Structural Health Monitoring,
Год журнала:
2025,
Номер
unknown
Опубликована: Апрель 27, 2025
With
the
rapid
development
of
deep
learning,
edge
intelligence
applications
(EIA)
have
achieved
numerous
results.
However,
redundant
parameters
model
and
strong
noise
pollution
pose
challenges
to
EIA
for
bearing
fault
diagnosis.
To
solve
these
challenges,
a
with
lightweight
network
antinoise
ability
was
proposed
First,
novel
pluggable
channel
slimming
module
designed
make
lightweight,
which
can
effectively
reduce
computation
model.
Second,
an
learning
is
proposed,
has
discriminator
enhance
network’s
feature
extraction
capability
through
supervised
learning.
Finally,
adaptive
input
generalization
model,
adaptively
adjust
information
under
different
application
environments
improve
stability
accuracy
The
performance
verified
test
rig
experiments
on
two
types
train
axle
box
bearings
datasets,
indicated
achieves
more
than
89%
diagnostic
at
−10
dB.
Measurement Science and Technology,
Год журнала:
2024,
Номер
35(12), С. 125109 - 125109
Опубликована: Авг. 27, 2024
Abstract
Traction
motor
bearings,
serving
as
a
critical
component
in
trains,
have
significant
impact
on
ensuring
the
safety
of
train
operations.
However,
there
is
scarcity
sample
data
for
bearing
failures
during
operations,
and
complex
variable
operating
conditions
bearings
result
differences
domain
distribution.
Traditional
cross-domain
fault
diagnosis
methods
are
no
longer
adequate
addressing
faults.
Therefore,
this
study
proposes
novel
adversarial
domain-adaptation
meta-learning
network
(NADMN)
purpose
diagnosing
Firstly,
deep
convolutional
neural
proposed,
which
enhances
model’s
feature
extraction
capability
by
incorporating
attention
mechanisms.
Moreover,
employing
adaptation
learning
strategy,
it
effectively
extracts
domain-invariant
features
from
both
source
target
domains,
thereby
achieving
generalization
across
different
domains.
Three
experiments
carried
out,
superiority
NADMN
proved
charts,
confusion
matrix
visualization
techniques.
Compared
with
other
five
methods,
showed
obvious
advantages
diagnostic
scenarios
characterized
changes
Sensors,
Год журнала:
2025,
Номер
25(3), С. 874 - 874
Опубликована: Янв. 31, 2025
Marine
Current
Turbines
(MCTs)
play
a
critical
role
in
converting
the
kinetic
energy
of
water
into
electricity.
However,
due
to
influence
marine
organisms,
current
equipment
often
experiences
imbalance
faults.
Additionally,
affected
by
underwater
environment,
fault
characteristics
are
submerged
disturbances
such
as
waves
and
turbulence.
Against
background
above
problems,
this
article
proposes
detection
strategy
based
on
Generalized
Likelihood
Ratio
Test
(GLRT)
detector.
Firstly,
simulation
model
MCT
system
is
established
obtain
prior
knowledge.
Then,
combining
Matrix
Pencil
Method
(MPM)
for
calculating
instantaneous
frequency,
metrics
selected
proposed
GLRT
At
end,
turbine
experimental
platform
established,
which
can
simulate
imbalanced
faults
environmental
disturbances,
helping
verify
effectiveness
strategy.
The
results
indicate
that
detect
complex
environments.
Imbalance
main
manifestation
blade
attachments.
Thus,
it
very
meaningful
accomplish
order
maintain
working
system.
