npj Computational Materials,
Journal Year:
2024,
Volume and Issue:
10(1)
Published: July 14, 2024
Abstract
Accelerating
the
design
of
Ni-based
single
crystal
(SX)
superalloys
with
superior
creep
resistance
at
ultrahigh
temperatures
is
a
desirable
goal
but
extremely
challenging
task.
In
present
work,
deep
transfer
learning
neural
network
physical
constraints
for
rupture
life
prediction
constructed.
Transfer
enables
model
breaks
through
generalization
performance
barrier
in
extrapolation
space
temperature
properties
case
very
small
dataset,
which
key
to
achieving
above
goal.
demonstrated
be
effective
utilizing
prior
compositional
sensitivities
information
contained
pre-trained
model,
and
motivates
fine-tuned
capture
particular
relationship
between
composition
temperature.
Aiming
find
advanced
SX
applied
1200
°C,
proposed
learning-based
guides
us
superalloy
verified
~170
h
80
MPa,
exceeds
state-of-art
value
by
30%.
The
improved
γ/γ′
interface
strengthening,
effectively
regulated
Mo/Ta
ratio
form
γ′
rafting
longer,
flatter
interfaces
achieve
stronger
interfacial
bonding,
revealed
as
dominant
mechanism
behind
combining
experiments
first-principles
calculations.
Moreover,
excellent
ability
further
confirmed
enhance
efficiency
active
reducing
its
dependence
on
initial
dataset
size.
This
study
provides
pioneering
AI-driven
approach
rapid
development
aero-engine
blades.
Journal of Composites Science,
Journal Year:
2023,
Volume and Issue:
7(9), P. 364 - 364
Published: Sept. 1, 2023
The
determination
of
mechanical
properties
plays
a
crucial
role
in
utilizing
composite
materials
across
multiple
engineering
disciplines.
Recently,
there
has
been
substantial
interest
employing
artificial
intelligence,
particularly
machine
learning
and
deep
learning,
to
accurately
predict
the
materials.
This
comprehensive
review
paper
examines
applications
intelligence
forecasting
different
types
composites.
begins
with
an
overview
then
outlines
process
predicting
material
properties.
primary
focus
this
lies
exploring
various
techniques
employed
Furthermore,
highlights
theoretical
foundations,
strengths,
weaknesses
each
method
used
for
Finally,
based
on
findings,
discusses
key
challenges
suggests
future
research
directions
field
prediction,
offering
valuable
insights
further
exploration.
is
intended
serve
as
significant
reference
researchers
engaging
studies
within
domain.
Annual Review of Chemical and Biomolecular Engineering,
Journal Year:
2022,
Volume and Issue:
13(1), P. 235 - 254
Published: March 18, 2022
Designing
functional
materials
requires
a
deep
search
through
multidimensional
spaces
for
system
parameters
that
yield
desirable
material
properties.
For
cases
where
conventional
parameter
sweeps
or
trial-and-error
sampling
are
impractical,
inverse
methods
frame
design
as
constrained
optimization
problem
present
an
attractive
alternative.
However,
even
efficient
algorithms
require
time-
and
resource-intensive
characterization
of
properties
many
times
during
optimization,
imposing
bottleneck.
Approaches
incorporate
machine
learning
can
help
address
this
limitation
accelerate
the
discovery
with
targeted
In
article,
we
review
how
to
leverage
reduce
dimensionality
in
order
effectively
explore
space,
property
evaluation,
generate
unconventional
structures
optimal
We
also
discuss
promising
future
directions,
including
integration
into
multiple
stages
algorithm
interpretation
models
understand
relate
npj Computational Materials,
Journal Year:
2024,
Volume and Issue:
10(1)
Published: May 8, 2024
Abstract
Acquiring
reliable
microstructure
datasets
is
a
pivotal
step
toward
the
systematic
design
of
materials
with
aid
integrated
computational
engineering
(ICME)
approaches.
However,
obtaining
three-dimensional
(3D)
often
challenging
due
to
high
experimental
costs
or
technical
limitations,
while
acquiring
two-dimensional
(2D)
micrographs
comparatively
easier.
To
deal
this
issue,
study
proposes
novel
framework
called
‘Micro3Diff’
for
2D-to-3D
reconstruction
microstructures
using
diffusion-based
generative
models
(DGMs).
Specifically,
approach
solely
requires
pre-trained
DGMs
generation
2D
samples,
and
dimensionality
expansion
(2D-to-3D)
takes
place
only
during
process
(i.e.,
reverse
diffusion
process).
The
proposed
incorporates
concept
referred
as
‘multi-plane
denoising
diffusion’,
which
transforms
noisy
samples
latent
variables)
from
different
planes
into
data
structure
maintaining
spatial
connectivity
in
3D
space.
Furthermore,
harmonized
sampling
developed
address
possible
deviations
Markov
chain
expansion.
Combined,
we
demonstrate
feasibility
Micro3Diff
reconstructing
connected
slices
that
maintain
morphologically
equivalence
original
images.
validate
performance
Micro3Diff,
various
types
(synthetic
experimentally
observed)
are
reconstructed,
quality
generated
assessed
both
qualitatively
quantitatively.
successful
outcomes
inspire
potential
utilization
upcoming
ICME
applications
achieving
breakthrough
comprehending
manipulating
space
DGMs.
