Sensors,
Год журнала:
2024,
Номер
24(24), С. 8111 - 8111
Опубликована: Дек. 19, 2024
Molding
sand
mixtures
in
the
foundry
industry
are
typically
composed
of
fresh
and
reclaimed
sands,
water,
additives
such
as
bentonite.
Optimizing
control
these
recycling
used
after
casting
requires
an
efficient
in-line
monitoring
method,
which
is
currently
unavailable.
This
study
explores
potential
AI-enhanced
electrical
impedance
spectroscopy
(EIS)
system
a
solution.
To
establish
fundamental
dataset,
we
characterized
various
containing
quartz
sand,
bentonite,
deionized
water
using
EIS
frequency
range
from
20
Hz
to
1
MHz
under
laboratory
conditions
also
measured
content
density
samples.
Principal
component
analysis
was
applied
data
extract
relevant
features
input
for
machine
learning
models.
These
features,
combined
with
density,
were
train
regression
models
based
on
fully
connected
neural
networks
estimate
bentonite
mixtures.
led
high
prediction
accuracy
(R2
=
0.94).
results
demonstrate
that
has
promising
bulk
material
industry,
paving
way
optimized
process
recycling.
The Journal of Physical Chemistry C,
Год журнала:
2025,
Номер
unknown
Опубликована: Янв. 5, 2025
Electrochemical
impedance
spectroscopy
(EIS)
is
an
important
analytical
technique
for
the
understanding
of
electrochemical
systems.
With
recent
advent
and
burgeoning
deployment
machine
learning
(ML)
in
EIS
analysis,
a
critical
yet
hitherto
unanswered
question
emerges:
what
appropriate
manner
to
preprocess
data
ML-based
analysis?
While
preprocessing
model's
input
known
be
successful
ML
model,
possess
multiple
classical
venues
representation,
moreover,
proper
normalization
protocol
comparative
studies
remains
elusive.
Here,
we
report
methodology
outcomes
that
evaluate
efficacy
methods
analysis.
Within
our
proof-of-concept
parameter
space,
plotting
training
data's
magnitude
(|Z|)
against
phase
angle
(φ)
while
individually
normalizing
each
curve
yields
highest
accuracy
robustness
correspondingly
established
residual
neural
network
(ResNet)
model.
Rationalized
by
additional
"importance"
analysis
data,
such
representation
method
extracts
information
hidden
features
more
effectively.
Nyquist
plot
widely
used
manual
different
seems
equally
plausible
Our
work
offers
future
researchers
decide
on
applications
electrochemistry
case-by-case
basis.
ACS Applied Materials & Interfaces,
Год журнала:
2025,
Номер
unknown
Опубликована: Фев. 28, 2025
The
continuous
global
effort
to
predict
material
properties
through
artificial
intelligence
has
predominantly
focused
on
utilizing
stoichiometry
or
structures
in
deep
learning
models.
This
study
aims
using
electrochemical
impedance
data,
along
with
frequency
and
time
parameters,
that
can
be
obtained
during
processing
stages.
target
material,
silica
aerogel,
is
widely
recognized
for
its
lightweight
structure
excellent
insulating
properties,
which
are
attributed
large
surface
area
pore
size.
However,
production
often
delayed
due
the
prolonged
aging
process.
Real-time
prediction
of
significantly
enhance
process
optimization
monitoring.
In
this
study,
we
developed
a
system
physical
specifically
diameter,
volume,
area.
integrates
3
×
array
Pd/Au
sensor,
exhibits
high
sensitivity
varying
pH
levels
aerogel
synthesis
capable
acquiring
data
set
(impedance,
frequency,
time)
real-time.
collected
then
processed
neural
network
algorithm.
Because
trained
stage,
it
enables
real-time
predictions
critical
thus
facilitating
final
performance
evaluation
demonstrated
an
optimal
alignment
between
true
predicted
values
mean
absolute
percentage
error
approximately
0.9%.
approach
holds
great
promise
improving
efficiency
effectiveness
by
providing
accurate
predictions.
Sensors,
Год журнала:
2025,
Номер
25(8), С. 2627 - 2627
Опубликована: Апрель 21, 2025
Mandarin
(Citrus
reticulata
L.)
is
consumed
worldwide.
Improper
storage
temperatures
cause
flavor
loss
and
shorten
shelf
lives,
reducing
marketability.
Mandarins’
quality
difficult
to
assess
visually,
as
they
show
no
apparent
changes
during
storage.
Therefore,
a
simple,
non-destructive
method
needed
their
freshness
affected
by
temperature.
This
work
utilized
non-invasive
bioimpedance
spectroscopy
(BIS)
on
mandarins
stored
at
different
temperatures.
Eight
machine
learning
(ML)
models
were
trained
with
data
classify
Also,
we
confirmed
whether
integrating
diameter
time-series
into
the
could
improve
ML
models’
accuracies
minimizing
sample
variations.
Additionally,
evaluated
effectiveness
of
equivalent
circuit
(EC)
parameters
derived
from
for
training.
Although
slightly
less
accurate
than
using
raw
data,
EC
can
efficiently
reduce
dimensionality.
Among
all
models,
SVM
model
in
integrated
achieved
highest
accuracy
0.92.
It
was
significant
improvement
compared
0.76
when
only
data.
Thus,
this
study
suggests
novel
temperature
mandarins.
approach
also
be
applied
other
fruits
utilizing
BIS.
Scientific Reports,
Год журнала:
2025,
Номер
15(1)
Опубликована: Июнь 3, 2025
Electrical
impedance
spectroscopy
(EIS)
is
a
powerful
tool
used
to
investigate
the
properties
of
materials
and
biological
tissues.
This
study
presents
one
first
applications
EIS
for
detection
classification
oral
potentially
malignant
disorders
(OPMDs)
cancer.
We
aimed
apply
in
conjunction
with
deep
learning
assist
clinical
diagnosis
OPMD
cancer
as
non-invasive
diagnostic
technology.
Currently,
relies
on
examination
histopathological
analysis
invasive
scalpel
tissue
biopsies,
which
stressful
patients,
time-consuming
clinicians
subject
interobserver
variation
diagnosis,
although
recent
advances
artificial
intelligence
may
circumvent
discrepancy.
Here
we
developed
novel
convolutional
neural
network
(CNN)-based
method
automatically
differentiate
normal,
tissues
using
measurements.
readings
were
initially
taken
from
untreated
or
glacial
acetic
acid-treated
porcine
mucosa
analyzed
via
CNN
determine
if
this
could
discriminate
between
normal
damaged
epithelium.
models
achieved
area
under
curve
(AUC)
values
0.92
±
0.03,
specificity
0.95
sensitivity
0.84,
showing
good
discrimination.
data
ventral
tongue
floor-of-the-mouth
collected
51
healthy
humans
11
patients
When
binary
(low
high
risk
malignancy)
was
applied,
best
model
an
AUC
0.91
0.1,
accuracy
0.05,
0.97
0.74.
These
results
demonstrate
considerable
potential
combination
adjunctive
