International Journal of Electrical and Electronics Engineering,
Год журнала:
2024,
Номер
11(7), С. 215 - 227
Опубликована: Июль 31, 2024
The
use
of
hybrid
Convolutional
Neural
Network-
Gated
Recurrent
Unit
(CNN-GRU)
models
for
solar
panel
Maximum
Power
Point
(MPP)
prediction
is
examined
in
this
work.
Improved
energy
forecasting
accuracy
essential
grid
integration
and
power-generating
optimization.
A
novel
CNN-GRU
architecture
that
captures
both
temporal
spatial
patterns
present
data
using
a
dataset
includes
temperature,
irradiance,
MPP
characteristics
utilized.
comparison
study
with
alternative
architectures
individual
GRU
CNN
models.
Model
performance
evaluated
by
evaluation
metrics
such
as
coefficient
determination
(R²),
Mean
Squared
Error
(MSE),
Absolute
(MAE).
Results
show
the
model
achieves
better
voltage
(Vmp)
current
(Imp)
at
than
architectures.
Furthermore,
residual
analysis
against
actual
comparisons
prove
efficacy
robustness
suggested
method.
practical
ramifications
renewable
management
stability
advance
methods.
Energies,
Год журнала:
2025,
Номер
18(2), С. 399 - 399
Опубликована: Янв. 17, 2025
To
address
the
challenges
of
issue
inaccurate
prediction
results
due
to
missing
data
in
PV
power
records,
a
photovoltaic
imputation
method
based
on
Wasserstein
Generative
Adversarial
Network
(WGAN)
and
Long
Short-Term
Memory
(LSTM)
network
is
proposed.
This
introduces
data-driven
GAN
framework
with
quasi-convex
characteristics
ensure
smoothness
imputed
existing
employs
gradient
penalty
mechanism
single-batch
multi-iteration
strategy
for
stable
training.
Finally,
through
frequency
domain
analysis,
t-Distributed
Stochastic
Neighbor
Embedding
(t-SNE)
metrics,
performance
validation
generated
data,
proposed
can
improve
continuity
reliability
tasks.
Scientific Reports,
Год журнала:
2025,
Номер
15(1)
Опубликована: Фев. 24, 2025
The
stochastic
and
variable
nature
of
power
generated
by
photovoltaic
(PV)
systems
can
impact
grid
stability.
Accurately
predicting
the
output
a
solar
PV
generation
system
is
crucial
for
addressing
this
challenge.
While
short-term
prediction
highly
accurate,
accuracy
medium-
to
long-term
predictions
will
face
great
challenges.
In
order
improve
medium
prediction,
unique
hybrid
deep
learning
model
named
interactive
feature
trend
transformer
(IFTformer)
has
been
designed.
Initially,
isolated
forest
(DIF)
local
anomaly
factor
(LOF)
are
used
construct
parallel
framework
that
serves
as
data
preprocessing
module,
removing
outliers
from
raw
data.
time
series
subsequently
decomposed
into
seasonal
components,
which
modelled
separately
independent
study.
Ultimately,
predicted
components
with
ProSparse
Self-attention
mechanism
based
on
information
interaction
fitted
multilayer
perceptron
(MLP)
prediction.
comprehensive
experimental
results
show
predictive
performance
IFTformer
superior
baseline
models,
normalised
root
mean
square
error
(NRMSE)
3.64%
absolute
(NMAE)
2.44%.
proposed
in
paper
an
effective
approach
mitigate
outliers,
enhance
extraction
ability,
accuracy,
generalizability
robustness
predictions,
providing
novel
perspective
methods
methods.
Energies,
Год журнала:
2025,
Номер
18(5), С. 1042 - 1042
Опубликована: Фев. 21, 2025
The
increasing
adoption
of
photovoltaic
(PV)
systems
has
introduced
challenges
for
grid
stability
due
to
the
intermittent
nature
PV
power
generation.
Accurate
forecasting
and
data
quality
are
critical
effective
integration
into
grids.
However,
records
often
contain
missing
system
downtime,
posing
difficulties
pattern
recognition
model
accuracy.
To
address
this,
we
propose
a
GAN-based
imputation
method
tailored
Unlike
traditional
GANs
used
in
image
generation,
our
ensures
smooth
transitions
with
existing
by
utilizing
data-guided
GAN
framework
quasi-convex
properties.
stabilize
training,
introduce
gradient
penalty
mechanism
single-batch
multi-iteration
strategy.
Our
contributions
include
analyzing
necessity
imputation,
designing
novel
conditional
network
validating
generated
using
frequency
domain
analysis,
t-NSE,
prediction
performance.
This
approach
significantly
enhances
continuity
reliability
tasks.
Romanian Journal of Information Science and Technology,
Год журнала:
2025,
Номер
28(1), С. 39 - 50
Опубликована: Март 14, 2025
The
research
in
the
field
of
renewable
energy
has
taken
centre
stage
study
reliable
and
effective
photovoltaic
(PV)
systems.
These
systems
are
essential
to
a
future
powered
by
energy,
where
solar
radiation
is
directly
converted
into
electrical
power.
However,
arrays
have
limited
conversion
efficiency.
Hence,
highly
accurate
forecasting
strategies
required
mitigate
impact
this
challenge.
This
focuses
on
proposing
serial
algorithms
that
combine
machine
learning
global
optimization
solve
stochastic
problems.
Gated
Recurrent
Unit
(GRU)
architecture,
Support
Vector
Machine
(SVM)
for
Regression
(SVR)
models
Differential
Evolution
algorithm
(DE)
used
developing
forecast
grid
power
generation
across
environmental
variations.
Initially,
four
GRU-SVR
will
be
trained
address
prediction
seasonal
evolution.
Afterwards,
hybrid
approach
GRU-SVR-DE
strategy
defined
integrate
models,
providing
robust
PV
generation.
In
end,
performances
predictions
analyzed
demonstrate
accuracy
long-term
forecasts.
Climate
variability
influences
renewable
electricity
supply
and
demand
hence
system
reliability.
Using
the
hidden
states
of
sea
surface
temperature
tropical
Pacific
Ocean
that
reflect
El
Niño-Southern
Oscillation
(ENSO)
dynamics
is
objectively
identified
by
a
nonhomogeneous
Markov
model,
we
provide
first
example
potential
predictability
monthly
wind
solar
energy
heating
cooling
for
1
to
6
months
ahead
Texas,
United
States,
region
has
high
penetration
susceptible
disruption
climate-driven
supply-demand
imbalances.
We
find
statistically
significant
oversupply
or
undersupply
anomalous
heating/cooling
depending
on
ENSO
state
calendar
month.
Implications
financial
securitization
application
forecasts
are
discussed.
