Buildings,
Год журнала:
2025,
Номер
15(4), С. 526 - 526
Опубликована: Фев. 9, 2025
The
construction
industry
accounts
for
approximately
28%
of
global
CO2
emissions,
and
emission
management
at
the
building
demolition
stage
is
important
achieving
carbon
neutrality
goals.
Systematic
studies
on
stage,
however,
are
still
lacking.
In
this
study,
research
development
optimal
machine
learning
(ML)
models
was
conducted
to
predict
emissions
stage.
were
predicted
by
applying
various
ML
algorithms
(e.g.,
gradient
boosting
[GBM],
decision
tree,
random
forest),
based
information
features
equipment
used
demolition,
as
well
energy
consumption
data.
GBM
selected
a
model
with
prediction
performance.
It
exhibited
very
high
accuracy
R2
values
0.997,
0.983,
0.984
training,
test,
validation
sets,
respectively.
also
showed
excellent
results
in
generalization
performance,
it
effectively
learned
data
patterns
without
overfitting
residual
analysis
mean
absolute
error
(MAE)
evaluation.
found
that
such
floor
area,
equipment,
wall
type,
structure
significantly
affect
area
key
factors.
developed
study
can
be
support
decision-making
initial
design
evaluate
sustainability,
establish
reduction
strategies.
enables
efficient
collection
processing
provides
scalability
analytical
approaches
compared
existing
life
cycle
assessment
(LCA)
approach.
future,
deemed
necessary
develop
tools
enable
comprehensive
through
system
boundary
expansion.
A
suitable
waste
management
strategy
is
crucial
for
a
sustainable
and
efficient
circular
economy
in
the
construction
sector,
requires
precise
data
on
volume
of
demolition
(DW)
gen-erated.
Therefore,
we
developed
an
optimal
machine
learning
(ML)
model
to
forecast
quantity
recycling
landfill
based
characteristics
DW.
dataset
comprising
infor-mation
150
buildings,
equipment
utilized,
five
types
generated
(i.e.,
recyclable
mineral,
combustible,
specified,
mix
waste,
minerals)
was
constructed.
ML
models
were
predict
quantities
such
waste.
Artificial
neural
network,
decision
tree,
gradient
boosting
machine,
k-nearest
neighbors,
linear
regression,
random
forest
(RF),
support
vector
regression
applied,
derived
via
hyperparameter
tuning.
The
RF
demonstrated
superior
performance.
In
both
validation
test
phases,
“recyclable
mineral
waste”
combustible
achieved
accuracies
0.987
0.972,
re-spectively.
metals”
“landfill
specified
0.953
0.858
or
higher,
respectively.
Moreover,
exhibited
accuracy
0.984
higher.
SHapley
Additive
exPlanations
analysis
highlighted
floor
area
as
primary
input
variable
influencing
type
employed
emerged
another
impacting
wastes
generated.
can
provide
management,
thereby
facilitating
decision-making
process
industry
professionals.
Buildings,
Год журнала:
2024,
Номер
14(11), С. 3695 - 3695
Опубликована: Ноя. 20, 2024
In
the
actual
estimation
of
construction
and
demolition
waste
(C&DW),
it
is
significantly
relevant
to
effective
management,
design,
planning
at
project
stages,
but
lack
reliable
methods
historical
data
prevents
C&DW
quantities
for
both
short-
long-term
planning.
To
address
this
gap,
study
aims
predict
in
projects
more
accurately
by
integrating
gray
wolf
optimization
algorithm
(GWO)
Archimedes
(AOA)
into
an
artificial
neural
network
(ANN).
This
uses
concerning
work
200
real-life
performed
Gaza
Strip.
Different
performance
parameters,
such
as
mean
absolute
error
(MAE),
square
(MSE),
root
squared
(RMSE),
coefficient
determination
(R2),
are
used
evaluate
effectiveness
models
developed.
The
results
have
shown
that
AOA-ANN
model
outperforms
other
terms
accuracy
(R2
=
0.023728,
MSE
0.00056304,
RMSE
MAE
0.0086648).
Moreover,
new
hybrid
yields
accurate
estimations
with
minimal
input
making
process
feasible.
Advances in environmental engineering and green technologies book series,
Год журнала:
2025,
Номер
unknown, С. 267 - 294
Опубликована: Янв. 16, 2025
Zero
waste,
as
defined
by
the
Waste
International
Alliance
(ZWIA)
refers
to
conservation
of
all
resources
means
responsible
production,
consumption,
reuse,
and
recovery
products,
packaging,
materials
without
burning
with
no
substantial
discharges
land,
water,
or
air
that
threaten
environment,
human
health,
various
other
life
forms.
An
estimated
11.2
billion
metric
tons
solid
waste
is
collected
every
year
worldwide,
approximately
5%
overall
greenhouse
gas
emissions
are
caused
decomposition
organic
elements
alone
in
environment.
It
projected
production
municipal
garbage
will
increase
from
2.3
2023
3.8
2050.
The
predicted
global
direct
cost
management
2020
was
$252
billion,
which
be
doubled
If
we
don't
find
a
solution
quickly,
it
may
become
unfixable
convert
earth
into
“gas
chamber.”.
using
AI-driven
technologies
sustainable
because
recycling
plastic
produces
hazardous
chemicals.
Buildings,
Год журнала:
2025,
Номер
15(4), С. 526 - 526
Опубликована: Фев. 9, 2025
The
construction
industry
accounts
for
approximately
28%
of
global
CO2
emissions,
and
emission
management
at
the
building
demolition
stage
is
important
achieving
carbon
neutrality
goals.
Systematic
studies
on
stage,
however,
are
still
lacking.
In
this
study,
research
development
optimal
machine
learning
(ML)
models
was
conducted
to
predict
emissions
stage.
were
predicted
by
applying
various
ML
algorithms
(e.g.,
gradient
boosting
[GBM],
decision
tree,
random
forest),
based
information
features
equipment
used
demolition,
as
well
energy
consumption
data.
GBM
selected
a
model
with
prediction
performance.
It
exhibited
very
high
accuracy
R2
values
0.997,
0.983,
0.984
training,
test,
validation
sets,
respectively.
also
showed
excellent
results
in
generalization
performance,
it
effectively
learned
data
patterns
without
overfitting
residual
analysis
mean
absolute
error
(MAE)
evaluation.
found
that
such
floor
area,
equipment,
wall
type,
structure
significantly
affect
area
key
factors.
developed
study
can
be
support
decision-making
initial
design
evaluate
sustainability,
establish
reduction
strategies.
enables
efficient
collection
processing
provides
scalability
analytical
approaches
compared
existing
life
cycle
assessment
(LCA)
approach.
future,
deemed
necessary
develop
tools
enable
comprehensive
through
system
boundary
expansion.
