Scientific Reports,
Год журнала:
2024,
Номер
14(1)
Опубликована: Ноя. 2, 2024
One
of
the
important
applications
real-time
crash
prediction
analysis
lies
in
field
proactive
traffic
management,
where
instantaneous
risk
evaluation
and
dynamic
decision-making
are
prerequisites.
This
research
proposes
an
integrated
advanced
framework
for
Variable
Speed
Limits
(VSL)
Hard
Shoulder
Running
(HSR)
implemented
freeways
considering
their
operational
periods.
Statistical
methods
utilized
to
identify
significant
contributing
factors
(related
traffic,
roadway
geometry,
weather
conditions)
explain
relationships
with
crashes.
Time-Embedded
Transformer
models
proposed
predict
likelihood
events.
The
sensitivity
false
alarm
rate
AM
peak
model
found
be
0.76
0.27,
respectively,
whereas
values
0.78
0.24,
respectively
PM
model.
Additionally,
results
indicate
substantial
improvements
performance
(i.e.,
increment
by
7.04%
8.33%
models,
respectively)
after
incorporating
general
safety
condition
a
freeway
segment
as
input
feature
while
estimating
models.
Practitioners
policymakers
could
apply
this
method
obtain
more
accurate
estimation,
precursors,
dynamically
update
algorithms,
enhance
aspects
operating
management
strategies
on
freeways.
Multimodal Transportation,
Год журнала:
2024,
Номер
3(3), С. 100139 - 100139
Опубликована: Май 31, 2024
The
main
aim
of
this
study
was
to
propose
a
comprehensive
risk
indicator
identify
the
potential
driving
caused
by
changing
environment
at
tunnel
entrance.
Driving
decisions
are
affected
many
external
factors,
especially
entrance
tunnels.
However,
indicators
mostly
considering
vehicle
movement
status
currently.
In
study,
safe
field
model
including
obstacle
field,
and
lighting
is
constructed
evaluate
influence
roads,
drivers,
vehicles,
change
conditions
on
risk.
Furthermore,
distribution
its
temporal
rate,
(CDRI)
established
magnitude
Finally,
comparison
between
CDRI
other
two
classic
indicates
that
proposed
in
paper
has
better
performance
safety
assessment
It
expected
finding
could
be
valuable
developing
control
measures
for
in-tunnel
declining.
Multidisciplinary Science Journal,
Год журнала:
2025,
Номер
7(8), С. 2025392 - 2025392
Опубликована: Фев. 12, 2025
This
research
examines
the
application
of
Sichel
(SI)
generalized
additive
models
for
location,
scale,
and
shape
(GAMLSS)
in
addressing
challenge
modeling
highly
dispersed
crash
data.
The
distribution,
which
combines
Poisson
distribution
with
inverse
Gaussian
is
particularly
suited
data
significant
dispersion,
where
traditional
often
prove
inadequate.
primary
objective
this
study
was
to
assess
performance
GAMLSS
comparison
widely-used
Negative
Binomial
(NB)
linear
model
(GLM).
To
achieve
this,
developed
evaluated
NB,
zero-inflated
NB
(ZINB),
(PIG)
SI
using
two
datasets
characterized
by
high
dispersion.
first
dataset
comprises
from
338
rural
interstate
road
sections
Indiana,
while
second
includes
undivided
4-lane
segments
Texas.
were
compared
a
range
goodness-of-fit
criteria.
Results
demonstrate
that
offers
better
fit
than
ZINB
PIG
GLMs
analyzed.
Moreover,
provides
advantageous
statistical
properties,
such
as
long-term
mean
never
zero
variance
function
driven
two-parameter
dispersion
mechanism.
Therefore,
presents
itself
promising
alternative
analysis
datasets.
PLoS ONE,
Год журнала:
2025,
Номер
20(4), С. e0320135 - e0320135
Опубликована: Апрель 9, 2025
The
use
of
statistical
modeling
methods
to
quantify
crash
causation
on
mountain
freeways
is
limited
by
data
availability
and
technical
challenges
posed
excess
zeros
heterogeneity,
resulting
in
a
lack
significant
targeting
proactive
prevention
measures
freeways.
We
collected
multidimensional
crash-related
information
mountainous
China,
including
road
design
characteristics,
traffic
conditions,
pavement
performance,
weather
conditions.
To
overcome
the
heterogeneity
techniques,
we
innovatively
developed
two
new
models:
Random
Parameter
Negative
Binomial
Lindley
(RPNB-L)
model
Generalized
Exponential
(RPNB-GE)
model.
goodness-of-fit
indicates
that
RPNB-L
RPNB-GE
models
stand
out
among
six
competing
models,
suggesting
GE
distributions
are
conducive
portraying
multi-zero
attributes,
while
regression
coefficients
randomization
treatment
provides
deeper
portrayal
heterogeneous
effects.
Moreover,
analysis
reveals
considerable
number
causes
for
frequency
China.
These
include
several
interesting
results,
such
as
special
segments
like
tunnels
interchanges,
Pavement
Damage
Condition
Index
(PCI)
stormy
rainfall
(TR),
which
have
not
been
extensively
studied
previous
research.
research
results
provided
important
reference
values
selection
active
safety
countermeasures
