Journal of Energy and Natural Resources,
Journal Year:
2024,
Volume and Issue:
13(4), P. 166 - 177
Published: Dec. 7, 2024
Lithium-ion
battery
is
one
of
the
core
components
electric
vehicles,
and
state
charge-state
health
estimation
results
it
key
to
restrict
safe
efficient
use
it,
which
then
affects
comprehensive
performance
vehicles.
However,
SOC
SOH
lithium-ion
batteries
have
a
coupling
relationship,
fast
slow
time-varying
characteristics
respectively,
with
inconsistent
time
scales.
Hence,
necessary
carry
out
SOC-SOH
collaborative
select
suitable
scale,
can
ensure
accuracy
robustness
without
consuming
too
much
calculation
cost.
This
article
proposed
an
innovative
hybrid
optimization
network
improve
ability
analysis
feature
extraction
capability
input
sequences
for
precise
estimation.
fully
combines
advantages
convolutional
neural
network,
bidirectional
long
short-term
memory,
attention
mechanism.
Additionally,
kepler
algorithm
applied
hyperparameter
first
according
our
knowledge,
also
estimated
accurately
more
ideal
results.
The
experimental
indicate
that
reach
under
different
working
conditions
ambient
temperatures.
mean
absolute
error
root
square
are
0.55%
0.72%
only
about
third
considering
SOH,
means
very
essential.
this
great
significance
development
smarter
management
system.
Processes,
Journal Year:
2025,
Volume and Issue:
13(4), P. 1229 - 1229
Published: April 18, 2025
The
plywood
industry
is
one
of
the
most
significant
sub-sectors
forestry
and
serves
as
a
cornerstone
sustainable
construction
within
bioeconomy
framework.
Plywood
panel
composed
multiple
layers
wood
sheets
bonded
together.
While
automation
process
monitoring
have
played
crucial
role
in
improving
efficiency,
data-driven
decision-making
remains
underutilized
industrial
sector.
Many
processes
continue
to
rely
heavily
on
expertise
operators
rather
than
data
analytics.
However,
advancements
storage
capabilities
availability
high-speed
computing
paved
way
for
algorithms
that
can
support
real-time
decision-making.
Due
biological
nature
numerous
variables
involved,
managing
manufacturing
operations
inherently
complex.
multitude
variables,
presence
non-linear
physical
phenomena
make
it
challenging
develop
accurate
robust
analytical
predictive
models.
As
result,
approaches—particularly
Artificial
Intelligence
(AI)—have
emerged
highly
promising
modeling
techniques.
Leveraging
exploring
application
AI
algorithms,
particularly
Machine
Learning
(ML),
predict
key
performance
indicators
(KPIs)
plants
represent
novel
expansive
field
study.
processing
evaluation
best
suited
remain
areas
research.
This
study
explores
supervised
(ML)
enhance
quality
control
veneers
production.
analysis
included
Random
Forest,
XGBoost,
K-Nearest
Neighbors
(KNN),
Support
Vector
(SVM),
Lasso,
Logistic
Regression.
An
initial
dataset
comprising
49
related
maceration,
peeling,
drying
was
refined
30
using
correlation
Lasso
variable
selection.
final
dataset,
encompassing
13,690
records,
categorized
into
9520
low-quality
labels
4170
high-quality
labels.
classification
revealed
differences;
Forest
reached
highest
accuracy
0.76,
closely
followed
by
XGBoost.
(KNN)
demonstrated
notable
precision,
while
(SVM)
exhibited
high
precision
but
low
recall.
Regression
showed
comparatively
lower
metrics.
These
results
highlight
importance
selecting
tailored
specific
characteristics
optimize
model
effectiveness.
highlights
critical
AI-driven
insights
operational
efficiency
product
veneer
manufacturing,
paving
enhanced
competitiveness.
