Information Technology Reference
In-Depth Information
work the Asymptotic Mean Integrated Squared Error (AMISE) with plug-in
bandwidth selection procedure is used to choose automatically the bandwidth
h [treated in Wand and Jones (
1994b
)]. In the proposed algorithm, KDE is used to
model a speci
c pattern for each motor condition, indeed the features of the current
signals are mapped in the two-dimensional principal component space, representing
speci
c signatures of the motor conditions.
2.1.3 Kullback
Leibler Divergence
-
Given two continuous PDFs f
1
ð
x
Þ
and fi
2
ð
x
Þ
, a measure of
“
divergence
”
or
“
dis-
tance
”
between fi
1
ð
x
Þ
versus f
2
ð
x
Þ
is given in Kullback and Leibler (
1951
), as:
Z
f
1
ð
x
Þ
I
1
:
2
ð
X
Þ
¼
f
1
ð
x
Þ
log
dx
;
ð
8
Þ
f
2
ð
x
Þ
d
R
and between f
2
ð
Þ
versus f
1
ð
Þ
x
x
is given by:
Z
f
2
ð
x
Þ
I
2
:
1
ð
X
Þ
¼
f
2
ð
x
Þ
log
dx
:
ð
9
Þ
f
1
ð
x
Þ
d
R
Therefore the K
-
L divergence between fi
1
ð
x
Þ
and fi
2
ð
x
Þ
is:
J
ð
f
1
;
f
2
Þ
¼I
1
:
2
ð
X
Þþ
I
2
:
1
ð
X
Þ
Z
ð
10
Þ
f
1
ð
x
Þ
¼
ð
f
1
ð
x
Þ
f
2
ð
x
Þ
Þ
log
dx
:
f
2
ð
x
Þ
d
R
L divergence, which repre-
sents a non negative measure between two PDFs. In the present work d is 2 and a
discrete form of K
The above equation is known as the symmetric K
-
-
L divergence is adopted:
n
grid
X
X
d
log
f
1
ð
x
ij
Þ
J
ð
f
1
;
f
2
Þ
¼
f
1
ð
x
ij
Þ
f
2
ð
x
ij
Þ
x
ij
Þ
:
ð
11
Þ
f
2
ð
i¼1
j¼1
ne a fault index: if if
X
is the PDF in the PCs
space estimated by KDE of the oncoming current measurements, the motor con-
dition is that which minimizes the K
The K
-
L divergence allows to de
-
L divergence between fi
X
and fi
i
that is the ith
PDF related to each motor condition:
ð
f
X
;
f
i
Þ;
ð
12
Þ
c
¼
arg min
i
J
where c is the classi
cation output.
Search WWH ::
Custom Search