TELKOM
NIKA
, Vol. 11, No. 12, Decem
ber 20
13, pp.
7599
~76
0
4
e-ISSN: 2087
-278X
7599
Re
cei
v
ed
Jun
e
29, 2013; Revi
sed Aug
u
st
17, 2013; Accepted Sept
em
ber 3, 201
3
Based on Artificial Immune Algorithm of Robot Multi-
Sensor Signal Variation Characteristics of the
Detection Method
Hong
w
e
i Yan*, Huijuan
Li, Xin L
i, Qia
ng Gao
Col
l
e
ge of Me
chan
ical En
gi
n
eeri
ng & Auto
matizatio
n
, Nor
t
h Univers
i
t
y
of
Chin
a, Xu
e
y
u
an Ro
ad 3,
T
a
iyua
n, 030
0
51, Chi
n
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: ya
w
e
ig
eh@s
ohu.com
A
b
st
r
a
ct
W
i
th the continuo
us impro
v
em
ent of robot intel
lig
ent, cons
tantly e
x
pan
din
g
ran
ge o
f
app
licati
ons, a
s
w
e
ll as mu
lti
-
sensor i
n
for
m
ation fusi
on te
chno
logy, the t
r
aditi
ona
l sin
g
l
e
sensor si
gn
a
l
transmissio
n
prob
le
m has
beco
m
e multi-
sensor trans
mission
prob
le
ms or
mu
ltipl
e
source si
g
n
a
l
transmissio
n
p
r
obl
ems. T
h
is
brou
ght a l
a
rg
e a
m
o
unt
of s
i
gn
al var
i
atio
n
and s
i
gn
als
mu
ltipl
e
var
i
ati
o
n
prob
le
ms. T
h
e
tradition
al
det
ection
alg
o
rith
m h
a
s be
en u
nab
le to
me
et the req
u
ir
e
m
en
ts; therefore, this
pap
er puts for
w
ard a kind of
robot multis
en
sory sign
al var
i
atio
n test met
hod b
a
se
d on
artificial
immun
e
alg
o
rith
m. First, establis
h the
dyna
mic cha
n
g
e
s of the
si
gna
l
variab
ility of e
quati
ons to g
e
t
the cross po
in
t
of the d
i
stribu
tion of th
e si
gna
l vari
ab
ility
of varia
b
il
ity, then
upd
ate
sign
al
var
i
ati
on ch
aracteris
t
ic
datab
ase, in
the data
base
selectio
n sig
nal var
i
atio
n characte
r
i
stics. T
he meth
od
overco
mes t
h
e
drawbacks of traditional algor
i
thm
s
; t
he experim
ents show
that this algor
ithm
can av
oid
t
he defect signal
varia
b
il
ity of mutation, to i
m
pr
ove t
he acc
u
ra
cy of signal var
i
atio
n detecti
on
.
Ke
y
w
ords
:
c
h
aracteristic d
a
tabas
e, mu
ltipl
e
source
sig
nal t
e
st, artificial i
m
mu
ne, sens
or
Copy
right
©
2013 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Due to th
e a
pplication of
roboti
c
s is i
n
cre
a
si
ngly wi
de no
w, ou
r
requi
rem
ent to the
environ
ment
accomm
odat
e ability of
ro
bot is in
cr
ea
singly
high. I
n
the
sa
me ti
me, du
e to t
h
e
compl
e
xity of robot
syste
m
and in
stab
ility of
environment, the e
n
vironm
ent a
nalysi
s
provide
d
by tradition
al
singl
e
sen
s
o
r
be
come
limit
ed, an
d d
on’t
meet the
req
u
irem
ent to t
he info
rmatio
n
of
accu
ra
cy and
tim
e
line
s
s.
In re
cent years with
th
e ri
se
of info
rmation
fusi
o
n
technol
ogy
of
multi-sen
s
or,
the multidi
m
ensi
onal i
n
formati
on
proce
s
sing m
e
thod of this techn
o
logy
can
effectively de
al with th
e fu
zzy
point
of si
ngle
se
n
s
o
r
,
more
a
c
c
u
rat
e
ly
ob
serv
e
a
nd inte
rp
ret t
he
surro
undi
ng
environ
ment,
effectively reduce t
he phe
nomen
on of missed an
d misdia
gno
si
s [1-
4]. But it also b
r
ing
s
the
problem
of
sig
nal va
ria
t
ion in th
e
signal tran
smi
ssi
on,
so it i
s
necessa
ry to detect the variation sig
nal in
the pro
c
e
s
sing of robot si
gnal tran
smi
s
sion.
1)
The
dete
c
tio
n
of va
riation
sig
nal i
s
a
complex
detection p
r
oble
m
, like
the
prob
lem of fa
ult
diagn
osi
s
.
2)
Freq
uently-u
sed fault diag
nosi
s
meth
od
s ar
e ge
neral
ly based
on t
he technolo
g
y
of sen
s
or
detectio
n
, it is effective u
s
ually only in singl
e f
ault condition, it is
usel
ess for th
e pro
b
lem of
multiple faults
3)
Due to
the v
a
riation
an
d
even several
variat
ion
s
si
gnal
s do
n’t b
e
a
ccu
rately
detecte
d an
d
diagn
osed, th
e inadve
r
tent
operation
s
o
f
robot
a
r
e m
ade, even th
e paralysi
s
of
system i
s
cau
s
e
d
, the effective way becom
e usele
ss.
4)
Thro
ugh the
extraction
of sign
al variatio
n f
eature
s
, th
e com
pari
s
o
n
betwee
n
si
g
nal variatio
n
and va
riation
feature d
a
taba
se,
sign
a
l
variat
ion
d
e
tection
will
be
complet
ed, but
th
e
sub
s
tantial d
e
viation bet
wee
n
sig
nal
variat
ion fe
ature a
nd original feature
cau
s
ed by
several variat
ion of sign
al in sho
r
t time can red
u
ce the
detection a
c
curacy of sig
nal variation
5)
Thro
ugh
the
artificial
im
mune
dete
c
tion meth
od i
n
biolo
g
ical i
s
u
s
e
d
an
d
the dynami
c
equatio
n of signal variatio
n feature is
establi
s
h
ed, the cross poi
nt di
stributio
n
of feature
variation of si
gnal variatio
n
will be got, and the det
e
c
tion of sign
al variation
will b
e
got [5-6].
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN: 2
087-278X
TELKOM
NIKA
Vol. 11, No
. 12, Dece
mb
er 201
3: 759
9 – 7604
7600
2.
The De
tec
t
io
n Principle of Robo
t Sen
s
ing Signal Variation
The Dete
ctio
n Modle of Rob
o
t Sensi
ng Signal Variation
:
Wh
en rob
o
t multi-sen
s
o
r
signal vari
ation is detected,
firs
t the si
gnal variation feature
will
be extracted,
that
signal feature
and
sign
al v
a
riation
featu
r
e d
a
taba
se
will be
co
mp
ared, fin
a
lly the det
ection
will be
finish
ed.
Princi
ple diag
ram a
s
sh
own in Figure 1 [6].
Assu
me the
sampl
e
n
u
m
ber
of sig
nal
variation fe
a
t
ure d
a
taba
se is
(
n
), the sample
numbe
r of
si
gnal vari
ation
feature i
s
(
i
),
the se
rial n
u
m
ber
of sig
n
al variation f
eature i
s
(
V
),
the time i
n
terval of b
r
ain
o
peratio
n i
s
(
t
),
the b
r
ain
op
e
r
ation
num
be
r i
s
(
l
)
,
th
e cha
r
ac
te
r
i
s
t
ic
numbe
r of si
gnal variatio
n
in same time is(
m
). The weighted value
of signal vari
ation ca
n be
cal
c
ulate
d
by the formula (1):
i
t
V
l
n
m
2
2
2
(1)
Acco
rdi
ng to
the formul
a
above, the
variat
ion d
egre
e
of m
u
lti-se
nsor
si
gnal i
s
descri
bed, th
e variation
coefficient of
sign
al
variati
on feature is cal
c
ulated.
The mutatio
n
coeffici
ent of sign
al variatio
n featur
e is d
e
scrib
ed by the formul
a (2
):
2
1
n
t
i
(2)
To en
su
re th
e situatio
n of
sud
den
ch
an
ge of
multi-se
nso
r
vari
ation
feature, the f
eature
mutation coef
ficient is
add
ed to the feat
ure
weig
ht
value calculatio
n. The n
e
xt formul
a (3
)
ca
n
get accurate weig
ht value.
1
)
1
(
2
t
n
(3)
Acco
rdi
ng to
the formul
a a
bove, the vari
ati
on feature
weig
ht value
of actual
sig
n
a
l ca
n
be got. The si
gnal variatio
n
can be d
e
tected and di
scri
minated [7-10
]
.
Figure 1. Signal Mutation
Dete
ction Pri
n
cipl
e Diag
ra
m
The De
fec
t
s
Of The Me
thod Of
Ro
bot Signal Variation De
tec
t
ion:
Due to the
external envi
r
onm
ental complexity
and
the
robot
intelligence requi
rement
s continue
t
o
increa
se, so the u
s
e of mu
lti sen
s
or fu
si
on tec
hnolo
g
y
, it is possi
b
l
e in a very short pe
riod
of
time has
sev
e
ral variations, result
ing in signal va
riability substanti
a
l deviati
on from the
ori
g
inal
feature, feat
ure mutations dramaticall
y
. Assu
me that the signal variability of mutations,
according to
the Equation (1) can le
arn, sig
nal variability increase, will ca
use the si
gn
al
variation
ch
a
r
acte
ri
stics of
mixed
co
efficient.
A
c
cord
ing to th
e Eq
uation
(2
)
ca
n lea
r
n,
sig
n
a
l
variability characteri
stics
of mix
ed coefficient increases will
caus
e signal variability coefficient
increa
se
mut
a
tion. Acco
rding to
the
Equation
(3)
can
lea
r
n
,
sign
al vari
ability mutation
coeffici
ent increases will ca
use the
signal variation det
ection method [9-10].
In orde
r to avoid the above
defects, thi
s
pape
r put
s forwa
r
d a
kind
of borrowi
ng i
n
th
e
biologi
cal a
r
tificial immun
e
this con
c
e
p
t detection
method, th
e rob
o
t multisen
so
r si
gn
al
variation test
method. Through the e
s
t
ablishment
of
signal vari
ation ch
ara
c
te
ristics dynami
c
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
Based o
n
Artificial Imm
une Algorithm
of
Rob
o
t Mu
lti-Sensor Sign
al Variation
…
(Hon
gwei Yan
)
7601
cha
nge e
qua
tion, and the
n
in the data
base select
signal vari
atio
n ch
ara
c
te
ristics. Thi
s
wa
y,
you avoid
th
e different
si
gnal va
riation
ch
aracte
risti
c
s cau
s
ed
b
y
the mixed
sign
al va
riation
cha
r
a
c
teri
stics of the def
ects
of muta
tion and
red
u
ce the
sig
n
a
l variation f
o
r dete
c
tion
of
resi
dual rate.
3.
Robo
t Sensi
ng Signal Multiple Muta
tion Detectio
n Method
Rob
o
t se
nsi
n
g sig
nal m
u
tation dete
c
tio
n
metho
d
, the re
sea
r
ch fi
eld of the
ro
bot is a
major i
s
sue,
the rob
o
t ca
n accu
rately reali
z
e the h
u
man requi
re
ments, qui
ckl
y
adapt to the
environ
ment
play a de
ci
si
ve role. T
he
use
of
traditi
onal
sign
al variation
dete
c
tion a
p
p
r
oa
ch,
can
not be
avoided
be
cau
s
e the si
gnal v
a
riation
ch
ar
a
c
teri
stics of
mixed sig
nal
cau
s
e
d
by th
e
variability of mutations i
n
the defect,
causing
the signal vari
ation detection missing rate
increa
se. Thi
s
pa
per
pre
s
ent
s a met
hod ba
se
d o
n
Artificial Immune Alg
o
r
ithm for
sig
nal
mutation dete
c
tion metho
d
. Signal variati
on dete
c
tion
diagram a
s
shown in Figu
re 2.
Figure 2. Signal Variatio
n Test Block Di
agra
m
The Dy
namic Alter Equa
tion of Sign
al Variation
Feature Is
Built:
c
H
is the d
a
ta
colle
ction of signal variatio
n feature,
H
is sample nu
mbe
r
of the signal
variation dat
a colle
ctio
n
in the spe
c
ific pe
riod. T
h
e
dynamic alte
r situ
at
ion of
sign
al variati
on feature
ca
n be d
e
scri
b
ed
by the formul
a (4).
u
y
H
y
H
u
u
H
u
u
H
e
e
)
(
)
(
)
(
(
4
)
u
is a n
e
w
rob
o
t multi-sen
s
or si
gnal va
ri
ation charact
e
risti
c
.
u
y
H
e
)
/
(
e
is
Signal varia
b
ility update time. Affinity c
oefficient is(
). Signal variability during
the updat
e
pro
c
e
ss i
s
repla
c
ed with
the feature
e
H
. Sensor info
rmation u
s
in
g (
y
)
n
or
ma
l op
e
r
a
t
ion
behavio
r de
scriptio
n. Each signal vari
ability of
the
update process inclu
d
e
s
sen
s
or u
p
d
a
te
cy
cle p
a
ra
me
t
e
rs (
) an
d affinity paramet
ers
(
). Need t
o
use the fo
rmula (5
):
1
)
(
)
1
(
)
(
)
1
(
u
u
u
u
u
(
5
)
In the above
formula, Th
rough the
cal
c
ulatio
n to o
b
tain the si
g
nal variation
and the
affinity between the norm
a
l sensor si
gn
al param
et
ers, used to de
scribe the de
gree of affinity
con
s
tant a
ccumulation. S
ensor info
rm
ation upd
ate
d
after ea
ch
iteration
s
, the cumulativ
e
pro
c
e
ssi
ng
sensor i
n
form
ation
cycle
p
a
ram
e
ter
values. Hypoth
e
s
ized affinity
re
gistration is
su
ccessful, th
en you
will n
eed to affinity para
m
et
ers
of accum
u
lation. Usually d
i
vided into th
e
f
o
llowin
g
t
h
re
e kind
s of
cir
c
umst
an
ce
s:
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN: 2
087-278X
TELKOM
NIKA
Vol. 11, No
. 12, Dece
mb
er 201
3: 759
9 – 7604
7602
1) Hypothe
si
s
u
, then the sig
n
a
l
variability is activate, sign
al variability is
trans
formed into a s
a
mp
le c
h
arac
teris
t
ic.
2) Hypothe
si
s
u
,
then the
se
nso
r
info
rma
t
ion ch
ara
c
te
ristics of lo
w
affinity accum
u
lation re
sult,
the continu
e
d
need for a
c
cumul
a
tion.
3) Hypothe
si
s
u
, then the
sig
n
a
l variation
chara
c
te
risti
c
s of cumul
a
tive
results consi
s
tent with
the
a
ffinity measure, it
can determi
ne signal
vari
ability has
been
compl
e
tely repla
c
ed
by ne
w featu
r
es. Am
ong
them, the
ch
ara
c
te
rist
ic
is the sign
a
l
variation cha
r
acteri
stic affin
i
ty profile metrics.
Signal varia
b
ility of dynamic tra
n
sfo
r
mation pr
oce
ss, the n
eed
for ch
ara
c
te
ristics of
cro
s
s processing. In the si
gnal
variatio
n
detection, si
gnal varia
b
ility hypothesi
s
rene
wal
spe
e
d
of (
V
)
co
ntinu
e
s to i
n
cre
a
se, then the
sensor
ope
rati
ng data
tra
n
smissi
on
sp
ee
d incre
a
sed
by
(
V
),
V=
l
1
v
d
e
scrib
e
d
the
rel
a
tionship
bet
wee
n
the
two.
L
1
is si
gnal
varia
b
ility u
pdate
rate
a
nd
operating
rat
e
of data tra
n
sfer
co
rrelat
ion coefficie
n
t. Set signal
variability dat
a coll
ectio
n
is
H
2
={
h
1
,h
2
,L,h
i
,…L,h
p
}, Arbit
r
ary
signal variability variability is(
h
i
) , the poi
nt mut
a
tion rate
is
(
q
i
).
To get u
pdat
es the
proce
s
sed
sign
al variability data
set
H
2
={
h
1
,h
2
,L,h
j
,…L,h
p
}. Among th
e
m
,
the si
gnal va
riation cha
r
a
c
teristi
c
s of variability in
the t
r
eatme
nt of i
n
tersectio
n
p
o
int di
stributi
o
n
can u
s
e the f
o
rmul
a (6
):
L
e
Y
P
i
,
3
,
2
,
1
,
!
)
(
1
2
(
6
)
Y
is signal va
riability of t
he numbe
r of cross point
s.
Upd
a
te Signa
l Variation Ch
ara
c
teri
stic Databa
se
:
V
is a sen
s
o
r
ope
rat
i
ng feature d
a
ta
set, whi
c
h in
clud
es two subsets of da
ta (
T
)an
d
(
ST
) , whe
r
e (
T
) is the normal sensor
informatio
n o
peratio
n data
sub
s
ets, (
ST
) sensor si
gnal
mutation ope
ration is a su
b
s
et of data.
(
T
) incl
ude
s th
e no
rmal
sen
s
or cha
r
a
c
teristic i
n
form
ation a
nd th
e af
finity param
eters,
se
nso
r
norm
a
l op
era
t
ing characte
ristics
of info
rmation in
clu
d
ed in
the o
p
e
r
ation
of data
flow,
control
terminal i
n
terf
ace i
n
form
ation, ro
bot mul
t
i sen
s
o
r
rel
a
ted pa
ram
e
ters. Sen
s
or no
rmal op
eratin
g
c
h
arac
teris
t
ic information through t
he formal
desc
r
iption:
{1000111010
0000101011
1011
1100
10
1010
}, The re
lationship bet
wee
n
(
T
) and (
ST
) can u
s
e the
formula (7):
ST
T
V
ST
T
(7)
Variation
in
si
gnal
dete
c
tio
n
p
r
o
c
e
ss, i
s
to
achieve
a
c
curate j
udgm
ent of
se
nsor sig
nal
operation be
havior wh
eth
e
r
to belo
n
g
to
the sig
n
a
l
mutation
o
peratio
n. Se
nso
r
info
rmat
ion
norm
a
l ope
ra
tion data in th
e data set to rep
r
e
s
ent
the
normal
ope
rating features from the dat
a
colle
ction, assumin
g
a
r
bitrary sampl
e
a
s
sign
al muta
tion op
eratio
n
sa
mple
dete
c
tion
antibo
d
y
,
then to th
e
advantag
e of
the a
r
tificial
immun
e
me
thod for co
mplete
self t
o
lera
nce, thu
s
obtainin
g
the
sign
al vari
ation featu
r
e
de
tection fo
rmul
a. According
to the (
T
) ch
ara
c
t
e
ri
st
ic
s of
data
set, To
be a
b
le to
se
lect o
ne of th
e sample
T·B
j
(
j=1,2,L
),U
s
e the fo
rmula
(8
) to
rev
e
r
s
e
transfo
rmatio
n operation:
)
,
2
,
1
(
L
j
B
T
P
j
(
8
)
The results e
s
tabli
s
h a
se
t of data, usi
ng
ST
={
P
1
,P
2
,L
} de
scriptio
n. The d
a
ta
of the
numbe
r of ele
m
ents in the
colle
ction a
r
e
1. From the
ST
data colle
ction of arbit
r
ary element
s
P
j
as
signal vari
ability to
detect candidate
detector,
P
j
a
nd sample
s i
n
the
sampl
e
are
cal
c
ul
ate
d
in T,
P
j
={
b
1
,b
2
,L,b
12
} ca
n
be
T·B
={
c
1
,c
2
,L,c
12
}(
j=
1
,
2,L
)
.
If me
as
ure
is
and
is
a con
s
tant,
then use the formul
a (9
) to cal
c
ulate
the
correl
ation co
efficient of two:
12
1
j
E
(9)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
Based o
n
Artificial Imm
une Algorithm
of
Rob
o
t Mu
lti-Sensor Sign
al Variation
…
(Hon
gwei Yan
)
7603
Hypothesi
s
β
j
=c
j
, you
ca
n
get
ε
=
1, Hyp
o
thesi
s
β
j
≠
c
j
,, you ca
n g
e
t
ε
=
0. If
E<
ε
, then can
determi
ne th
e sampl
e
si
gnal
mutatio
n
op
eration,
othe
rwi
s
e, t
he
sam
p
le i
s
n
o
rm
al b
r
ain
operation. A
c
cordi
ng to
the ab
ove d
e
scrib
ed,
to
establi
s
h th
e
sign
al muta
tion ope
ratio
n
cha
r
a
c
teri
stics of the dynamic
eq
uatio
n, in the update sig
nal variation d
a
ta
base sel
e
ctio
n
intrusi
on feat
ure, thu
s
com
p
leting the si
gnal variatio
n
detection.
4.
The Simulation Res
u
lts
Variation in
sign
al dete
c
tion process, need
to extra
c
t the sig
nal
variability, will signa
l
variability and signal vari
ability in the featur
e
database sampl
e
is
co
mpared, can
com
p
lete
signal vari
ation detection. Mult
i
sensor signal variability may be
i
n
a very
short period of
time
has several v
a
riations,
causing the robot
sensor si
gnal vari
ability
substantial devi
a
tion from the
origin
al feature, feature mu
tations
drama
t
ically. Assum
e
that the
sig
nal variation
cha
r
a
c
teri
stics
of mutation,
usin
g the tra
d
itional al
gori
t
hms of
sign
al variatio
n d
e
tection,
can
not be
avoid
e
d
becau
se the
signal va
ri
ation ch
aracteristics of f
a
st multip
le
variation, caused by the
cha
r
a
c
teri
stics of deg
radati
on, redu
ce
s
the sig
nal mut
a
tion dete
c
tio
n
accuracy.
In orde
r to verify the effectiveness of this
algo
rithm, the nee
d fo
r contra
st experi
m
ent.
Establishme
n
t
of robot m
u
l
t
i sen
s
o
r
mod
e
l, re
sp
e
c
tively using th
e traditional
algo
rithm an
d the
algorith
m
of robot multi
se
nso
r
mo
del 1
0
dete
c
tion.
Whe
r
ein,
rele
vant para
m
et
ers are set a
s
follows
:
450
,
500
,
5
,
39
.
0
,
389
,
1000
m
l
t
V
i
n
, each expe
rime
nt on 100
ro
bot multi
sen
s
o
r
stro
n
g
variatio
n i
n
si
gnal
det
ection,
re
co
rd do
es not
su
ccessf
ully detecte
d sig
nal
numbe
rs as t
he missin
g ra
te measure, f
o
r ea
ch
sign
a
l
variation det
ection
re
sults for calib
ratio
n
,
spe
c
ific di
stri
bution a
s
sh
o
w
n in Figu
re
3.
Figure 3. Different Te
sting
Method
s Miss Rate
Contrast
From Fi
gure 3, can b
e
se
en cle
a
rly, using th
is al
gori
t
hm for ro
bot
multi sen
s
o
r
signa
l
variability to detect missin
g rate cu
rve is far lo
wer th
an the traditi
onal algo
rith
m, this algorit
hm
in signal vari
ability expressed mi
xed
case, signal vari
ation detecti
on has the
certai
n superiority.
The exp
e
rim
ental d
a
ta ca
n be
re
co
rde
d
, in Ta
ble
1
and
Tabl
e
2. Wh
erei
n,
Table
1
sign
al
variability is i
ndep
ende
nt
of context, si
gnal vari
ab
ilit
y in the dete
c
tion of releva
nt data. Tabl
e 2
sign
al variabil
i
ty is mixed case, si
gnal va
ri
ability in the detectio
n
of relevant data.
Variation i
n
si
gnal d
e
tectio
n process, thr
ough th
e d
a
ta in Ta
ble
1
are
analy
z
ed
to learn
that, if signal
variability is indepe
nde
nt of each
oth
e
r, so th
e u
s
e of this
al
gorithm
sign
al
variation d
e
tection
error
and the t
r
adit
i
onal al
gor
ith
m
is b
a
si
call
y the sam
e
. Table
2 thro
ug
h
analyze the data to learn that, assu
ming rob
o
t multi sen
s
or signal varia
b
ility are mixed
together, th
e
n
u
s
ing t
h
is
a
l
gorithm
si
gn
al variatio
n d
e
tection
erro
r is fa
r le
ss th
an the
traditio
nal
algorith
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN: 2
087-278X
TELKOM
NIKA
Vol. 11, No
. 12, Dece
mb
er 201
3: 759
9 – 7604
7604
Table 1. Biolo
g
ical Sign
al Variability Inde
pend
ent Data
Table
Signal variation detection data
The tra
d
itional algorithm
This algorithm
Freque
nc
y
of tes
t
ing
100
100
Detection signal variation number
24
26
The actual signal variation numbe
r
32
32
Detection of err
o
r
23.3%
20%
Table 2. Biolo
g
ical Sign
al Variation
Cha
r
acteri
stics of
Mixed Data T
able
Signal variation detection data
The tra
d
itional algorithm
This algorithm
Freque
nc
y
of tes
t
ing
100
100
Detection signal variation number
20
22
The actual signal variation numbe
r
30
30
Detection of err
o
r
36.7%
23.3%
5. Conclu
sion
This p
ape
r p
r
esents
a bo
rro
ws this
co
nce
p
t in biol
ogy, artificial
immune d
e
tectio
n
method to d
e
t
ect the variat
ion of ro
bot
multi sen
s
o
r
sign
al. Throu
gh the e
s
tabli
s
hme
n
t of sig
nal
variability
of dynamic equation,
to update the si
gnal variation
characteristi
c
database, thus
reali
z
ing
the
sign
al vari
ation d
e
tectio
n. The
ex
pe
ri
ment p
r
ove
s
, this
algo
rith
m imp
r
oves t
h
e
accuracy of d
e
tecting
sign
al variation.
Ackn
o
w
l
e
dg
ement
The wo
rk d
e
scrib
ed in this
pape
r ha
s be
en
sup
p
o
r
ted
by the Shanxi youth scien
ce and
techn
o
logy rese
arch fou
n
dation (No.
2011
0210
26
): Control val
v
e hydrauli
c
system
sca
l
e
synchro
nou
s control a
n
d
less d
egree
s of freed
om
mixed drive
parallel m
a
nipulato
r
de
sign
theory re
se
arch. The a
u
tho
r
s
would li
ke to expre
ss the
i
r gratitud
e for the su
ppo
rt of this study.
Referen
ces
[1] Hu
HW.
Mu
lti-
source
infor
m
a
t
ion fus
i
on
in
r
obot
obstacl
e
detectio
n
a
p
p
l
i
c
ation
. P
o
p
u
la
r busi
ness.
201
0; 11(1): 29
5.
[2]
Zhu DQ, Liu YG.
Informati
o
n
fusion
meth
od for
fault di
agn
o
s
is
. Co
ntrol
an
d d
e
cisi
on.
20
07; 2
2
(1
2):
132
1-13
28.
[3]
Huang YR, Wei YY.
B
a
sed
on
t
he multi sensor infor
m
ation
fusio
n
a
nd neur
al net
w
o
rk
to
fault
dia
gnos
is of stea
m turbi
n
e
. C
h
in
a Electric P
o
w
e
r. 20
10 ; 4
3
(3): 46-5
0
.
[4]
Yang
XR, Sh
en JY, Luo H
.
Artificial immune the
o
r
y
ap
plicati
on i
n
ne
t
w
ork i
n
trusi
o
n
detection
.
Co
mp
uter Engi
neer
ing.
Be
iji
n
g
. 2003.
[5]
Guo Q, Ho
ng
BR. T
r
acking
control
for m
obil
e
r
obots b
y
usin
gthe
w
a
velet
i
n
terp
olat
ion
m
e
tho
d
.
Journ
a
l of Co
mputer Res
earc
h
and D
e
ve
lop
m
e
n
t
. 2003 ; 4
0
(6): 856-
86
0.
[6]
W
ang CZ
,
Ji YB,
Sun DH. Rese
arch on
app
l
y
in
g
w
a
ve
l
e
t
an
al
ysis
to path
s
y
nt
hes
is
for
co
upl
er
curves of pla
n
a
r
four bar link
a
ges.
Chi
nes
e Journ
a
l of Mech
anic
a
l Eng
i
n
e
e
r
ing
. 20
04 ; 40(
8): 34-39.
[7]
W
u
XS,
Re
n
HY. An
optimi
z
ed fuzz
y
co
ntrol
appr
oac
h t
o
the
traject
o
r
y
track
i
n
g
of
n
onh
olo
nom
ic
mobil
e
ro
bots.
Journ
a
l
o
f So
uth Ch
in
a Un
i
v
ersity of T
e
c
hno
logy:
Natur
a
l Sci
enc
e Ed
ition
. 20
04
;
32(3): 44-
49.
[8]
Che
ng Z
X
, Ya
ng SZ
, F
e
n
g
XX. Pro
g
ress
e
s
in th
eor
etica
l
al
gorit
hm a
n
d
a
ppl
icatio
ns
of
w
a
v
e
let
ana
l
y
sis. Natio
nal D
e
fens
e Industr
y
Pr
ess, Beiji
ng. 20
07.
[9]
Li S, Ma GL,
Hu W
L
. T
r
acking co
ntrol
of car-like m
o
b
ile r
obot b
a
se
d o
n
back ste
ppi
ng
.
Journa
l of
Southe
ast Univ
ersity
. 2005; 3
5
(2): 248-
25
2.
[10]
Li X, Z
h
a
ng
HY, Li J.
T
r
ajector
y
co
ntro
l
of
w
h
e
e
l
ed
mobil
e
rob
o
t base
d
on fuzz
y PID.
Rob
o
t
T
e
chni
que and
Applic
ation
. 2
002; 2(5): 3
0
-3
3.
Evaluation Warning : The document was created with Spire.PDF for Python.