PROCESS CONTROL SYSTEM IDENTIFICATION

System Identification is an art of dealing with a problem of generating workable model
of dynamic response based on the observed dataset from the actual system. The
modelling process is based on the observed input and output data of a system.
The objective of this project is to design and implement System Identification for
Liquid System Pilot Plant in UTP by applying both the conventional System
Identification technique known as empirical modeling and the intelligent techniques, a
computer based method named System Identification Toolbox. Then, the comparison
study between intelligent techniques and conventional modelling technique is
conducted for a better performance determination.
System Identification procedure involves the construction of a model from actual data
and the model validation process. The construction of a model engages with three basic
entities that are data record, model structure and determination of the best model. By
following the System I dentification procedure, the four steps taken in accomplishing
this project were: (1) experimental design, (2) modelling via empirical modelling, (3)
simulation of System Identification via MATLAB-Simulink and (4) investigate
performance Comparison between empirical modelling and model predictor using
System Identification Toolbox.
Empirical modelling is a simple graphical and calculation technique. A linear transfer
function that is obtained from this method is adequate for the project implementations.
The second method is intelligent method which is carried out with the aid of MATLAB
software. All the selected best models are capable to reproduce the observed data with
minimum predicted error.
At the end of the project, based on some comparison and analysis, the author concludes
that an intelligent technique gives a better performance compared to the conventional
technique.
n