ACS ES&T Water,
Год журнала:
2023,
Номер
4(3), С. 784 - 804
Опубликована: Авг. 24, 2023
Wastewater
treatment
companies
are
facing
several
challenges
related
to
the
optimization
of
energy
efficiency,
meeting
more
restricted
water
quality
standards,
and
resource
recovery
potential.
Over
past
decades,
computational
models
have
gained
recognition
as
effective
tools
for
addressing
some
these
challenges,
contributing
economic
operational
efficiencies
wastewater
plants
(WWTPs).
To
predict
performance
WWTPs,
numerous
deterministic,
stochastic,
time
series-based
been
developed.
Mechanistic
models,
incorporating
physical
empirical
knowledge,
dominant
predictive
models.
However,
represent
a
simplification
reality,
resulting
in
model
structure
uncertainty
constant
need
calibration.
With
increasing
amount
available
data,
data-driven
becoming
attractive.
The
implementation
can
revolutionize
way
manage
WWTPs
by
permitting
development
digital
twins
process
simulation
(near)
real-time.
In
is
not
explicitly
specified
but
instead
determined
searching
relationships
data.
Thus,
main
objective
present
review
discuss
machine
learning
prediction
WWTP
effluent
characteristics
inflows
well
anomaly
detection
studies
consumption
WWTPs.
Furthermore,
an
overview
considering
merging
both
mechanistic
hybrid
presented
promising
approach.
A
critical
assessment
gaps
future
directions
on
mathematical
modeling
processes
also
presented,
focusing
topics
such
explainability
use
Transfer
Learning
processes.
Results in Engineering,
Год журнала:
2023,
Номер
20, С. 101428 - 101428
Опубликована: Сен. 26, 2023
Wastewater
treatment
plants
(WWTPs)
are
energy-intensive
facilities
that
play
a
critical
role
in
meeting
stringent
effluent
quality
regulations.
Accurate
prediction
of
energy
consumption
WWTPs
is
essential
for
cost
savings,
process
optimization,
regulatory
compliance,
and
reducing
carbon
footprint.
This
paper
introduces
an
efficient
approach
predicting
WWTPs,
leveraging
deep
learning
models,
data
augmentation,
feature
selection.
Specifically,
Spline
Cubic
interpolation
enriches
the
dataset,
while
Random
Forest
model
identifies
important
features.
The
study
investigates
impact
lagged
to
capture
temporal
dependencies.
Comparative
analysis
five
models
on
original
augmented
datasets
from
Melbourne
WWTP
demonstrates
substantial
performance
improvement
with
data.
Incorporating
further
enhances
accuracy,
providing
valuable
insights
effective
management.
Notably,
Long
Short-Term
Memory
(LSTM)
Bidirectional
Gated
Recurrent
Unit
(BiGRU)
achieve
Mean
Absolute
Percentage
Error
(MAPE)
values
1.36%
1.436%,
outperforming
state-of-the-art
methods.
Water,
Год журнала:
2023,
Номер
15(13), С. 2349 - 2349
Опубликована: Июнь 25, 2023
Wastewater
treatment
plants
(WWTPs)
are
energy-intensive
facilities
that
fulfill
stringent
effluent
quality
norms.
Energy
consumption
prediction
in
WWTPs
is
crucial
for
cost
savings,
process
optimization,
compliance
with
regulations,
and
reducing
the
carbon
footprint.
This
paper
evaluates
compares
a
set
of
23
candidate
machine-learning
models
to
predict
WWTP
energy
using
actual
data
from
Melbourne
WWTP.
To
this
end,
Bayesian
optimization
has
been
applied
calibrate
investigated
machine
learning
models.
Random
Forest
XGBoost
(eXtreme
Gradient
Boosting)
were
assess
how
incorporated
features
influenced
prediction.
In
addition,
study
consideration
information
past
improving
accuracy
by
incorporating
time-lagged
measurements.
Results
showed
dynamic
outperformed
static
reduced
The
shows
including
lagged
measurements
model
improves
accuracy,
results
indicate
K-nearest
neighbors
dominates
state-of-the-art
methods
reaching
promising
predictions.
ACS ES&T Water,
Год журнала:
2024,
Номер
4(4), С. 1904 - 1915
Опубликована: Апрель 2, 2024
Models
are
increasingly
being
utilized
to
improve
the
understanding
and
operation
of
wastewater
treatment
plants
(WWTPs)
in
face
escalating
water
resource
challenges.
Abundant
operational
data
provide
extensive
opportunities
for
development
machine
learning
(ML)
deep
(DL)
models.
However,
coupling
time
lag
among
features
exacerbate
black-box
nature
such
models,
hindering
their
application
WWTPs.
In
this
study,
we
construct
a
DL
model
using
long
short-term
memory
(LSTM)
algorithm
capable
accurately
predicting
effluent
quality
full-scale
WWTP
with
finely
tuned
hyperparameters
rationally
chosen
input
features.
Comprehensive
explanation
based
on
Shapley
additive
explanations
(SHAP)
is
implemented
clarify
contributions
multivariate
series
(MTS)
inputs
predicted
results
terms
feature
dimensions.
The
LSTM
models
exhibit
excellent
accuracy
(R2
0.96,
0.95,
0.76
MAPE
5.49,
7.17,
13.37%,
respectively)
chemical
oxygen
demand
(COD),
total
phosphorus
(TP),
nitrogen
(TN)
better
than
other
baseline
ML
SHAP
quantify
what
most
important
when
they
exert
influence
how
impact
results.
analysis
from
temporal
dimension
further
explains
characteristics
process
justifies
introduction
MTS.
Compared
correlation
without
engineering,
selection
method
by
significantly
enhances
predictive
accuracy.
combinations
adjusted
values,
strong
interactions
significant
output
identified.
This
novel
attempt
both
explainability
MTS
prediction
work
shows
potential
applying
WWTPs
performance.