Advanced Energy Materials,
Год журнала:
2023,
Номер
13(39)
Опубликована: Авг. 18, 2023
Abstract
Precise
prediction
of
lithium‐ion
cell
level
aging
under
various
operating
conditions
is
an
imperative
but
challenging
part
ensuring
the
quality
performance
emerging
applications
such
as
electric
vehicles
and
stationary
energy
storage
systems.
Accurate
real‐time
battery‐aging
models,
which
require
exact
understanding
degradation
mechanisms
battery
components
materials,
could
in
turn
provide
new
insights
for
materials
basic
research.
Furthermore,
primary
barrier
to
meaningful
artificial
intelligence/machine
learning
accelerating
period
exploitation
accurate
mechanistic
descriptors.
This
review
comprehensively
summarizes
evolution
deterioration
at
material
different
environments
usage
scenarios,
including
intricate
relationships
between
mechanisms,
modes,
external
influences,
are
cornerstones
modeling
simulation
machine
techniques.
Recent
advances
electrochemical
models
coupled
with
internal
well
identification
tracking
parameters
shown,
particular
emphasis
on
electrode
balance
anticipated
trend
learning‐assisted
reliable
remaining
useful
life
prediction.
will
continue
play
essential
role
advanced
smart
research
management,
enhancing
its
while
shortening
experimental
sequences.
Journal of Energy Chemistry,
Год журнала:
2023,
Номер
82, С. 103 - 121
Опубликована: Апрель 1, 2023
Lithium-ion
batteries
are
the
most
widely
used
energy
storage
devices,
for
which
accurate
prediction
of
remaining
useful
life
(RUL)
is
crucial
to
their
reliable
operation
and
accident
prevention.
This
work
thoroughly
investigates
developmental
trend
RUL
with
machine
learning
(ML)
algorithms
based
on
objective
screening
statistics
related
papers
over
past
decade
analyze
research
core
find
future
improvement
directions.
The
possibility
extending
lithium-ion
battery
lifetime
using
results
also
explored
in
this
paper.
ten
ML
first
identified
380
relevant
papers.
Then
general
flow
an
in-depth
introduction
four
signal
pre-processing
techniques
presented.
common
given
time
a
uniform
format
chronological
order.
compared
from
aspects
accuracy
characteristics
comprehensively,
novel
directions
or
opportunities
including
early
prediction,
local
regeneration
modeling,
physical
information
fusion,
generalized
transfer
learning,
hardware
implementation
further
outlooked.
Finally,
methods
extension
summarized,
feasibility
as
indicator
Battery
can
be
extended
by
optimizing
charging
profile
serval
times
according
online
future.
paper
aims
give
inspiration
strategy.
Progress in Energy and Combustion Science,
Год журнала:
2023,
Номер
100, С. 101120 - 101120
Опубликована: Сен. 22, 2023
Transportation
electrification
is
a
promising
solution
to
meet
the
ever-rising
energy
demand
and
realize
sustainable
development.
Lithium-ion
batteries,
being
most
predominant
storage
devices,
directly
affect
safety,
comfort,
driving
range,
reliability
of
many
electric
mobilities.
Nevertheless,
thermal-related
issues
batteries
such
as
potential
thermal
runaway,
performance
degradation
at
low
temperatures,
accelerated
aging
still
hinder
wider
adoption
To
ensure
safe,
efficient,
reliable
operations
lithium-ion
monitoring
their
states
critical
safety
protection,
optimization,
well
prognostics,
health
management.
Given
insufficient
onboard
temperature
sensors
inability
measure
battery
internal
temperature,
accurate
timely
estimation
particular
importance
state
monitoring.
Toward
this
end,
paper
provides
comprehensive
review
techniques
in
systems
regarding
mechanism,
framework,
representative
studies.
The
metrics
used
characterize
are
discussed
detail
first
considering
spatiotemporal
attributes
strengths
weaknesses
applying
management
also
analyzed.
Afterward,
various
methods,
including
impedance/resistance-based,
model-based,
data-driven
estimations,
elucidated,
analyzed,
compared
terms
strengths,
limitations,
improvements.
Finally,
key
challenges
real
applications
identified,
future
opportunities
for
removing
these
barriers
presented
discussed.
Nature Communications,
Год журнала:
2024,
Номер
15(1)
Опубликована: Май 21, 2024
Abstract
Accurate
state-of-health
(SOH)
estimation
is
critical
for
reliable
and
safe
operation
of
lithium-ion
batteries.
However,
stable
battery
SOH
remains
challenging
due
to
diverse
types
operating
conditions.
In
this
paper,
we
propose
a
physics-informed
neural
network
(PINN)
accurate
SOH.
Specifically,
model
the
attributes
that
affect
degradation
from
perspective
empirical
state
space
equations,
utilize
networks
capture
dynamics.
A
general
feature
extraction
method
designed
extract
statistical
features
short
period
data
before
fully
charged,
enabling
our
applicable
different
charge/discharge
protocols.
Additionally,
generate
comprehensive
dataset
consisting
55
lithium-nickel-cobalt-manganese-oxide
(NCM)
Combined
with
three
other
datasets
manufacturers,
use
total
387
batteries
310,705
samples
validate
method.
The
mean
absolute
percentage
error
(MAPE)
0.87%.
Our
proposed
PINN
has
demonstrated
remarkable
performance
in
regular
experiments,
small
sample
transfer
experiments
when
compared
alternative
networks.
This
study
highlights
promise
machine
learning
modeling
estimation.
Reliability Engineering & System Safety,
Год журнала:
2023,
Номер
241, С. 109603 - 109603
Опубликована: Авг. 29, 2023
Predictive
health
assessment
is
of
vital
importance
for
smarter
battery
management
to
ensure
optimal
and
safe
operations
thus
make
the
most
use
life.
This
paper
proposes
a
general
framework
aging
prognostics
in
order
provide
predictions
knee,
lifetime,
state
degradation,
rate
variations,
as
well
health.
Early
information
used
predict
knee
slope
other
life-related
via
deep
multi-task
learning,
where
convolutional-long-short-term
memory-bayesian
neural
network
proposed.
The
structure
also
online
degradation
detection
accelerating
aging.
two
probabilistic
predicted
boundaries
identify
regions
assessment.
To
avoid
wrong
premature
alarms,
empirical
model
data
preprocessing
together
with
learning.
A
cloud-edge
considered
fine-tuning
adopted
performance
improvement
during
cycling.
proposed
flexible
adjustment
different
practical
requirements
can
be
extrapolated
batteries
aged
under
conditions.
results
indicate
that
early
are
improved
using
method
compared
multiple
single
feature-based
benchmarks,
integration
algorithm
improved.
sequence
prediction
reliable
lengths
root
mean
square
errors
less
than
1.41%,
guide
predictive
management.
