Development of neural network based models to control temperature and estimate composition of a debutaniser column / Nasser Mohamed Ramli
The current method for composition measurement of an industrial distillation column specifically offline method, is slow, tedious and could lead to inaccurate results. Among the advantages of using online composition designed are to overcome the long time delay introduced by laboratory sampling a...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2015
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/6139/ http://studentsrepo.um.edu.my/6139/1/Appendices_ips.pdf http://studentsrepo.um.edu.my/6139/2/chapter_1_to_chapter_7_include_reference_ips_format_margin.pdf http://studentsrepo.um.edu.my/6139/3/table_content_and_appendices_ips_format_margin.pdf http://studentsrepo.um.edu.my/6139/4/title_ips_format.pdf |
| Summary: | The current method for composition measurement of an industrial distillation column
specifically offline method, is slow, tedious and could lead to inaccurate results. Among
the advantages of using online composition designed are to overcome the long time
delay introduced by laboratory sampling and provide better estimation, which is suitable
for online monitoring purposes. Principal component and partial least square analysis
are used to determine the important variables surrounding the column prior to
implementing the neural network. It is due to the different types of data available for the
plant, which requires proper screening in determining the right input variables to the
dynamic model. Statistical analysis is used as a model adequacy test for the composition
prediction of n-butane and i-butane in the column. Simulation results showed that the
Artificial Neural Network (ANN) can reliably predict the online composition of the
column. The major contribution of the current research is the development of
composition prediction of n-butane and i-butane using equation based neural network
(NN) models. Based on statistical analysis, the results indicate that neural network
equation, which is more robust in nature, predicts better than the PLS equation and RA
equation based methods. The temperature predictions using neural network equation are
also compared with partial least square (PLS) and regression analysis (RA) equations
methods. A new technique for nonlinear system, which is based on hybrid neural
network modeling, is proposed. The hybrid model consists of combination of residual
composition and residual temperature with first principle in terms of mass and energy
balance. Hybrid neural network equation performs better than the hybrid neural
network, and neural network predictions to estimate composition and temperature for
the column. The use of an inverse neural network and forward neural network are used
for the direct control of a distillation column. The neural network used for the control
strategy to track the set point of the top and bottom temperature. Neural network
iv
estimators are used to track the set point of the top and bottom composition together
with disturbances. There are two types of controller used for control strategies which are
the direct inverse control (DIC) and internal model controller (IMC). Based on the
results, IMC and DIC were found to perform better in controlling the temperature with
respect to set point changes and disturbances compared to conventional PID controllers. |
|---|