Solar
energy
plays
an
important
role
in
the
future
system.
However,
inherent
uncertainty
of
solar
brings
great
difficulties
to
grid
connection
and
short-term
planning
dispatching.
Deep
learning
method
makes
it
possible
predict
with
its
powerful
ability,
but
huge
training
process
parameter
adjustment
bring
actual
deployment.
Therefore,
this
paper
proposes
a
new
lightweight
multi-modal
model
for
irradiance
prediction
based
on
knowledge
distillation
strategy,
which
greatly
reduces
complexity
while
ensuring
acceptable
accuracy,
facilitating
Firstly,
teacher
inputs
Informer
framework
is
built
guide
student
model.
Then,
constructed
obtain
same
input
reduced
trainable
parameters.
The
optimal
settings
loss
function
ratio
are
studied.
Results
show
that
can
reduce
parameters
inference
time
by
97.7%
52.5%,
respectively.
normalized
root
mean
square
error
24.87%
compared
without
distillation,
verifying
effectiveness
proposed
method.
soft
uses
loss,
0.3,
best
results
structure
3
residual
blocks
LSTM
layers
proved
be
task.
Energy Reviews,
Journal Year:
2024,
Volume and Issue:
3(2), P. 100071 - 100071
Published: Feb. 9, 2024
Recent
studies
show
that
artificial
intelligence
(AI),
such
as
machine
learning
and
deep
learning,
models
can
be
adopted
have
advantages
in
fault
detection
diagnosis
for
building
energy
systems.
This
paper
aims
to
conduct
a
comprehensive
systematic
literature
review
on
(FDD)
methods
heating,
ventilation,
air
conditioning
(HVAC)
covers
the
period
from
2013
2023
identify
analyze
existing
research
this
field.
Our
work
concentrates
explicitly
synthesizing
AI-based
FDD
techniques,
particularly
summarizing
these
offering
classification.
First,
we
discuss
challenges
while
developing
HVAC
Next,
classify
into
three
categories:
those
based
traditional
hybrid
AI
models.
Additionally,
also
examine
physical
model-based
compare
them
with
methods.
The
analysis
concludes
FDD,
despite
its
higher
accuracy
reduced
reliance
expert
knowledge,
has
garnered
considerable
interest
compared
physics-based
However,
it
still
encounters
difficulties
dynamic
time-varying
environments
achieving
resolution.
Addressing
is
essential
facilitate
widespread
adoption
of
HVAC.
Applied Energy,
Journal Year:
2024,
Volume and Issue:
372, P. 123773 - 123773
Published: June 26, 2024
This
paper
proposes
a
novel
method
namely
WaveletKernelNet-Convolutional
Block
Attention
Module-BiLSTM
for
intelligent
fault
diagnosis
of
drilling
pumps.
Initially,
the
random
forest
is
applied
to
determine
target
signals
that
can
reflect
characteristics
Accordingly,
Module
Net
constructed
noise
reduction
and
feature
extraction
based
on
signals.
The
Convolutional
embedded
in
WaveletKernelNet-CBAM
adjusts
weight
enhances
representation
channel
spatial
dimension.
Finally,
Bidirectional
Long-Short
Term
Memory
concept
introduced
enhance
ability
model
process
time
series
data.
Upon
constructing
network,
Bayesian
optimization
algorithm
utilized
ascertain
fine-tune
ideal
hyperparameters,
thereby
ensuring
network
reaches
its
optimal
performance
level.
With
hybrid
deep
learning
presented,
an
accurate
real
five-cylinder
pump
carried
out
results
confirmed
applicability
reliability.
Two
sets
comparative
experiments
validated
superiority
proposed
method.
Additionally,
generalizability
verified
through
domain
adaptation
experiments.
contributes
safe
production
oil
gas
sector
by
providing
robust
industrial
equipment.
