Department of Computer Science and Electrical Engineering
The aim of the thesis was to develop a method for Expert System
Based Transformer Fault Diagnosis. It is determined that the most
suitable approach to the problem is Dissolved Gas Analysis (DGA). The paper
presents the development and final results of a classification system based
on mathematical and relational techniques that can provide early detection
of an abnormality in oil filled high voltage transformers.
It is known that there is little in this field that has not been pursued and hence the author develops a technique that combines the performance and advantages of several existing diagnostic methods while including some changes and common sense approaches that were developed from the literature review.
Following gas chromatography, gas composition data is entered into the expert system through user interaction with a software graphical user interface (GUI) where it is manipulated by a range of analytical processes as according to IEEE StdC57.104-1991.
To evaluate the success of the developed system it is necessary to analyse real life test data. Manipulating variables such as age, rated voltage and gas levels within the transformer generated output diagnoses and confidence factors.
Results obtained from testing proved that the diagnosis system accurately identifies fault conditions and generates correct advice thereafter. The expert system proves itself to be versatile with very few incorrect diagnoses and the combination of more than one diagnosis technique proves itself to be noteworthy to the user.
It is concluded that the developed expert system is a useful and
efficient tool for determining many fault conditions that may occur in
a transformer.
My complete conference paper is provided here:
Conference_Paper.pdf (62kB)
A copy of my complete thesis (including appendices) can be obtained by clicking below:
Thesis.pdf (1,512kB)
Additional materials:
Appendices - Short Description of My Thesis
Appendices
| Appendix A - Rogers' Ratio Source and Development Codes. (114kB) | |
| I did not develop the content of Appendix A and hence had to scan it for use, resulting in a large file. |  |
Appendix B - MATLAB Code. (12kB)
Some information about the Author is available
or please E-Mail me with
queries or comments.
Thankyou for visiting.
