Metals,
Journal Year:
2024,
Volume and Issue:
14(10), P. 1106 - 1106
Published: Sept. 26, 2024
Automotive
chassis
components,
constructed
as
lap
joints
and
produced
by
gas
metal
arc
welding
(GMAW),
require
fatigue
durability.
The
properties
of
the
weld
in
a
joint
are
largely
determined
geometry
factors.
When
there
is
no
gap
or
consistent
joint,
improving
toe
can
alleviate
stress
concentration
enhance
properties.
However,
due
to
machining
tolerances,
it
difficult
completely
eliminate
consistently
manage
joint.
In
case
lap-welded
with
an
inconsistent
gap,
necessary
identify
factors
related
Evaluating
behavior
materials
welded
requires
significant
time
cost,
meaning
that
research
seeks
predict
essential.
More
needed
on
predicting
automotive
particularly
studies
gaps.
This
study
proposed
regression
model
for
based
crucial
gaps
using
statistical
analysis.
Welding
conditions
were
varied
order
build
various
geometries
configured
0,
0.2,
0.5,
1.0
mm,
87
S–N
curves
derived.
As
input
variables,
17
(7
lengths,
7
angles,
3
area
factors)
selected.
slope
curve
Basquin
from
safe
strength
selected
output
variables
prediction
develop
model.
Multiple
linear
models,
multiple
non-linear
second-order
polynomial
models
Backward
elimination
was
applied
simplify
reduce
overfitting.
Among
three
had
coefficient
determination
greater
than
0.86.
gaps,
representing
identified
through
standardized
coefficients,
four
proposed.
Fatigue & Fracture of Engineering Materials & Structures,
Journal Year:
2025,
Volume and Issue:
unknown
Published: March 26, 2025
ABSTRACT
This
study
conducts
an
in‐depth
analysis
of
the
mechanical
property
changes
6005A‐T6
aluminum
alloy
under
different
fatigue
aging
states
(the
process
in
which
material's
performance
gradually
deteriorates
over
time
cyclic
loading).
First,
evolution
surface
displacement
fields
was
analyzed
using
digital
image
correlation
combined
with
various
levels
pretreatment.
Through
single‐cycle
tests
and
tensile
tests,
field
responses
material
degradation
were
examined,
ultimate
strength,
yield
elongation,
section
shrinkage
further
analyzed.
Based
on
existing
strength‐tensile
strength‐fatigue
strength
(Y‐T‐F)
model,
improved
approach,
Y‐T‐F‐II
proposed
to
account
for
effects
validated
prediction,
achieving
a
maximum
error
only
0.17%.
The
results
showed
that
significantly
affects
ductility,
toughness
alloy,
model
provides
more
accurate
predictions
states.
Fatigue & Fracture of Engineering Materials & Structures,
Journal Year:
2025,
Volume and Issue:
unknown
Published: May 24, 2025
ABSTRACT
A
novel
approach
was
proposed
and
implemented
to
assess
the
confidence
of
individual
class
predictions
made
by
convolutional
neural
networks
trained
identify
type
fracture
in
metals.
This
involves
utilizing
contextual
evidence
form
images
scores,
which
serve
as
indicators
for
determining
certainty
predictions.
first
tested
on
both
shallow
deep
employing
four
publicly
available
image
datasets:
MNIST,
EMNIST,
FMNIST,
CIFAR10,
subsequently
validated
an
in‐house
steel
dataset—FRAC,
containing
ductile
brittle
images.
The
effectiveness
method
is
producing
scores
data
other
datasets
selected
from
datasets.
CIFAR‐10
dataset
yielded
lowest
mean
score
78
model,
with
over
50%
representative
test
instances
receiving
a
below
90,
indicating
lower
model's
In
contrast,
CNN
model
used
achieved
99,
0%
suggesting
high
level
its
enhances
interpretability
provides
greater
insight
into
their
outputs.
This
study
proposed
a
regression
model
for
predicting
fatigue
properties
based
on
crucial
weld
geometry
factors
in
lap-welded
joints
with
gaps
using
statistical
analysis.
Welding
conditions
were
varied
to
build
various
geometries
configured
lap
from
of
0,
0.2,
0.5,
and
1.0
mm,
87
S-N
curves
the
derived.
As
input
variables,
17
(7
lengths,
7
angles,
3
area
factors)
selected.
The
slope
curve
Basquin
safe
strength
selected
as
output
variables
prediction
develop
model.
Multiple
linear
models,
multiple
non-linear
second-order
polynomial
models
predict
properties.
Backward
elimination
was
applied
simplify
reduce
overfitting.
Among
three
had
coefficient
determination
greater
than
0.86.
In
gaps,
representing
identified
through
standardized
coefficients,
four
related
stress
concentration
proposed.
