NanoWorld Journal,
Journal Year:
2023,
Volume and Issue:
9
Published: Nov. 3, 2023
This
research
presents
a
unique
method
for
estimating
nanofluid
viscosity
by
building
smart
generalized
model
on
top
of
deep
neural
network
(DNN).The
DNN
was
trained
using
the
nadam
optimization
approach
large
experimental
dataset
that
contained
Alumina
(Al
2
O
3
)
nanoparticles.Nonlinearities
may
be
automatically
learned
proposed
from
training
dataset.This
paper
details
innovative
aspects
this
investigation
and
how
they
combine
with
benefits
learning.To
author's
knowledge,
no
prior
attempt
made
to
predict
based
learning.The
comprehensive
model's
efficiency
demonstrates
it
outperforms
all
competing
models
while
also
avoiding
their
pitfalls.Additionally,
our
provides
remarkably
accurate
predictions
unseen
data
can
in
fraction
time
mandatory
conventional
data-driven
models.This
intelligent
has
been
subjected
sensitivity
study.With
coefficient
determination
0.9999,
DNN-based
is
best
at
predicting
nanofluids.
Journal of Materials Research and Technology,
Journal Year:
2023,
Volume and Issue:
24, P. 7570 - 7598
Published: May 1, 2023
This
study
presents
a
review
of
the
effect
nano-additives
in
improving
mechanical
properties
composites.
Nano-additives
added
to
composites,
also
termed
nanocomposites,
have
promising
applications
aerospace,
medical,
biomedical,
automotive,
and
military.
The
nanoparticles
alter
either
surface,
bulk,
or
both,
depending
upon
process,
dramatically
change
thermal
conductivity,
tensile
strength,
flexural
fatigue
impact
resistance,
vibration
buckling,
post-buckling,
surface
modification,
application
machine
learning
as
well
optimization
methods
nanocomposite
materials.
Such
transformations
composite
materials
are
extensively
studied
by
researchers
positive
implications
successfully
deployed
various
applications.
Interestingly,
recent
findings
revealed
that
weak
chemical
bonding
between
fiber
matrix
phase
is
main
reason
for
delamination,
however,
addition
nanoparticles,
chances
delamination
reduced
even
under
excessive
loading.
Graphene
multi-walled
carbon
nanotubes
(MWCNTs)
most
excessively
reported
nanomaterials
enhancing
behavior
energy
absorption
capacity,
decreasing
adverse
effects
due
porosity
within
structure.
Also,
techniques
showed
be
way
further
improve
while
reducing
total
cost
fabrication
process
predicting
providing
optimum
characteristics
with
acceptable
accuracy
compared
realistic
conditions.
Journal of Tribology,
Journal Year:
2024,
Volume and Issue:
146(5)
Published: Jan. 2, 2024
Abstract
In
the
current
work,
AZ91
hybrid
composites
are
fabricated
through
utilization
of
stir
casting
technique,
incorporating
aluminum
oxide
(Al2O3)
and
graphene
(Gr)
as
reinforcing
elements.
Wear
behavior
AZ91/Gr/Al2O3
was
examined
with
pin-on-disc
setup
under
dry
conditions.
this
study,
factors
such
reinforcement
percentage
(R),
load
(L),
velocity
(V),
sliding
distance
(D)
have
been
chosen
to
investigate
their
impact
on
wear-rate
(WR)
coefficient
friction
(COF).
This
study
utilizes
a
full
factorial
design
conduct
experiments.
The
experimental
data
critically
analyzed
examine
each
wear
parameter
(i.e.,
R,
L,
V,
D)
WR
COF
composites.
mechanisms
at
extreme
conditions
maximum
minimum
rates
also
investigated
by
utilizing
scanning
electron
microscope
(SEM)
images
specimen's
surface.
SEM
revealed
presence
delamination,
abrasion,
oxidation,
adhesion
surface
experiencing
wear.
Machine
learning
(ML)
models,
decision
tree
(DT),
random
forest
(RF),
gradient
boosting
regression
(GBR),
employed
create
robust
prediction
model
for
predicting
output
responses
based
input
variables.
trained
tested
95%
5%
points,
respectively.
It
noticed
that
among
all
GBR
exhibited
superior
performance
in
WR,
mean
square
error
(MSE)
=
0.0398,
root-mean-square
(RMSE)
0.1996,
absolute
(MAE)
0.1673,
R2
98.89,
surpassing
accuracy
other
models.
