MODELLING OF NONLINEAR SYSTEMS USING INTEGRATED OBFARX PLUS NEURAL NETWORK MODELS
The objective of this project is to develop a new model, which is by combining OBFARX linear model with nonlinear NN model. The results obtained will be compared with the previous models to show performance improvement by the new model. The new model development is based on the model developed by...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Year Project |
| Language: | English |
| Published: |
Universiti Teknologi Petronas
2014
|
| Subjects: | |
| Online Access: | http://utpedia.utp.edu.my/13853/1/DISSERTATION_fyp_September_2013_13034.pdf http://utpedia.utp.edu.my/13853/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The objective of this project is to develop a new model, which is by combining
OBFARX linear model with nonlinear NN model. The results obtained will be compared
with the previous models to show performance improvement by the new model. The new
model development is based on the model developed by (Zabiri et al 2011) which is OBF
linear model combination with nonlinear NN model. The OBF-NN model cannot work
efficiently on some problems due to the limitations of the OBF part of the equation. So it
is important to analyze the new model which is OBFARX-NN with OBF-NN model. The
scope for this project will be the development of the parallel OBFARX-NN model,
methods for estimating the model parameter, simulation analysis using MATLAB and
evaluation on OBFARX-NN model performance. The method for completing the project
will be firstly, make sure all the necessary information about the individual model is
available. Then develop a theoretically working OBFARX-NN model. After that,
analysis of the performance of the created model is done and also alterations here and
there for better clarification. All in all, the result are the improve performance of process
control by OBFARX-NN model compared to OBF-NN model.The most important aspect
of the model development is the extrapolation capabilities of the model itself. When a
model is forced to perform prediction in regions beyond the space of original training,
then it can be said that the model can function well even when the process parameter is
changed. This aspect is very important because in practical plant, the process conditions
are continually changing making extrapolation inevitable. Thus, by testing the
extrapolation capabilities of the OBFARX-NN model, the project had come up with the
subsequent RMSE value and compared with previous model. The RMSE value indicates
superior performance in the extrapolation region. |
|---|