International Journal of Parallel Emergent and Distributed Systems,
Год журнала:
2024,
Номер
unknown, С. 1 - 17
Опубликована: Дек. 12, 2024
We
consider
discounted
Markov
Decision
Processes
(MDPs)
with
large
state
spaces,
aiming
to
reduce
computational
complexity
and
execution
time.
Existing
hierarchical
techniques
often
decompose
the
space
into
strongly
connected
components
(SCCs)
across
levels.
However,
they
overlook
importance
of
SCC
size
at
each
level,
significantly
affecting
efficiency.
propose
Parallel
Hierarchical
Value
Iteration
(PHVI)
algorithm,
which
efficiently
handles
MDPs
by
considering
dimensionality.
This
approach
optimizes
multithreading
distribution,
leading
improved
performance
reduced
times.
Experimental
results
demonstrate
PHVI
algorithm's
effectiveness
superiority
over
traditional
methods
in
solving
complex
MDPs.
Energy and AI,
Год журнала:
2024,
Номер
17, С. 100378 - 100378
Опубликована: Май 22, 2024
The
increasing
drive
towards
eco-friendly
environment
motivates
the
generation
of
energy
from
renewable
sources
(RESs).
rising
share
RESs
in
power
poses
potential
challenges,
including
uncertainties
output,
frequency
fluctuations,
and
insufficient
voltage
regulation
capabilities.
As
a
solution
to
these
storage
systems
(ESSs)
play
crucial
role
storing
releasing
as
needed.
Battery
(BESSs)
provide
significant
maximize
efficiency
distribution
network
benefits
different
stakeholders.
This
can
be
achieved
through
optimizing
placement,
sizing,
charge/discharge
scheduling,
control,
all
which
contribute
enhancing
overall
performance
network.
In
this
paper,
we
comprehensive
overview
BESS
operation,
optimization,
modeling
applications,
how
mathematical
artificial
intelligence
(AI)-based
optimization
techniques
charging
discharging
scheduling.
We
also
discuss
some
future
opportunities
challenges
AI
BESSs,
emerging
technologies,
such
internet
things,
AI,
big
data
impact
development
BESSs.
Journal of Physics Conference Series,
Год журнала:
2025,
Номер
2936(1), С. 012018 - 012018
Опубликована: Янв. 1, 2025
Abstract
This
study
considers
a
cross-building
energy
storage
system
in
which
the
objective
function
of
each
step
is
piecewise
linear
decision
variables
and
state
variables.
Therefore,
can
be
modeled
as
programming
then
transformed
into
mixed
integer
(MILP)
problem.
However,
multi-stage
stochastic
problem
we
utilize
approximate
dynamic
(ADP)
to
tackle
computational
issues,
need
solve
multiple
times.
To
further
decrease
cost,
propose
several
algorithms
determine
variable
splitting,
degrades
We
use
techniques
design
experiments
verify
our
conclusion.
Numerical
show
that
algorithm
greatly
reduces
time
needed
under
condition
minimal
loss
accuracy.
The
simulation
experiment
Python
environment
proves
management
based
on
routers
control
centers
better
than
building
working
alone
maximize
benefits.
