R204
264.99
2013/10/19 0
4
:0
0
243.36
2013/09/23 2
2
:0
0
21.63
R205
362.38
2013/10/19 0
6
:0
0
291.75
2013/10/15 0
5
:0
0
70.63
R210
-23.01
2013/09/20 2
2
:0
0
-10.10
2013/10/19 0
8
:0
0
-12.90
R211
121.24
2013/10/19 0
4
:0
0
104.52
2013/09/22 2
2
:0
0
16.72
R212
277.42
2013/10/19 0
4
:0
0
252.82
2013/09/23 2
2
:0
0
24.60
R213
169.49
2013/10/19 0
6
:0
0
151.83
2013/09/21 2
0
:0
0
17.66
R214
23.79
2013/10/19 0
8
:0
0
2.16
2013/09/23 2
2
:0
0
21.63
R215
92.77
2013/10/19 0
8
:0
0
80.12
2013/09/21 2
0
:0
0
12.65
R216
45.03
2013/10/19 0
8
:0
0
32.85
2013/09/20 2
2
:0
0
12.19
4. Conclusio
n
Hardware aut
omation
equi
pment in
Chi
na h
a
s be
co
me mo
re
mat
u
re, b
u
t the
method
of
pro
c
e
ssi
ng the automati
c
data develo
ped slo
w
ly
whi
c
h ca
nnot
meet the need
s of pra
c
tical
engin
eeri
ng
of dam. In term
s of soft
ware, su
ch
a
s
data p
r
o
c
e
ssi
ng an
d an
alysis, auto
m
atic
pro
c
e
ssi
ng te
chni
que h
a
s t
o
be furthe
r p
r
omote
d
.
Thro
ugh
com
parative an
al
ysis an
d eng
ineeri
ng
practice, the effective combi
n
ation of
median
alg
o
rithm and
ei
g
envalue
stati
s
tics
,
by pro
g
rammi
ng wi
th
co
mpute
r
langu
age, ca
n
reali
z
e initial
reo
r
ga
nization function of
automatic d
a
ta . Becau
s
e
of its simpli
city in operation
,
it
is suit for sm
all and middl
e dams a
u
to
matic data.
The M
D
A
cal
c
ulate
s
qui
ckl
y. In the a
ge
of big
data, th
is d
enoi
sin
g
method
is si
mple
an
d
gene
ral an
d can qui
ckly provide the req
u
ired e
ngin
e
e
r
ing data fo
r analysi
s
.
With the
dev
elopme
n
t of
monitori
ng
sy
stem, d
a
ta p
r
ocessin
g
will
be
com
e
mo
re
and
more th
eoretical, scie
ntific an
d ration
al.
Speciali
zation, integra
t
ion and
net
work, ma
nag
ed
servi
c
e
will be professio
nal. It will be a main
st
re
am that the
dams
acqui
re a
p
r
ofe
ssi
onal
hostin
g
se
rvice.
Referen
ces
[1]
CHING-YUN K
,
CHIN-HSIUN
G
L. Monitor
i
n
g
of l
o
n
g
-term
static def
ormation
data
of
F
e
i-T
s
ui arc
h
dam us
in
g arti
fi
cial ne
ural
ne
t
w
ork-
bas
ed a
ppro
a
ches[J].
Structural C
o
n
t
rol an
d H
e
a
l
th Mon
i
tori
ng
,
201
3(20): 2
82-
303.
[2]
Hu W
,
YANG
X G, Z
H
OU J W
,
et al. Gro
ss error deno
i
s
ing meth
od for slop
e monit
o
rin
g
data at
h
y
dro
p
o
w
e
r
station[J].
T
e
lko
m
n
i
ka Ind
o
n
e
si
an Jour
nal
of Electrical E
ngi
ne
, 2013, 11(
10): 554
5-55
52.
[3]
Su H Z
,
W
u
Z
R, W
en Z
P. Identificati
o
n mo
d
e
l for
D
A
M beh
avi
o
r
base
d
o
n
w
a
v
e
let n
e
t
w
ork[J]
.
Co
mp
uter-Aid
e
d
Civi
l
a
ndInfras
t
ructureEng
in
e
e
rin
g
, 200
7, 22
(6): 438-4
48.
[4]
KIBLER K M,
T
U
LLOS D D,
KONDOLF
G M. Learn
i
n
g
fr
om dam
remo
val mo
nitori
ng:
chal
le
nges
to
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Autom
a
ted Data Processin
g
for Monitori
ng Base
d on
Median Alg
o
ri
thm
(Yan Du)
7279
selecti
ng e
x
pe
rimental
des
ig
n an
d estab
l
ish
i
ng
si
gn
i
f
i
c
an
ce
o
f
ou
tcome
s
[J].
River researc
h
an
d
app
licati
ons
, 2
011, 27(
6): 967
-975.
[5]
BUCHHEIT
R
B, GARRETT
J H, MC
NEIL
S, et a
l
. Autom
a
ted
proc
edur
e
s
for im
provi
n
g
the
accur
a
c
y
of Sens
or-Bas
ed mo
nitor
i
ng
data[C]//
Appl
i
c
ations
of Ad
vance
d
T
e
ch
n
o
lo
gies
in T
r
a
n
sportati
on
.
Boston, 20
02: 402-
409.
[6]
JAYAWARDENA A W, GURUNG A B. Noi
s
e red
u
ctio
n
a
nd pr
edicti
on
of h
y
drom
eteo
rolo
gica
l ti
m
e
series
_ d
y
n
a
mi
cal s
y
st
ems ap
proac
h vs. stochastic ap
pro
a
c
h
[J].
Journal of
Hydrol
ogy
, 20
00, 22
8(3/4):
242-
264.
[7]
RANDOLPH T
W, YASUI Y. Multiscale pr
ocessing of mass spectrometry
data[J].
Biom
etric
, 20
06,
62(6): 58
9-5
9
7
.
[8]
CHRIST
ER A H, WANG W.
A simpl
e
con
d
i
t
ion mo
ni
tori
ng
mode
l for a
di
rect monitor
i
ng
process[J]
.
Europ
e
a
n
Jour
nal of Operati
o
nal R
e
searc
h
, 199
5, 82(3): 25
8-26
9.
[9]
Z
H
ANG D Q, HAN Y B, T
A
NG X
Y. No
nli
near
_No
n
g
aus
sian tim
e
ser
i
e
s
pred
ictio
n
b
a
s
ed o
n
RBF
-
HMM-GMM m
odel[J].
T
e
lkomnika Ind
o
n
e
sia
n
Journ
a
l of El
ectrical En
gin
e
, 2012, 10(
6): 1214-
122
6.
[10]
F
ang B, GUO X Q. Mod
i
fie
d
al
lan v
a
ri
an
ce
an
al
ysis
o
n
ran
dom
errors of MINS[J].
T
e
lkomni
k
a
Indon
esi
an Jou
r
nal of Electric
al Eng
i
ne
, 2
0
1
3
, 11(3): 12
27-
123
5.
[11]
W
U
C-h, YAN
G C-h, LO S
C, et al. A
u
to
mati
c dat
a mi
nin
g
for te
lem
e
tr
y
data
base
of comp
ute
r
s
y
stems[J].
Microel
ectronics r
e
lia
bi
lity
, 201
1, 51(2): 263-
26
9.
[12]
LIANG G L, XU W Y, T
A
N X
L, et al.
DE
NOISING PROCESSING OF S
A
FE
T
Y
MONITORING DAT
A
FOR HIGH R
O
CK SLOPE BASED ON WAVELET
TR
ANSFORM [J].
Chines
e Jou
r
nal of Ro
c
k
Mecha
n
ics an
d
Engin
eeri
n
g
, 2
008, 27(
9): 183
7-18
44.
[13]
YAN Du, XIE Mo-
w
en, MAN
Hu. Mecha
n
is
m of lake
area
variatio
ns and
w
a
ter l
e
vel c
h
ang
es in NA
M
CO lake[C]//
Advance
d
Materi
als Res
earch
．
Shenz
he
n, 2013: 16
85-
169
2.
[14]
ZHAO S H, X
I
ANG
T
Y, LAI
H G.
Anal
y
s
is of compil
atio
n optimiz
ation
of h
y
dro
l
og
ical
da
ta in upstre
a
m
and
d
o
w
n
stre
a
m
stations
of
T
h
ree Gor
ges
Project befor
e and
after impo
undm
ent[J].
Yangt
z
e
Riv
er
,
201
2, 43(2
3
): 18-22.
[15]
SUN E J, ZHA
N
G X K, LI Z
X
.
T
he Intern
et of thin
gs (IOT
) and
cl
oud
co
mputin
g (CC)
base
d
tai
lin
g
s
dam[J].
Safety Science
, 201
2, 50(4): 811-
81
5.
Evaluation Warning : The document was created with Spire.PDF for Python.