International Journal for Research in Applied Science and Engineering Technology,
Journal Year:
2022,
Volume and Issue:
10(12), P. 2054 - 2060
Published: Dec. 29, 2022
Abstract:
Agriculture
is
one
in
all
the
formost
significant
roles
within
growth
and
development
of
our
nation
economy.
The
identification
diseases
that
key
forestall
losses
yield
quantity
agriculture
product.
Diseases
detection
on
plant
incredibly
critical
for
sustainable
agriculture.
It’s
challenging
to
watch
manually
especially
people
who
are
new
farming.
It
requires
excessive
time
interval.
Therefore
a
correct
prediction
disease
will
reduce
utilization
fertilizer
field,
which
helps
from
soil
impurities.
In
this
present
paper
we
have
explained
how
train
model
with
normal
dataset
augmented
achieved
accuracy
greater
than
95%
Energy Storage and Saving,
Journal Year:
2024,
Volume and Issue:
3(2), P. 96 - 105
Published: Feb. 23, 2024
Fault
detection
and
diagnosis
(FDD)
of
heating,
ventilating,
air
conditioning
(HVAC)
systems
can
help
to
improve
the
energy
saving
in
building
systems.
However,
most
data-driven
trained
FDD
models
have
limited
generalizability
only
be
applied
specific
The
diversity
HVAC
high
cost
data
acquisition
present
challenges
for
practical
application
FDD.
Transfer
learning
technology
employed
mitigate
this
problem
by
training
a
model
on
with
sufficient
then
transfer
it
other
data.
In
study,
novel
approach
is
proposed.
First,
transformer
modified
incorporate
one
encoder
two
decoders
connected,
enabling
outputs.
This
accommodates
absent
features
target
domain
serves
as
robust
foundation
learning.
It
has
effective
performance
complex
achieves
an
accuracy
91.38%
system
16
faults
multiple
fault
severity
levels.
Second,
adapter-based
parameter-efficient
method,
facilitating
simply
inserting
small
adapter
modules,
investigated
strategy.
Results
demonstrate
that
satisfactory
similar
full
fine-tunning
fewer
trainable
parameters.
works
well
amount
domain.
Furthermore,
findings
highlight
significance
adapters
positioned
near
bottom
top
layers,
emphasizing
their
critical
role
successful
IEEE Transactions on Instrumentation and Measurement,
Journal Year:
2023,
Volume and Issue:
72, P. 1 - 13
Published: Jan. 1, 2023
The
extreme
unbalance
of
training
samples
among
different
working
conditions
caused
by
complex
and
variable
external
environment
makes
the
fault
diagnosis
chiller
based
on
domain
adaptation
(DA)
poor
performance.
Although
recently
emerging
methods
generalization
(DG)
can
learn
domain-invariant
knowledge
from
multiple
source
domains
generalize
to
unseen
target
domains,
these
still
rely
similar
data
rarely
consider
how
enhance
ability
distinguish
joint
distribution
features
extracted
samples.
To
address
problems,
a
generative
domain-generalized
framework
for
chillers
diagnosis,
namely,
vision
transformer
adversarial
(VIT-GADG),
is
proposed.
In
VIT-GADG,
novel
VIT
generation
network
(VIT-DGN)
firstly
designed
reduce
DG's
dependence
multi-source
improving
diversity
Then,
new
called
conditional
(VIT-CADGN)
extract
latent
that
be
generalized
domains.
Specifically,
module
effectively
global
statistical
feature
input
samples,
which
conducive
identification
distribution.
Simultaneously,
collaborative
discrimination
strategy
introduced
improve
while
simultaneously
aligning
its
addition,
personalized
adaptive
weight
proposed
performance
VIT-CADGN.
Finally,
comprehensive
case
study
shows
VIT-GADG
has
satisfactory
invariant
features,
improves
accuracy
in
domain.
