2021 China Automation Congress (CAC),
Journal Year:
2023,
Volume and Issue:
unknown, P. 1177 - 1182
Published: Nov. 17, 2023
Electro-Hydrostatic
Actuator
(EHA)
is
widely
used
in
the
aerospace
industry
due
to
its
ability
of
high-precision
control
and
high
load-carrying
capacity.
During
their
extensive
service
life,
aviation
EHA's
hydraulic
systems
inevitably
experience
failures,
leading
a
degradation
performance
potential
safety
incidents.
Moreover,
diagnosis
challenged
by
intricate
structural,
elusive
failure
mechanisms,
difficulties
obtaining
representative
fault
samples.
To
address
this,
novel
approach
based
on
Muti-source
fusion
hypergraph
convolutional
neural
networks
(MF-HGCN)
proposed.
In
this
study,
employed
integrate
multiple
features
extracted
from
system.
By
calculating
Euclidean
distance
between
various
signal
sources,
structure
constructed.
Each
node
corresponds
specific
channel.
Subsequently,
data
fed
into
designed
network,
enabling
classification
entire
graph.
Finally,
trained
network
model
utilized
for
intelligent
The
experimental
results
EHA
system
test
rig
Nanjing
University
Science
Technology
demonstrate
that
proposed
method
has
better
diagnostic
under
limited
IEEE Access,
Journal Year:
2023,
Volume and Issue:
11, P. 55370 - 55379
Published: Jan. 1, 2023
Internet
of
Things
(IoT)
devices
are
becoming
increasingly
ubiquitous
in
daily
life.
They
utilized
various
sectors
like
healthcare,
manufacturing,
and
transportation.
The
main
challenges
related
to
IoT
the
potential
for
faults
occur
their
reliability.
In
classical
fault
detection,
client
device
must
upload
raw
information
central
server
training
model,
which
can
reveal
sensitive
business
information.
Blockchain
(BC)
technology
a
detection
algorithm
applied
overcome
these
challenges.
Generally,
fusion
BC
algorithms
give
secure
more
reliable
ecosystem.
Therefore,
this
study
develops
new
Assisted
Data
Edge
Verification
with
Consensus
Algorithm
Machine
Learning
(BDEV-CAML)
technique
Fault
Detection
purposes.
presented
BDEV-CAML
integrates
benefits
blockchain,
IoT,
ML
models
enhance
network's
trustworthiness,
efficacy,
security.
technology,
that
possess
significant
level
decentralized
decision-making
capability
attain
consensus
on
efficiency
intrablock
transactions.
For
network,
deep
directional
gated
recurrent
unit
(DBiGRU)
model
is
used.
Finally,
African
vulture
optimization
(AVOA)
optimal
hyperparameter
tuning
DBiGRU
helps
improving
rate.
A
detailed
set
experiments
were
carried
out
highlight
enhanced
performance
algorithm.
comprehensive
experimental
results
stated
improved
over
other
existing
maximum
accuracy
99.6%.
IEEE Sensors Journal,
Journal Year:
2024,
Volume and Issue:
24(14), P. 22601 - 22609
Published: May 29, 2024
This
paper
investigates
the
fault
detection
and
fault-tolerant
control
(FTC)
problems
for
discrete-time
multi-agent
systems
(MASs)
with
sensor
faults.
Firstly,
dynamic
linearization
method
is
introduced
to
describe
unknown
MASs
Afterward,
a
decentralized
based
on
data-driven
observers
proposed.
And
estimator
RBF
neural
networks
estimating
multiple
faults
designed
Then,
basis
of
estimator,
distributed
model-free
sliding
mode
FTC
strategy
provided
ensure
stability
considered
when
suffering
from
certain
Finally,
simulated
example
used
illustrate
efficiency
proposed
method.
IEEE Transactions on Instrumentation and Measurement,
Journal Year:
2024,
Volume and Issue:
73, P. 1 - 12
Published: Jan. 1, 2024
This
article
proposes
a
hybrid
fault
diagnosis
method
based
on
perturbation
estimation
convolution
network
(PECN)
of
multiple
open-circuit
switch
faults
for
cascaded
H-bridge
(CHB)
multilevel
converter.
The
proposed
observer
as
the
model-based
can
extract
characteristics
output
current
and
voltage.
deviations
measured
states
observed
states,
which
are
introduced
estimation,
well
capacitor
voltages
form
input
data
neural
(CNN).
multilayer
is
applied
to
deeply
signatures
determine
type
location
faulty
switches
rather
than
manually
setting
empirical
thresholds
in
traditional
methods.
PECN
improves
accuracy
adaptability
through
combining
advantages
both
data-driven
method,
detect
locate
under
different
operation
conditions.
Simulations
results
confirm
effectiveness
robustness
further
demonstrated
hardware-in-the-loop
(HIL)
testing
platform.
IEEE Transactions on Instrumentation and Measurement,
Journal Year:
2023,
Volume and Issue:
72, P. 1 - 14
Published: Jan. 1, 2023
The
dynamic
monitoring
of
the
temperature
distribution
power
equipment
is
crucial
for
safe
operation.
method
based
on
digital
twin
technology
has
received
extensive
attention
due
to
its
ability
provide
more
timely
and
comprehensive
analysis.
However,
existing
methods
only
use
physics-driven
or
data-driven
separately,
which
cannot
simultaneously
meet
application
requirements
such
as
low
cost,
high
efficiency,
effective
results
sufficient
training
data.
Therefore,
a
new
thermal
knowledge
conditional
generative
adversarial
network
(CGAN)
are
proposed
in
this
paper,
with
voltage
cable
joint
taken
an
example.
First,
steady-state
numerical
model
field
established.
image
set
under
different
operation
conditions
obtained
through
simulation.
CGAN
applied
learn
data
laws
between
images,
constructs
mapping
relationship
them.
Then,
used
reduce
feature
dimensions
extract
input
features
from
real-time
measurement
These
drive
generator
generate
images
dynamically.
By
comparing
generation
experimental
results,
can
achieve
joints,
advantages
regards
time
computational
accuracy
generalization
ability.
It
provides
insights
into
equipment.
