Energies,
Год журнала:
2024,
Номер
17(16), С. 4098 - 4098
Опубликована: Авг. 18, 2024
Renewable
energy
accommodation
in
power
grids
leads
to
frequent
load
changes
plants.
Sensitive
turbine
fault
monitoring
technology
is
critical
ensure
the
stable
operation
of
system.
Existing
techniques
do
not
use
information
sufficiently
and
are
sensitive
early
signs.
To
solve
this
problem,
an
unsupervised
warning
method
based
on
hybrid
gain
a
convolutional
autoencoder
(CAE)
for
intermediate
flux
proposed.
A
high-precision
intermediate-stage
prediction
model
established
using
CAE.
The
calculation
proposed
filter
features
multi-dimensional
sensors.
Hampel
time
series
outlier
detection
introduced
deal
with
factors
such
as
sensor
faults
noise.
achieves
highest
diagnosis
accuracy
through
experiments
real
data
compared
traditional
methods.
Real
show
that
relatively
improves
diagnostic
by
average
2.12%
gate
recurrent
unit
networks,
long
short-term
memory
other
models.
Meanwhile,
can
effectively
improve
models,
maximum
1.89%
relative
improvement.
noteworthy
its
superiority
applicability.
Symmetry,
Год журнала:
2025,
Номер
17(3), С. 427 - 427
Опубликована: Март 12, 2025
For
mechanical
equipment
to
operate
normally,
rolling
bearings—which
are
crucial
parts
of
rotating
machinery—need
have
their
faults
diagnosed.
This
work
introduces
a
bearing
defect
diagnosis
technique
that
incorporates
three-channel
feature
fusion
and
is
based
on
enhanced
Residual
Networks
Bidirectional
long-
short-term
memory
networks
(ResNet-BiLSTM)
model.
The
can
effectively
establish
spatial-temporal
relationships
better
capture
complex
features
in
data
by
combining
the
powerful
spatial
extraction
capability
ResNet
bidirectional
temporal
modeling
BiLSTM.
Specifically,
one-dimensional
vibration
signals
first
transformed
into
two-dimensional
images
using
Continuous
Wavelet
Transform
(CWT)
Markov
Transition
Field
(MTF).
upgraded
ResNet-BiLSTM
network
then
used
extract
combine
original
signal
along
with
from
two
types
images.
Finally,
experimental
validation
performed
datasets.
results
show
compared
other
state-of-the-art
models,
computing
cost
greatly
reduced,
params
flops
at
15.4
MB
715.24
MB,
respectively,
running
time
single
batch
becomes
5.19
s.
fault
accuracy
reaches
99.53%
99.28%
for
datasets,
successfully
realizing
classification
task.
Machine Learning Science and Technology,
Год журнала:
2024,
Номер
6(1), С. 015005 - 015005
Опубликована: Дек. 17, 2024
Abstract
The
complexity
and
fusion
dynamism
of
the
modern
industrial
chemical
sectors
have
been
increasing
with
rapid
progress
IR
4.0–5.0.
transformative
characteristics
Industry
4.0–5.0
not
fully
explored
in
terms
fundamental
importance
explainability.
Traditional
monitoring
techniques
for
automatic
anomaly
detection,
identifying
potential
variables,
root
cause
analysis
fault
information
are
intelligent
enough
to
tackle
intricate
problems
real-time
practices
sectors.
This
study
presents
a
novel
dynamic
machine
learning
based
explainable
approach
address
issues
process
systems.
methodology
aims
detect
faults,
identify
their
key
causes
feature
analyze
path
propagation
time
magnitude
one
variable
another
impact.
proposed
using
domain
multivariate
granger-entropy-aided
independent
component
(DICA)—distributed
canonical
correlation
approach,
incorporating
dynamics
wrapping
supported
delay-signed
directed
graph.
utilized
application
processes
verified
continuous
stirred
tank
reactor
Tennessee
Eastman
as
practical
benchmarks.
framework’s
validations
efficiency
evaluated
established
such
classic
computed
ICA
DICA
standard
model
scenarios.
outcomes
results
showed
that
newly
developed
strategy
is
preferable
previous
approaches
regarding
explainability
robust
detection
identification
actual
high
FDRs
low
FARs.
Machines,
Год журнала:
2025,
Номер
13(2), С. 125 - 125
Опубликована: Фев. 7, 2025
Power
transformers
(PTs)
play
a
vital
role
in
the
electrical
power
system.
Assessing
their
health
to
predict
remaining
useful
life
is
essential
optimise
maintenance.
Scheduling
right
maintenance
for
equipment
at
time
ultimate
goal
of
any
system
utility.
Optimal
has
number
benefits:
human
and
social,
by
limiting
sudden
service
interruptions,
economic,
due
direct
indirect
costs
unscheduled
downtime.
PT
now
produces
large
amounts
easily
accessible
data
increasing
use
IoT,
sensors,
connectivity
between
physical
assets.
As
result,
transformer
prognostics
management
(PT-PHM)
methods
are
increasingly
moving
towards
artificial
intelligence
(AI)
techniques,
with
several
hundreds
scientific
papers
published
on
topic
PT-PHM
using
AI
techniques.
On
other
hand,
world
undergoing
new
evolution
third
generation
models:
large-scale
foundation
models.
What
current
state
research
PT-PHM?
trends
challenges
where
do
we
need
go
management?
This
paper
provides
comprehensive
review
art
analysing
more
than
200
papers,
mostly
journals.
Some
elements
guide
given
end
document.