Chemical
process
control
relies
on
a
tightly
controlled,
narrow
range
of
margins
for
critical
variables,
ensuring
stability
and
safeguarding
equipment
from
potential
accidents.
The
availability
historical
data
is
limited
to
specific
setpoint
operation.
This
challenge
raises
issues
monitoring
in
predicting
adjusting
deviations
outside
the
operational
parameters.
Therefore,
this
paper
proposes
simulation-assisted
deep
transfer
learning
optimizing
final
purity
production
capacity
glycerin
purification
process.
proposed
network
trained
by
simulation
domain
generate
base
feature
extractor,
which
then
fine-tuned
using
few-shot
techniques
target
learner
extend
working
model
beyond
practice.
result
shows
that
improved
prediction
performance
99%
water
content
79.72%
over
conventional
model.
Additionally,
implementation
identified
product
quality
improvement
enhancing
Processes,
Год журнала:
2024,
Номер
12(4), С. 661 - 661
Опубликована: Март 26, 2024
Chemical
process
control
relies
on
a
tightly
controlled,
narrow
range
of
margins
for
critical
variables,
ensuring
stability
and
safeguarding
equipment
from
potential
accidents.
The
availability
historical
data
is
limited
to
specific
setpoint
operation.
This
challenge
raises
issues
monitoring
in
predicting
adjusting
deviations
outside
the
operational
parameters.
Therefore,
this
paper
proposes
simulation-assisted
deep
transfer
learning
optimizing
final
purity
production
capacity
glycerin
purification
process.
proposed
network
trained
by
simulation
domain
generate
base
feature
extractor,
which
then
fine-tuned
using
few-shot
techniques
target
learner
extend
working
model
beyond
practice.
result
shows
that
improved
prediction
performance
24.22%
water
content
79.72%
over
conventional
model.
Additionally,
implementation
identified
product
quality
improvements
enhancing
