Machine learning for computational science and engineering,
Journal Year:
2025,
Volume and Issue:
1(1)
Published: April 16, 2025
Abstract
This
study
introduces
a
physics-based
machine
learning
(
$$\phi
$$
ϕ
ML)
framework
for
modeling
both
brittle
and
ductile
fractures
in
elastic-viscoplastic
materials.
It
integrates
physical
principles,
including
governing
equations
constraints,
directly
into
the
neural
network
architecture.
Specifically,
feedforward
is
designed
to
embed
laws
within
its
architecture,
ensuring
thermodynamic
consistency.
Building
on
this
foundation,
synthetic
datasets
generated
from
finite
element-based
phase-field
fracture
simulations
are
employed
train
proposed
framework,
focusing
capturing
homogeneous,
one-dimensional
responses.
Detailed
analyses
performed
stored
elastic
energy
dissipated
work
due
plasticity
fracture,
demonstrating
capability
of
predict
essential
features.
The
ML
overcomes
shortcomings
classical
models,
which
rely
heavily
large
lack
guarantees
principles.
By
leveraging
physics-integrated
design,
demonstrates
exceptional
performance
predicting
key
properties
with
limited
training
data.
ensures
reliability,
efficiency,
consistency,
establishing
foundational
approach
integrating
computational
mechanics.
