nth modified approximation coefficient at level
j
|
nth modified detail coefficient at level j
|
|
λ
|
threshold of cj[n]
|
|
MAD(cL − 1[n])
|
mean absolute deviation of cL − 1[n]
|
|
modified approximation coefficients
|
|
avg
|
arithmetic mean or average of
|
|
std
|
standard derivation of
|
|
max
|
maximal magnitude of
|
|
min
|
minimal magnitude of
|
|
ptp
|
peak‐to‐peak amplitude of
|
|
kurt
|
kurtosis for measuring the peakedness of
|
|
skew
|
skewness for measuring the asymmetry of
|
|
RMS
|
root mean square value of
|
|
CF
|
crest factor of
|
|
T
|
duration of a continuous‐time signal
|
|
t
|
specific time t of a continuous‐time signals
|
|
cr
xy
|
correlation coefficient value of two signals: x(t) and y(t)
|
|
CR
xy
|
cross‐correlation value of two signals: x(t) and y(t)
|
|
γ
xy
|
covariance value of two signals: x(t) and y(t)
|
|
m
|
displacement unit of a signal
|
|
*
|
complex conjugate of a signal
|
|
SF
CR(xy)
|
a feature set of CRxy
|
|
FFT[n]
|
frequency amplitude at the nth Hz by FFT
|
|
SF
FFT(q)
|
qth certain frequency band delimited by a lower frequency and an upper frequency
|
|
Q
|
number of frequency band
|
|
uf
q
|
qth upper frequency of the critical characteristics; and
|
|
lf
q
|
qth lower frequency of the critical characteristics.
|
|
SF
FFT(q)
|
qth FFT‐based SF
|
|
u
|
uth wavelet packet at level L, u= 1, 2, …, L;
|
|
v
|
subband length for each wavelet packet at level L, v = N/2L
|
|
SF
WPT(u)
|
uth WPT‐based SF
|
|
h
|
compressed code of the middle layer in AEN;
|
|
output reconstructed from c in the middle layer of AEN;
|
|
f
EN
|
encoder layer of AEN;
|
|
f
DE
|
decoder layer of AEN;
|
|
f
a
|
activation function of AEN;
|
|
W
EN
|
network weight for node in the encoder;
|
|
W
DE
|
network weight for node decoder;
|
|
b
EN
|
bias for node in the encoder layer;
|
|
b
DE
|
bias for node in the decoder layer
|
|
SF
AEN
|
SFs extracted from AEN
|
References
СКАЧАТЬ
