Wang, Dian (2011) Ontology-based fault diagnosis for power transformers. Masters thesis, University of Liverpool.
|PDF (Renamed file) - Accepted Version |
Available under License Creative Commons Attribution No Derivatives.
Electrical engineers are frequently required to undertake tasks on condition assessment, fault diagnosis, operation decision-making and maintenance of power transformers, based on their knowledge, experience and expertise by comparing present and past measurement data. In some cases, it requires a team of experts in different areas and a huge amount of information has been produced and stored for such tasks. The volume of data has exceeded the capability of data analysis of engineers with limited individual knowledge, as these data are in association with complex and comprehensive concepts and knowledge of power system operations. Therefore, new techniques for knowledge representation, automated data analysis and decision-making are required, in order to reduce the need for human intervention in handling the complex data and individual knowledge. Expert-system is widely used for transformer fault diagnosis, which is provided with strong pertinence yet the expansibility is comparatively weak. Pursuant to the diversified inference mechanism and knowledge library structure, the former problems on knowledge exchange in such systems could never be solved. The current diagnosis systems for transformer faults are mostly based on detecting variations of a transformer. Given no effective integration for the current methods of diagnosis, it is necessary to introduce a new system which can integrate a variety of diagnostic methods to enhance the diagnosis efficiency. Ontology is a mechanism that describes concepts and their system relationships. An ontology-based knowledge representation has several attractive features and holds the fact that it focuses on the classification and constraints of allowable taxonomies and definitions. The formal nature of ontology also enables the integration of data from heterogeneous sources. In this thesis, ontology is employed to enhance the exchanging-study ability between heterogeneous systems as well as realizing authentic knowledge exchange. As the knowledge foundation of the whole system, an ontology knowledge library guarantees the realization of a higher-level knowledge exchange. In order to develop an ontology system for the fault diagnosis of power transformers, it is necessary to analyse numerous concepts and relationships exhibited for power transformers. This thesis proposes a power transformer fault diagnosis system with ontology, which is concerned as a part of power system ontology. This ontology provides a semantic model for knowledge representation and information management. It can be used to integrate a number of transformers diagnostic methods, such as transformer thermal condition monitoring and diagnosis, dissolved gas analysis, partial discharge analysis and frequency response analysis etc. A new approach to transformer fault diagnosis is introduced in this thesis based on the idea of exchanging information with explicit, formal and machine accessible descriptions of meaning by using the Semantic Web. An ontology model is developed for accurate and efficient fault diagnosis for power transformers. Through the use of this model, various transformer faults diagnostic methods can be integrated to describe an inference among fault phenomena, fault sources and causes of faults. The proposed ontology model provides a dedicated semantic model for knowledge representation and information management concerning fault diagnosis of power transformers. Finally, a systematic summary is given. Challenges are discussed and future research work is suggested.
|Item Type:||Thesis (Masters)|
|Departments, Research Centres and Related Units:||Academic Faculties, Institutes and Research Centres > Faculty of Engineering > Department of Electrical Engineering and Electronics|
|Deposited On:||07 Aug 2012 09:50|
|Last Modified:||07 Aug 2012 09:50|
Repository Staff Only: item control page