Electronics,
Journal Year:
2023,
Volume and Issue:
13(1), P. 113 - 113
Published: Dec. 27, 2023
In
order
to
enhance
the
predictive
accuracy
and
control
capabilities
pertaining
low-
medium-frequency
road
noise
in
automotive
contexts,
this
study
introduces
a
methodology
for
Structural-borne
Road
Noise
(SRN)
prediction
optimization.
This
approach
relies
on
multi-level
target
decomposition
hybrid
model
combining
Convolutional
Neural
Network
(CNN)
Support
Vector
Regression
(SVR).
Initially,
analysis
method
is
proposed,
grounded
hierarchical
of
vehicle
along
chassis
parts,
delineated
layer
by
layer,
accordance
with
vibration
transmission
path.
Subsequently,
CNN–SVR
model,
predicated
framework,
proposed.
Notably,
exhibits
superior
exceeding
0.97,
surpassing
both
traditional
CNN
SVR
models.
Finally,
are
deployed
sensitivity
parameters
relation
noise,
as
well
optimization
SRN
vehicles.
The
outcomes
underscore
high
such
dynamic
stiffness
rear
axle
bushing
large
front
swing
arm
influencing
SRN.
results,
facilitated
align
closely
measured
outcomes,
displaying
negligible
relative
error
0.82%.
Furthermore,
results
indicate
noteworthy
enhancement
4.07%
driver’s
right-ear
Sound
Pressure
Level
(SPL)
following
proposed
improvements
compared
original
state.
TURKISH JOURNAL OF ELECTRICAL ENGINEERING & COMPUTER SCIENCES,
Journal Year:
2024,
Volume and Issue:
32(1), P. 68 - 92
Published: Feb. 7, 2024
This
study
presents
a
fast
hyperparameter
optimization
algorithm
based
on
the
benefits
and
shortcomings
of
standard
grid
search
(GS)
for
support
vector
regression
(SVR).
presented
GS-inspired
algorithm,
called
(FGS),
was
tested
benchmark
datasets,
impact
FGS
prediction
accuracy
primarily
compared
with
GS
which
it
is
based.
To
validate
efficacy
proposed
conduct
comprehensive
comparison,
two
additional
techniques,
namely
particle
swarm
Bayesian
optimization,
were
also
employed
in
development
models
given
datasets.
The
evaluation
models'
predictive
performance
conducted
by
assessing
root
mean
square
error,
absolute
percentage
error.
In
addition
to
these
metrics,
number
evaluated
submodels
time
required
used
as
determinative
measures
models.
Experimental
results
proved
that
FGS-optimized
SVR
yield
precise
performance,
supporting
reliability,
validity,
applicability
algorithm.
As
result,
can
be
offered
faster
alternative
optimizing
hyperparameters
terms
execution
time.
International Journal of Low-Carbon Technologies,
Journal Year:
2025,
Volume and Issue:
20, P. 1024 - 1035
Published: Jan. 1, 2025
Abstract
The
objective
of
this
study
is
to
oversee
the
operation
several
converter-based
distributed
generations
in
order
assure
efficient
power
distribution
inside
an
island-microgrid
(MG).
commences
by
introducing
a
MG
model
that
integrates
virtual
impedances
with
phase-locked
loop.
It
subsequently
presents
unique
method
for
analyzing
small-signal
stability
islanded
MGs.
A
impedance
setting
strategy
created
using
gray
wolf
optimization
algorithm.
was
found
voltage
and
frequency
stay
within
acceptable
boundaries.
index
much
increased
reactive
imbalances
were
eliminated.
