Download the full-sized PDF of An Intelligent Fault Diagnosis Approach for Power Transformers Based on Support Vector MachinesDownload the full-sized PDF



Permanent link (DOI):


Export to: EndNote  |  Zotero  |  Mendeley


This file is in the following communities:

Graduate Studies and Research, Faculty of


This file is in the following collections:

Theses and Dissertations

An Intelligent Fault Diagnosis Approach for Power Transformers Based on Support Vector Machines Open Access


Other title
support vector machine
power transformer
fault diagnosis
Type of item
Degree grantor
University of Alberta
Author or creator
Xu, Hao
Supervisor and department
Tian, Zhigang (Mechanical Engineering)
Examining committee member and department
Zuo, Ming (Mechanical Engineering)
Tian, Zhigang (Mechanical Engineering)
Ma, Yongsheng (Mechanical Engineering)
Department of Mechanical Engineering

Date accepted
Graduation date
2017-11:Fall 2017
Master of Science
Degree level
Power transformers are essential for the operations of industrial systems such as metal production plants, and for the transmission and distribution of electricity to end users. Power transformer failures can cause huge loss of production, expensive downtime, significant costs for repair or replacement, and disruptions to city and community operations. Transformers are desired to operate at a high-reliability level, and they should be maintained carefully through effective condition monitoring and fault diagnosis, for evaluating transformer health conditions based on condition monitoring data and performing suitable maintenance actions. Dissolved gas analysis (DGA) is a primary way of monitoring the health conditions of transformers by analyzing the insulation oil via periodic sampling. Different gases can be decomposed from the insulation material and the liquid oil under certain thermal, electrical, or mechanical stresses, and these gases will dissolve into the transformer oil. Existing transformer fault diagnosis methods mainly include rule-based methods documented in IEEE Standards, which are based on analyzing key gases, gas concentration ratios, or certain gas proportions. In addition, artificial intelligence (AI)-based methods were proposed, based on artificial neural network, fuzzy logic or support vector machine (SVM) tools. However, the existing rule-based and AI-based methods suffer from limited and imbalanced datasets and the capability to deal with low concentration DGA data, and the fault diagnosis accuracy needs to be further improved. In this thesis, a new intelligent approach based on SVM is proposed for condition monitoring and fault diagnosis of power transformers based on DGA data. The proposed method integrates a gas concentration filter and a plurality-voting SVM model. Low concentration data are typical for new transformers, but existing ratio-based methods are generally not effective in utilizing such data. A gas concentration filter is proposed to process low gas concentrations data, and it is combined with the SVM model to generate fault diagnosis results. The plurality-voting SVM model is designed with a new plurality-voting structure and integrates the synthetic minority over-sampling technique (SMOTE) to overcome the problem of imbalanced data, where the dataset sizes are significantly different for different health conditions. A parameter optimization approach based on genetic algorithm is employed. The proposed SVM-based approach is compared with existing DGA-based power transformer diagnosis methods, including rule-based methods and various AI methods. The comparative study results demonstrate the effectiveness of the proposed SVM-based power transformer fault diagnosis approach.
This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for the purpose of private, scholarly or scientific research. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.
Citation for previous publication

File Details

Date Uploaded
Date Modified
Audit Status
Audits have not yet been run on this file.
File format: pdf (PDF/A)
Mime type: application/pdf
File size: 3993005
Last modified: 2017:11:08 18:02:06-07:00
Filename: Xu_Hao_201709_MSc.pdf
Original checksum: 04741a883129e20c457eb1ceead0e6c3
Activity of users you follow
User Activity Date