| Summary: | 博士 === 國立清華大學 === 化學工程學系 === 85 === The main purpose of this study is to develop an algorithm to
solve the interaction and nonlinear problems in chemical
processes. The other purpose is to develop a method for
determination of the optimal operation in a complicate process.
The temperature control exhibits poor performance in a high
purity distillation column. The on-line analysis control poses
difficulties due to the presence of the long dead-time. A
dynamic state variable estimator (DSVE) is proposed in this
study. We demonstrate the potential and versatility of NNM for
developing DSVE particularly useful for inferential product
composition estimation and control in a high purity distillation
column. A nonlinear simplified decoupler is developed based on a
neural network model(NNM). The NNM decoupler is applied to a
simulated CSTR and distillation column, respectively. The
results of the simulation study show the good capability of the
NNM decoupler for multivariable systems.The crude fractionation
column is a complicate process. It contains 13 manipulated
variables and 7∼8 controlled variables. In this study, a crude
fractionation column model is developed based on a neural
network. The predicted results of neural network model match the
plant operating data well. A combination of the neural network
model and a successive quadratic programming (SQP) algorithm is
employed to determine the optimal operating conditions of the
crude fractionation column.Many control strategies applied to a
high purity distillation column with a sidestream are
illustrated in this study. We compare the differential
temperature control and double differential temperature control,
the direct internal reflux control and indirect internal reflux
control.The simulation results show the excellent capability of
the feed-forward control to disturbance rejection. An indirect
model-base control strategy applied to product composition
control of a high purity distillation column is also developed.
The indirect internal reflux control, feed-forward control, and
indirect model-base control strategies are successfully
implemented to a commercial distillation column.
|
|---|
