Revue des composites et des matériaux avancés,
Год журнала:
2024,
Номер
34(6), С. 755 - 765
Опубликована: Дек. 28, 2024
Finding
the
California
Bearing
Ratio
(CBR)
of
soil
stabilised
by
an
environmentally
friendly
binder
composite
is
one
most
important
steps
in
designing
appropriate
mix.By
utilising
artificial
neural
network
(ANN)
to
forecast
parameters
and
additions
Portland
cement
Bamboo
Leaf
Ash
(BLA),
this
study
aims
estimate
treated
cement-lateritic
soils.The
precise
accurate
findings
are
obtained
selecting
six
factors
as
input
variables.Maximum
Dry
Density
(MDD)
(kg/m
3
),
Plasticity
Index
(PI)
(%),
Liquid
Limit
(LL)
Cement
(BLA)
OMC
(%)
were
variables.In
contrast,
output
variables
CBR
soaked
unsoaked
(%).1288
samples
from
a
database
used
investigation.Training
done
using
multilayer
perceptronbackpropagation
algorithm.The
topology
acquired
after
fixing
several
hidden
neurones.With
99.5%
accuracy
rate,
model
can
predict
results.
Purpose
Artificial
intelligence,
particularly
deep
learning
(DL),
has
increasingly
influenced
various
scientific
fields,
including
soil
mechanics.
This
paper
aims
to
present
a
novel
DL
application
of
long
short-term
memory
(LSTM)
networks
for
predicting
behaviour
during
constant
rate
strain
(CRS)
tests.
Design/methodology/approach
LSTMs
are
adept
at
capturing
long-term
dependencies
in
sequential
data,
making
them
suitable
the
complex,
nonlinear
stress–strain
soil.
evaluates
LSTM
configurations,
optimising
parameters
such
as
step
size,
batch
data
sampling
and
training
subset
size
balance
prediction
accuracy
computational
efficiency.
The
study
uses
comprehensive
set
from
numerical
finite
element
method
simulations
conducted
with
PLAXIS
2D
laboratory
CRS
Findings
proposed
model,
trained
on
lower
stress
levels,
accurately
forecasts
higher
levels.
optimal
setup
achieved
median
error
3.59%
5.10%
3.86%
presenting
setup’s
effectiveness.
Originality/value
approach
reduces
required
time
complete
extensive
testing,
aligning
sustainable
industrial
practices.
findings
suggest
that
can
enhance
geotechnical
engineering
applications
by
efficiently
behaviour.
Engineering With Computers,
Год журнала:
2025,
Номер
unknown
Опубликована: Фев. 11, 2025
Abstract
Real-time
assessment
of
unsaturated
soils
through
deflection
tests
is
challenging
due
to
the
complex
effects
water
and
air
in
soil
pores,
which
significantly
impact
test
outcomes
but
are
difficult
quantify,
especially
when
key
data
like
gravimetric
content
suction
incomplete
or
missing.
While
human
expertise
intuition
valuable
high-pressure
scenarios
ground
during
compaction,
they
prone
biases.
AI-driven
solutions
excel
at
processing
datasets
often
require
highly
specialised
inputs,
may
not
always
be
readily
available.
This
paper
aims
develop
a
robust
pragmatic
approach
decision-support
by
combining
insight
with
AI’s
computational
power
principles
from
mechanics.
outlines
limitations
current
practices
discusses
challenges
developing
reliable
using
on
soils.
To
address
these
challenges,
an
augmented
intelligence
framework
introduced
that
leverages
fuzzy
inputs
for
missing
information
incorporates
sophisticated
self-improving
mechanism
estimate
data,
based
insights
gained
calibration.
enhances
after
validation
recent
field
trial
particularly
uncertain
subsurface
conditions.
The
study
also
demonstrates
framework’s
resilience
qualitative
assessments,
maintaining
accuracy
across
range
assumptions
about
content.
Purpose
This
study
explores
the
integration
of
generative
artificial
intelligence
(AI)
into
numerical
analysis
workflows
in
geotechnical
engineering
to
address
challenges
generating
synthetic
datasets.
aims
create
a
framework
that
allows
practitioners
with
limited
programming
skills
automate
complex
simulations,
enabling
development
extensive
data
sets
for
AI
and
machine
learning
applications.
Design/methodology/approach
The
proposes
seven-step
methodology
using
finite
element
method
Python
auotmate
modelling.
Generative
AI,
specifically
ChatGPT,
is
used
as
virtual
assistant
guide
through
automation.
validated
pilot
predicting
excavation-induced
ground
displacement
Sydney’s
Hawkesbury
Sandstone.
Findings
Integrating
accelerates
generation
improves
quality
indicates
generated
datasets
closely
align
real-world
measurements,
confirming
robustness
reliability
proposed
framework.
Research
limitations/implications
study’s
accuracy
may
be
affected
by
assumptions
input
parameter
quality.
Future
research
should
explore
more
conditions,
such
3D
effects,
further
validate
enhance
methodology.
Practical
implications
provides
an
efficient
solution
generate
datassets
training,
reducing
reliance
on
experienced
programmers.
It
streamlines
enhances
data-driven
decision-making
engineering.
Originality/value
paper
introduces
novel
workflows,
offering
innovative
approach
generation.
serves
valuable
tool
advancing
applications
engineering,
particularly
those
experience.
