Sensors,
Год журнала:
2024,
Номер
24(24), С. 8053 - 8053
Опубликована: Дек. 17, 2024
Applying
deep
learning
to
unsupervised
bearing
fault
diagnosis
in
complex
industrial
environments
is
challenging.
Traditional
detection
methods
rely
on
labeled
data,
which
costly
and
labor-intensive
obtain.
This
paper
proposes
a
novel
approach,
WDCAE-LKA,
combining
wide
kernel
convolutional
autoencoder
(WDCAE)
with
large
attention
(LKA)
mechanism
improve
under
unlabeled
conditions,
the
adaptive
threshold
module
based
multi-layer
perceptron
(MLP)
dynamically
adjusts
thresholds,
boosting
model
robustness
imbalanced
scenarios.
Experimental
validation
two
datasets
(CWRU
customized
ball
screw
dataset)
demonstrates
that
proposed
outperforms
both
traditional
state-of-the-art
methods.
Notably,
WDCAE-LKA
achieved
an
average
diagnostic
accuracy
of
90.29%
varying
scenarios
CWRU
dataset
72.89%
showed
remarkable
conditions;
compared
advanced
models,
it
shortens
training
time
by
10–26%
improves
5–10%.
The
results
underscore
potential
as
robust
effective
solution
for
intelligent
applications.
Applied Sciences,
Год журнала:
2024,
Номер
14(11), С. 4380 - 4380
Опубликована: Май 22, 2024
This
paper
puts
forth
a
systematic
approach
for
evaluating
the
maturity
level
of
production
process
automation
in
context
digital
transformation
manufacturing
companies.
The
method
was
developed
to
address
absence
sector-specific
framework
assessing
growth,
line
with
Industry
5.0
guidelines
(incorporating
sustainability,
circular
economy,
and
human-centeredness).
survey
covers
six
core
areas
companies:
automation,
robotization
processes,
digitalization
warehouse
flexibility,
intralogistics,
end-to-end
integration
key
data
management
processes.
study
aimed
advance
through
improved
maturity.
surveyed
200
small-
medium-sized
businesses
operating
Poland
from
2022
2024.
presents
enterprise
operational
maturity,
covering
current
planned
levels
development
plans
next
three
years.
Transekonomika Akuntansi Bisnis dan Keuangan,
Год журнала:
2024,
Номер
4(3), С. 343 - 358
Опубликована: Июнь 25, 2024
Era
5.0
is
characterized
by
deeper
integration
between
technology
and
human
life,
where
Artificial
Intelligence
(AI)
plays
a
key
role
in
various
sectors,
including
digital
marketing.
This
study
aims
to
explore
how
AI
the
MSME
world
of
Palembang
city
applied
marketing,
as
well
its
impact
on
marketing
management
strategies
results.
research
uses
qualitative
approach
with
case
method
analyze
applications
promotions.
The
results
show
that
able
increase
operational
efficiency,
personalize
content,
data
analysis.
Apart
from
this,
also
helps
advertising
optimization,
market
trend
prediction,
improving
customer
experience.
application
faces
challenges,
such
privacy
issues,
implementation
costs,
need
for
specialized
expertise.
has
great
potential
revolutionize
MSMEs,
but
requires
strategic
ethical
overcome
existing
challenges.
Sustainability,
Год журнала:
2024,
Номер
16(15), С. 6275 - 6275
Опубликована: Июль 23, 2024
The
digitalisation
of
production
has
a
positive
impact
on
manufacturing
processes
in
terms
resource
efficiency
and
environmental
impact,
particularly
the
form
increased
as
well
cost
savings.
However,
use
technologies
is
also
associated
with
efforts
such
costs,
CO2
emissions,
raw
material
consumption.
When
planning
or
deciding
systems,
it
therefore
necessary
to
assess
whether
these
pay
off
sustainability
over
their
life
cycle.
This
literature
review
(based
PRISMA
guidelines)
analyses
relevance
assessment
its
methods
for
research.
reveals
that
research
focuses
benefits
manufacturing,
while
infancy.
There
need
further
holistic
technologies.
In
particular,
there
lack
consistently
link
economic
dimensions
sustainability,
guidance
application
production.
REID (Research and Evaluation in Education),
Год журнала:
2024,
Номер
10(1), С. 50 - 63
Опубликована: Июнь 30, 2024
This
study
aims
to
reveal
the
content
validity,
construct
and
reliability
of
instrument
for
evaluating
teaching
process
in
higher
education.
research
is
development
applying
ADDIE
model
from
Molenda.
The
indicators
evaluated
consist
context,
inputs,
processes,
products.
sample
consisted
1200
students
eight
faculties,
each
represented
by
three
programs.
Data
analysis
uses
stages:
validity
test
using
V-Aiken
method
involving
six
panellists
or
experts;
Confirmatory
Factor
Analysis
(CFA).
Quantitative
descriptive
interpretive
qualitative
used
Miles
Huberman
method.
results
showed
that
developed
evaluation
had
good
proof
content,
with
an
average
score
0.752,
which
was
high
category.
Universitas
Negeri
Yogyakarta's
instrument,
through
already
meets
exemplary
a
loading
factor
value
(
0.3).
It
has
composite
above
0.7
Cronbach's
alpha
0.6.
show
all
empirical
criteria
indicate
data
fit
against
model.
Sensors,
Год журнала:
2024,
Номер
24(24), С. 8053 - 8053
Опубликована: Дек. 17, 2024
Applying
deep
learning
to
unsupervised
bearing
fault
diagnosis
in
complex
industrial
environments
is
challenging.
Traditional
detection
methods
rely
on
labeled
data,
which
costly
and
labor-intensive
obtain.
This
paper
proposes
a
novel
approach,
WDCAE-LKA,
combining
wide
kernel
convolutional
autoencoder
(WDCAE)
with
large
attention
(LKA)
mechanism
improve
under
unlabeled
conditions,
the
adaptive
threshold
module
based
multi-layer
perceptron
(MLP)
dynamically
adjusts
thresholds,
boosting
model
robustness
imbalanced
scenarios.
Experimental
validation
two
datasets
(CWRU
customized
ball
screw
dataset)
demonstrates
that
proposed
outperforms
both
traditional
state-of-the-art
methods.
Notably,
WDCAE-LKA
achieved
an
average
diagnostic
accuracy
of
90.29%
varying
scenarios
CWRU
dataset
72.89%
showed
remarkable
conditions;
compared
advanced
models,
it
shortens
training
time
by
10–26%
improves
5–10%.
The
results
underscore
potential
as
robust
effective
solution
for
intelligent
applications.
