2022 IEEE Energy Conversion Congress and Exposition (ECCE),
Journal Year:
2023,
Volume and Issue:
unknown, P. 4457 - 4464
Published: Oct. 29, 2023
Stator
inter-turn
faults
(SITFs)
are
electrical
abnormalities
in
the
windings
of
a
motor
or
generator,
resulting
from
short
circuits
between
adjacent
coil
turns,
potentially
leading
to
reduced
performance
even
catastrophic
failures.
This
paper
aims
detect
SITFs
and
classify
their
level
severity
using
combination
prominence-based
features
recently
developed
neural
networks
that
rely
on
self-attention
mechanisms.
The
approach
involves
transforming
3-phase
currents
extended
Park
Vector
(EVPA),
extracting
based
prominence
frequency
spectrum,
studying
geometry
gain
important
insights
about
data.
After
this
feature-engineering
data
exploration
step,
neural-based
classifiers
have
been
trained
tested.
Through
comparative
study
with
other
approaches,
Transformer
Encoder
achieves
highest
classification
accuracy
97.25%
when
tested
experimental
data,
outperforming
networks.
authors
also
present
importance
maps
for
exploring
interpretability
Encoder,
revealing
significant
contribution
classification.
iScience,
Journal Year:
2024,
Volume and Issue:
27(5), P. 109673 - 109673
Published: April 4, 2024
Machine
learning
interatomic
potential
(MLIP)
overcomes
the
challenges
of
high
computational
costs
in
density-functional
theory
and
relatively
low
accuracy
classical
large-scale
molecular
dynamics,
facilitating
more
efficient
precise
simulations
materials
research
design.
In
this
review,
current
state
four
essential
stages
MLIP
is
discussed,
including
data
generation
methods,
material
structure
descriptors,
six
unique
machine
algorithms,
available
software.
Furthermore,
applications
various
fields
are
investigated,
notably
phase-change
memory
materials,
searching,
properties
predicting,
pre-trained
universal
models.
Eventually,
future
perspectives,
consisting
standard
datasets,
transferability,
generalization,
trade-off
between
complexity
MLIPs,
reported.
Journal of Materials Informatics,
Journal Year:
2023,
Volume and Issue:
3(3)
Published: July 13, 2023
Virtual
sample
generation
(VSG),
as
a
cutting-edge
technique,
has
been
successfully
applied
in
machine
learning-assisted
materials
design
and
discovery.
A
virtual
without
experimental
validation
is
defined
an
unknown
sample,
which
either
expanded
from
the
original
data
distribution
for
modeling
or
designed
via
algorithms
predicting.
This
review
aims
to
discuss
applications
of
VSG
techniques
discovery
based
on
research
progress
recent
years.
First,
we
summarize
commonly
used
expansion
training
set,
including
Bootstrap,
Monte
Carlo,
particle
swarm
optimization,
mega
trend
diffusion,
Gaussian
mixture
model,
random
forest,
generative
adversarial
networks.
Next,
frequently
employed
searching
are
introduced,
efficient
global
proactive
progress.
Then,
universally
adopted
inverse
methods
presented,
genetic
algorithm,
Bayesian
pattern
recognition
projection.
Finally,
future
directions
proposed.
Journal of Manufacturing and Materials Processing,
Journal Year:
2025,
Volume and Issue:
9(2), P. 51 - 51
Published: Feb. 6, 2025
Additive
manufacturing
(AM)
has
revolutionised
the
production
of
customised
components
across
industries
such
as
aerospace,
automotive,
healthcare,
electronics,
and
renewable
energy
industries.
Offering
unmatched
design
freedom,
reduced
time-to-market,
minimised
material
waste,
AM
enables
fabrication
high-quality,
products
with
greater
sustainability
compared
to
traditional
methods
like
machining
injection
moulding.
Additionally,
reduces
consumption,
resource
requirements,
CO2
emissions
throughout
a
material’s
lifecycle,
aligning
global
goals.
This
paper
highlights
insights
into
polymers,
comparing
bio-based
polymers.
Bio-based
polymers
exhibit
lower
carbon
footprints
during
but
may
face
challenges
in
durability
mechanical
performance.
Conversely,
while
more
robust,
require
higher
inputs
contribute
emissions.
Polymer
composites
tailored
for
further
enhance
properties
support
development
innovative,
eco-friendly
solutions.
Special
Issue
brings
together
cutting-edge
research
on
polymer
AM,
focusing
processing
techniques,
microstructure–property
relationships,
performance,
sustainable
practices.
These
advancements
underscore
AM’s
transformative
potential
deliver
versatile,
high-performance
solutions
diverse
Nanomaterials,
Journal Year:
2024,
Volume and Issue:
14(5), P. 445 - 445
Published: Feb. 28, 2024
The
band
gap
is
a
key
parameter
in
semiconductor
materials
that
essential
for
advancing
optoelectronic
device
development.
Accurately
predicting
gaps
of
at
low
cost
significant
challenge
science.
Although
many
machine
learning
(ML)
models
prediction
already
exist,
they
often
suffer
from
interpretability
and
lack
theoretical
support
physical
perspective.
In
this
study,
we
address
these
challenges
by
using
combination
traditional
ML
algorithms
the
‘white-box’
sure
independence
screening
sparsifying
operator
(SISSO)
approach.
Specifically,
enhance
accuracy
predictions
binary
semiconductors
integrating
importance
rankings
vector
regression
(SVR),
random
forests
(RF),
gradient
boosting
decision
trees
(GBDT)
with
SISSO
models.
Our
model
uses
only
intrinsic
features
constituent
elements
their
calculated
Perdew–Burke–Ernzerhof
method,
significantly
reducing
computational
demands.
We
have
applied
our
to
predict
1208
theoretically
stable
compounds.
Importantly,
highlights
critical
role
electronegativity
determining
material
gaps.
This
insight
not
enriches
understanding
principles
underlying
but
also
underscores
potential
approach
guiding
synthesis
new
valuable
materials.
