Fractals,
Journal Year:
2024,
Volume and Issue:
32(03)
Published: Jan. 1, 2024
Metaheuristic
techniques
are
capable
of
representing
optimization
frames
with
their
specific
theories
as
well
objective
functions
owing
to
being
adjustable
and
effective
in
various
applications.
Through
the
deep
learning
models,
metaheuristic
algorithms
inspired
by
nature,
imitating
behavior
living
non-living
beings,
have
been
used
for
about
four
decades
solve
challenging,
complex,
chaotic
problems.
These
can
be
categorized
evolution-based,
swarm-based,
nature-based,
human-based,
hybrid,
or
chaos-based.
Chaos
theory,
a
useful
approach
understanding
neural
network
optimization,
has
basic
idea
viewing
dynamical
system
which
equation
schemes
utilized
from
space
pertaining
learnable
parameters,
namely
trajectory,
itself,
enables
description
evolution
training
behavior,
is
say
number
iterations
over
time.
The
examination
recent
studies
reveals
importance
chaos
sensitive
initial
conditions
randomness
properties
that
principally
emerging
on
complex
multimodal
landscape.
Chaotic
this
regard,
accelerates
speed
algorithm
while
also
enhancing
variety
movement
patterns.
significance
hybrid
developed
through
applications
different
domains
concerning
real-world
phenomena
well-known
benchmark
problems
literature
evident.
applied
networks
(DNNs),
branch
machine
learning.
In
respect,
features
DNNs
extensive
use
overviewed
explained.
Accordingly,
current
review
aims
at
providing
new
insights
into
deal
algorithms,
hybrid-based
metaheuristics,
chaos-based
metaheuristics
besides
presenting
information
development
essence
science
opportunities,
applicability-based
aspects
generation
well-informed
decisions.
Materials Testing,
Journal Year:
2022,
Volume and Issue:
64(5), P. 706 - 713
Published: May 1, 2022
Abstract
The
adaptability
of
metaheuristics
is
proliferating
rapidly
for
optimizing
engineering
designs
and
structures.
imperative
need
the
fuel-efficient
design
vehicles
with
lightweight
structures
also
a
soaring
demand
raised
by
different
industries.
This
research
contributes
to
both
areas
using
hybrid
Taguchi
salp
swarm
algorithm-Nelder–Mead
(HTSSA-NM)
manta
ray
foraging
optimization
(MRFO)
algorithm
optimize
structure
shape
automobile
brake
pedal.
results
HTSSA-NM
MRFO
are
compared
some
well-established
such
as
horse
herd
algorithm,
black
widow
squirrel
search
Harris
Hawks
verify
its
performance.
It
observed
that
robust
superior
in
terms
least
mass
Also,
realize
best
value
present
problem
rest
optimizer.
Geoscience Frontiers,
Journal Year:
2021,
Volume and Issue:
13(2), P. 101313 - 101313
Published: Oct. 9, 2021
Deep
excavation
during
the
construction
of
underground
systems
can
cause
movement
on
ground,
especially
in
soft
clay
layers.
At
high
levels,
excessive
ground
movements
lead
to
severe
damage
adjacent
structures.
In
this
study,
finite
element
analyses
(FEM)
and
hardening
small
strain
(HSS)
model
were
performed
investigate
deflection
diaphragm
wall
layer
induced
by
braced
excavations.
Different
geometric
mechanical
properties
investigated
study
behavior
clays.
Accordingly,
1090
hypothetical
cases
surveyed
simulated
based
HSS
FEM
evaluate
behavior.
The
results
then
used
develop
an
intelligent
for
predicting
using
functional
linked
neural
network
(FLNN)
with
different
expansions
activation
functions.
Although
FLNN
is
a
novel
approach
predict
deflection;
however,
order
improve
accuracy
deflection,
three
swarm-based
optimization
algorithms,
such
as
artificial
bee
colony
(ABC),
Harris's
hawk's
(HHO),
hunger
games
search
(HGS),
hybridized
generate
models,
namely
ABC-FLNN,
HHO-FLNN,
HGS-FLNN.
hybrid
models
compared
basic
MLP
models.
They
revealed
that
good
solution
application
functions
has
significant
effect
outcome
predictions
deflection.
It
remarkably
interesting
performance
was
better
than
mean
absolute
error
(MAE)
19.971,
root-mean-squared
(RMSE)
24.574,
determination
coefficient
(R2)
0.878.
Meanwhile,
only
obtained
MAE
20.321,
RMSE
27.091,
R2
0.851.
Furthermore,
also
indicated
proposed
i.e.,
HGS-FLNN,
yielded
more
superior
performances
those
terms
prediction
walls
range
11.877
12.239,
15.821
16.045,
0.949
0.951.
be
alternative
tool
simulate
deflections
under
conditions
degree
accuracy.
Energy Reports,
Journal Year:
2022,
Volume and Issue:
8, P. 11769 - 11798
Published: Sept. 22, 2022
The
management
of
renewable-powered
smart
grids
deals
with
nonlinear
optimization
problems
featuring
a
variety
linear
or
constraints,
discrete
continuous
variables,
involving
high
dimensionality
the
solution
space,
and
strict
time
requirements
to
identify
optimal
near-optimal
solution.
One
promising
approach
for
addressing
such
is
apply
bio-inspired
population-based
algorithms,
many
metaheuristics
emerging
lately.
In
this
paper,
we
have
identified
highest
impact
published
recently
reviewed
their
applications
in
energy
using
Preferred
Reporting
Items
Systematic
reviews
Meta-Analyses
(PRISMA)
methodology
Web
Science
Core
Collection
as
reference
database.
Four
main
grid
application
domains
been
analyzed:
(i)
prediction
models'
reduce
uncertainty
(ii)
resources
coordination
handle
stochastic
nature
renewables,
(iii)
demand
response
controllable
loads
flexibility
while
considering
consumers'
needs
constraints
(iv)
efficiency
costs.
results
showed
advantages
decentralized
low
computational
resource
overhead.
At
same
time,
several
issues
need
be
addressed
increase
adoption
scenarios:
lack
standard
testing
methodologies
benchmarks,
efficient
exploration
exploitation
search
guidelines
clear
links
type
problems,
etc.
