Frontiers in Sustainable Food Systems,
Год журнала:
2023,
Номер
7
Опубликована: Апрель 13, 2023
Evapotranspiration
is
considered
as
one
of
the
most
crucial
surface
fluxes
describing
water
movement
from
land
to
atmosphere
in
form
evaporation
soil
and
transpiration
plants.
Several
evapotranspiration
models
exist,
but
their
accuracy
subject
change
because
differences
between
underlying
assumptions
used
formulation
conditions
application
at
hand.
The
appropriate
selection
an
model
necessary
ensure
accurate
estimation
crop
requirements.
This
work
compares
20
different
for
cucumber
crops
grown
a
cooling-based
greenhouse
with
CO
2
enrichment
located
high
solar
radiation
region.
are
classified
into
temperature-based,
radiation-based,
mass
transfer-based,
combination
models.
These
assessed
against
direct
gas
exchange
measurements
crops.
performance
evaluated
using
nine
statistical
indicators
determine
suitable
under
study.
Results
demonstrate
that
among
temperature-based
models,
Schendel
Blaney
Criddle
resulted
best
prediction,
contrary
Hargreaves
Samani
which
presented
worst
performance.
Transpiration
estimates
Rohwer
were
closest
Trabert
furthest
measured
data
amongst
other
mass-transfer
based
Abtew
was
predicting
model,
while
Priestley
Taylor
exhibited
radiation-based
category.
combination-based
FAO56
Penman
Monteith
entailed
all
can
be
method
estimate
enriched
HVAC
greenhouses
regions.
Nonetheless,
parametrization
this
still
should
achieve
better
accurately
evaluate
effect
radiation,
cooling
agricultural
application.
Applied Energy,
Год журнала:
2023,
Номер
343, С. 121190 - 121190
Опубликована: Май 11, 2023
The
greenhouse
microclimate,
especially
temperature,
is
highly
complex,
and
controlling
it
requires
significant
resources
due
to
the
greenhouses'
inefficient
design.
application
of
model
predictive
control
a
promising
strategy
for
temperature
efficient
management.
However,
does
not
account
inaccuracies
uncertainties
existing
in
system,
leading
sub-optimal
temperatures.
Therefore,
this
study
proposes
comprehensive
data-driven
robust
framework
its
energy
utilisation
assessment
presence
uncertainties.
First,
an
analytical
based
on
mass
balance
artificial
neural
network
developed,
their
prediction
performance
compared.
demonstrates
higher
accuracy
used
as
system
proposed
framework.
A
strategy,
minimax
objective
function
particle
swarm
optimisation
algorithm,
developed
handle
system.
Results
illustrate
that
uncertainties,
outperforms
climate
management
basic
with
RMSE
0.32
°C
0.60
two-day
simulation
period
winter
summer,
respectively.
Furthermore,
leads
reduction
9.67%
23.61%
summer.
flexible
general
can
be
applied
other
greenhouses
different
configurations
cultivated
crops
by
fine-tuning
new
data
set.
Food and Energy Security,
Год журнала:
2025,
Номер
14(1)
Опубликована: Янв. 1, 2025
ABSTRACT
The
growing
global
challenges
of
environmental
degradation
and
resource
scarcity
demand
innovative
agricultural
solutions.
Intelligent
control
systems
integrating
sensors,
automation,
artificial
intelligence
(AI)
optimize
crop
production
sustainability
in
vertical
farming.
This
review
explores
the
critical
role
these
technologies
monitoring
adjusting
key
parameters,
including
light,
temperature,
humidity,
nutrient
delivery,
CO₂
enrichment.
use
real‐time
data
from
sensor
networks
to
continuously
maintain
optimal
conditions.
Sensors
measure
changes
environment,
such
as
light
intensity
humidity
levels.
Automation
enables
tasks
be
performed
without
human
intervention,
ensuring
consistent
adjustments
AI
predicts
plant
responses
proactive
management
strategies
this
context.
also
examines
how
integrate,
highlighting
successful
case
studies
addressing
like
energy
management,
scalability,
system
harmonization.
Looking
ahead,
AI's
potential
predictive
maintenance
emerging
trends
farming
highlight
transformative
intelligent
enhancing
efficiency
sustainability.
IEEE Access,
Год журнала:
2022,
Номер
10, С. 32190 - 32212
Опубликована: Янв. 1, 2022
This
literature
review
extends
and
contributes
to
research
on
the
development
of
data-driven
optimal
control.
Previous
reviews
have
documented
model-based
control
in
isolation
not
critically
reviewed
reinforcement
learning
approaches
for
adaptive
frameworks.
The
presented
discusses
model-free
controllers,
highlighting
use
data
In
frameworks,
methods
may
be
used
derive
policy
dynamical
systems.
Attractive
characteristics
these
include
requiring
a
mathematical
model
complex
systems,
their
inherent
capabilities,
being
an
unsupervised
technique
decision-making
abilities,
which
are
both
advantage
motivation
behind
this
approach.
considers
previous
topics,
including
recent
work
methods.
addition,
shows
system
dynamics,
determine
using
feedback
information,
tune
fixed
controllers.
Furthermore,
summarises
various
corresponding
characteristics.
Finally,
provides
taxonomy,
timeline
concise
narrative
underlines
limitations
techniques
due
lack
theoretical
analysis.
Areas
further
analysis
stability
robustness
explainability
black-box
evaluation
impact
extension
simulators
digital
twins.
Scientific African,
Год журнала:
2023,
Номер
19, С. e01578 - e01578
Опубликована: Фев. 5, 2023
For
cleaner
and
sustainable
greenhouse
crops
production,
it
is
essential
to
successfully
manage
the
needs
resources.
Thus
prediction
of
microclimate,
especially
temperature
relative
humidity
great
interest.
The
research
done
in
this
area
is,
however,
still
limited,
a
number
machine
learning
techniques
have
not
yet
been
sufficiently
exploited.
objective
paper
evaluate
two
modeling
(machine
(Artificial
Neural
Networks
(ANN),
Support
Vector
Machine
(SVM),
Bagging
trees
(BG)
Boosting
(BT))
Computational
Fluid
Dynamics
(CFD)
methods
assess
impact
seasonal
changes
on
performances.
study
was
carried
out
commercial
located
Agadir,
Morocco,
experimental
data
were
collected
during
October
March.
Results
show
that
all
predictive
models
are
capable
predicting
inside
air
(Tin)
(Rhin)
with
quite
good
precision
(R>0.98,
nRMSE<7%).
However,
time
required
by
much
more
less
than
one
CFD
model.
reason,
selected
for
further
analysis
assessment
seasonality
their
prove
efficiency
Tin
Rhin
agreement.
A
"combined
data"
model,
built
from
months,
tested
proved
its
March
separately
at
same
nRMSE
<9%).
