Applied Sciences,
Journal Year:
2020,
Volume and Issue:
10(19), P. 6653 - 6653
Published: Sept. 23, 2020
The
bio-inspired
algorithms
are
novel,
modern,
and
efficient
tools
for
the
design
of
electrical
machines.
However,
from
mathematical
point
view,
these
problems
belong
to
most
general
branch
non-linear
optimization
problems,
where
cannot
guarantee
that
a
global
minimum
is
found.
numerical
cost
accuracy
depend
on
initialization
their
internal
parameters,
which
may
themselves
be
subject
parameter
tuning
according
application.
In
practice,
even
more
challenging,
because
engineers
looking
robust
designs,
not
sensitive
tolerances
manufacturing
uncertainties.
These
criteria
further
increase
computationally
expensive
due
additional
evaluations
goal
function.
this
paper
give
an
overview
widely
used
techniques
in
machinery
summarize
challenges
open
applications
prospects
case
newly
emerging
technologies.
Computational Intelligence and Neuroscience,
Journal Year:
2021,
Volume and Issue:
2021(1)
Published: Jan. 1, 2021
Adaptive
Neuro‐Fuzzy
Inference
System
(ANFIS)
blends
advantages
of
both
Artificial
Neural
Networks
(ANNs)
and
Fuzzy
Logic
(FL)
in
a
single
framework.
It
provides
accelerated
learning
capacity
adaptive
interpretation
capabilities
to
model
complex
patterns
apprehends
nonlinear
relationships.
ANFIS
has
been
applied
practiced
various
domains
provided
solutions
commonly
recurring
problems
with
improved
time
space
complexity.
Standard
certain
limitations
such
as
high
computational
expense,
loss
interpretability
larger
inputs,
curse
dimensionality,
selection
appropriate
membership
functions.
This
paper
summarizes
that
the
standard
is
unsuitable
for
human
tasks
require
precise
handling
machines
systems.
The
state‐of‐the‐art
practice
research
questions
have
discussed,
which
primarily
focus
on
applicability
diversifying
field
engineering
sciences.
We
conclude
architecture
vastly
when
amalgamated
metaheuristic
techniques
further
moderated
nature‐inspired
algorithms
through
calibration
tuning
parameters.
significant
adapting
automating
currently
depend
discretion,
prominent
mechanical,
electrical,
geological
fields.
IEEE Access,
Journal Year:
2021,
Volume and Issue:
9, P. 56066 - 56092
Published: Jan. 1, 2021
This
study
integrates
a
tunicate
swarm
algorithm
(TSA)
with
local
escaping
operator
(LEO)
for
overcoming
the
weaknesses
of
original
TSA.
The
LEO
strategy
in
TSA–LEO
prevents
searching
deflation
TSA
and
improves
convergence
rate
search
efficiency
agents.
proposed
was
verified
on
CEC'2017
test
suite,
its
performance
compared
seven
metaheuristic
algorithms
(MAs).
comparisons
revealed
that
significantly
helps
by
improving
quality
solutions
accelerating
rate.
further
tested
real-world
problem,
namely,
segmentation
based
objective
functions
Otsu
Kapur.
A
set
well-known
evaluation
metrics
used
to
validate
TSA–LEO.
TSA-LEO
outperforms
other
MA
terms
fitness,
peak
signal-to-noise
ratio,
structural
similarity,
feature
findings.
PLoS ONE,
Journal Year:
2021,
Volume and Issue:
16(11), P. e0260725 - e0260725
Published: Nov. 30, 2021
Metaheuristic
optimization
algorithms
are
one
of
the
most
effective
methods
for
solving
complex
engineering
problems.
However,
performance
a
metaheuristic
algorithm
is
related
to
its
exploration
ability
and
exploitation
ability.
Therefore,
further
improve
African
vultures
(AVOA),
new
algorithm,
an
improved
based
on
tent
chaotic
mapping
time-varying
mechanism
(TAVOA),
proposed.
First,
map
introduced
population
initialization.
Second,
individual's
historical
optimal
position
recorded
applied
individual
location
updating.
Third,
designed
balance
To
verify
effectiveness
efficiency
TAVOA,
TAVOA
tested
23
basic
benchmark
functions,
28
CEC
2013
functions
3
common
real-world
design
problems,
compared
with
AVOA
5
other
state-of-the-art
algorithms.
According
results
Wilcoxon
rank-sum
test
5%,
among
has
significantly
better
than
that
13
functions.
Among
9
AVOA,
17
similar
AVOA.
Besides,
six
algorithms,
also
shows
good
in
IEEE Access,
Journal Year:
2021,
Volume and Issue:
9, P. 77416 - 77437
Published: Jan. 1, 2021
As
a
meta-heuristic
algorithm
that
simulates
the
intelligence
of
gray
wolves,
grey
wolf
optimizer
(GWO)
has
wide
range
applications
in
practical
problems.
kind
local
search,
chaotic
search
(CLS)
strong
ability
to
get
rid
optimum
due
its
integration
maps.
To
enhance
GWO,
CLS
is
always
incorporated
into
GWO
increase
population
diversity
and
accelerate
algorithm's
convergence.
However,
it
still
unclear
how
may
maps
should
be
used
embed
them
GWO.
address
these
challenging
issues,
this
paper
studies
both
single
multiple
GWOs.
Extensive
comparative
experiments
are
conducted
based
on
IEEE
Congress
Evolutionary
Computation
(CEC)
benchmark
test
suit.
The
results
show
GWOs
generally
perform
better
than
original
suggesting
effectiveness
such
hybridization.
Moreover,
remarkable
finding
work
piecewise
linear
map
(PWLCM)
Gaussian
have
most
potential
improve
performance
Additionally,
also
significantly
some
other
state-of-the-art
algorithms.
This
study
not
only
gives
more
insights
mechanism
makes
influence
but
finds
suitable
choice
for
it.
Energy Reports,
Journal Year:
2021,
Volume and Issue:
7, P. 6700 - 6717
Published: Oct. 19, 2021
Forecasting
of
solar
radiation
(Radn)
can
provide
an
insight
vision
for
the
amount
green
and
friendly
energy
sources.
Owing
to
non-linearity
non-stationarity
challenges
caused
by
meteorological
variables
in
forecasting
Radn,
a
variational
mode
decomposition
method
is
integrated
with
simulated
annealing
random
forest
(VMD-SA-RF)
resolving
this
problem.
Firstly,
input
parameters
are
separated
into
training
testing
phases
after
generating
one-day
ahead
significant
lags
at
(t
–
1).
Secondly,
set
factorize
multivariate
data
train
test
sets,
independently,
their
band-limited
signals.
Thirdly,
simulate
based
feature
selection
system
engaged
select
best
Finally,
using
pertinent
signals,
daily
Radn
forecasted
via
(RF)
model.
The
outcomes
benchmarked
other
comparative
models.
hybrid
fusion
VMD-SA-RF
model
tested
geographically
Australia,
generates
reliable
performance
forecast
Radn.
combining
features,
as
predictors
have
substantial
implications
renewable
sustainable
resource
management.
IEEE Access,
Journal Year:
2021,
Volume and Issue:
9, P. 14369 - 14386
Published: Jan. 1, 2021
Designing
Convolutional
Neural
Networks
from
scratch
is
a
time-consuming
process
that
requires
specialist
expertise.
While
automated
architecture
generation
algorithms
have
been
proposed,
the
underlying
search
strategies
generally
are
computationally
expensive.
The
existing
methods
also
do
not
explore
space
efficiently,
and
often
lead
to
sub-optimal
solutions.
In
this
research,
we
propose
novel
Particle
Swarm
Optimization
(PSO)-based
model
for
deep
address
above
challenges.
Our
proposed
solution
incorporates
three
new
components.
Firstly,
group-based
encoding
strategy
devised,
which
enforces
candidate
networks
always
follow
best
practices.
Specifically,
it
ensures
number
of
groups
can
be
adjusted
in
accordance
with
input
image
size.
By
restricting
groups,
adapt
frequency
pooling
operations
toward
As
such,
ascertains
position
maximum
result
valid
network
without
need
additional
complex
governing
rules.
Secondly,
velocity
updating
mechanism
creates
architectures
by
identifying
key
configuration
differences.
Thirdly,
using
weighted
strengths
devised.
Both
mechanisms
facilitate
PSO-based
intermediate
positions
particles'
trajectories,
allowing
better
trade-off
between
diversification
intensification
achieved.
We
employ
eight
well-known
data
sets,
including
Convex,
Rectangles,
MNIST
its
variants,
evaluation.
achieves
up
7.58%
improvement
accuracy
63%
reduction
computational
cost,
comparison
those
current
state-of-the-art
methods.
