Prediction of local scour depth in bridge piers: Physical information and machine learning based modeling
Advances in Structural Engineering,
Год журнала:
2025,
Номер
unknown
Опубликована: Март 19, 2025
Local
scour
is
one
of
the
main
reasons
for
bridge
collapse.
To
solve
difficult
problem
detecting
local
depth
underwater
pier
structures,
this
paper
explores
an
optimal
method
predicting
structures
based
on
various
ensemble
learning
methods.
Firstly,
collects
487
sets
data
samples
containing
nine
input
parameters
with
corresponding
depths
from
open-source
database
in
practical
project.
Secondly,
employs
five
algorithms
commonly
used
learning,
that
is,
Random
Forest
(RF),
Gradient
Boosted
Decision
Tree
(GBDT),
Extreme
Boosting
(XGBoost),
Adaptive
(AdaBoost),
and
Light
Machine
(LightGBM),
to
build
a
prediction
model
depth.
In
addition,
Bayesian
hyperparameter
optimization
applied
search
best
combination
model.
Then,
eight
evaluation
indices,
including
Mean
Absolute
Error
(MAE),
Bias
(MBE),
Percentage
(MAPE),
Root
Square
(RMSE),
coefficient
determination
(R
2
),
Nash-Sutcliffe
Efficiency
(NSE),
Percent
(Pbias),
Willmott
Index
(WI),
were
compare
analyse
established
model,
importance
coefficients
each
parameter
evaluated
Finally,
Conditional
Generative
Adversarial
Network
(CGAN)
was
augment
supplement
existing
database,
verify
its
effectiveness.
The
results
show
parameter-optimized
LightGBM
achieves
performance.
Moreover,
CGAN
can
effectively
insufficient
lack
specific
sample
data.
Язык: Английский
Experimental Study of Scouring and Deposition Characteristics of Riprap at Embankment Toe Due to Overflow
Geotechnics,
Год журнала:
2024,
Номер
4(3), С. 773 - 785
Опубликована: Июль 16, 2024
In
this
study,
the
effects
of
grain
size
and
gradation
riprap,
overtopping
flow
depth,
downstream
slope
embankment
on
scouring
deposition
characteristics
at
toe
were
investigated.
For
experiment,
three
different
slopes
(1:2,
1:3,
1:4),
overflow
depths
(1,
2,
3
cm),
sizes
riprap
particles
(d50
16.41
mm,
8.48
3.39
herein
referred
to
as
coarse
gravel,
medium
granule,
respectively)
used
in
laboratory.
The
experimental
results
demonstrated
that
scour
depth
height
increased
with
increasing
energy
head
for
each
condition.
Among
particle
sizes,
gravel
shows
lowest
highest
height.
1:2
slope,
was
62%
75%
less
resistant
than
granule
particles,
respectively.
1:3
case,
31%
46%,
1:4
39%
49%
Язык: Английский
A novel structural reliability analysis method combining the improved beluga whale optimization and the arctangent function‐based maximum entropy method
Quality and Reliability Engineering International,
Год журнала:
2024,
Номер
40(8), С. 4439 - 4461
Опубликована: Авг. 22, 2024
Abstract
A
novel
structural
reliability
analysis
method
that
combines
the
improved
beluga
whale
optimization
(IBWO)
and
arctangent
function‐based
maximum
entropy
(AMEM)
is
proposed
in
this
paper.
It
aims
to
augment
accuracy
of
failure
probability
prediction
based
on
traditional
(MEM).
First,
function
introduced
avoid
effects
truncation
error
numerical
overflow
MEM.
The
can
nonlinearly
transform
performance
defined
infinite
interval
into
a
transformed
bounded
interval.
Subsequently,
undetermined
Lagrange
multipliers
density
(MEPDF)
are
obtained
using
IBWO
at
swifter
convergence
speed
with
heightened
accuracy.
Finally,
MEPDF
be
by
combining
AMEM,
predicted.
metro
bogie
frame
as
an
engineering
example
reveals
compared
MEM
genetic
algorithm
solve
multipliers,
diminishes
relative
from
20.51%
only
0.09%.
This
significantly
enhances
probability.
Язык: Английский