Use this URL to cite or link to this record in EThOS:
Title: An investigation into partial discharge activity within three-phase belted cables
Author: Hunter, Jack A.
ISNI:       0000 0004 2732 016X
Awarding Body: University of Southampton
Current Institution: University of Southampton
Date of Award: 2013
Availability of Full Text:
Access from EThOS:
Access from Institution:
Industrially driven interest in the field of partial discharge (PD) diagnostics has rapidly increased in recent years. Utilities are turning to continuous asset monitoring methods to inform them on the real-time health of plant. The majority of London's medium voltage (MV) distribution network is constructed from paper insulated lead covered (PILC) belted cables. The vast majority of this cable was commissioned in the 60's and 70's and is now nearing the end of its design life. PD diagnostics have been proposed as a possible tool for the condition monitoring of these distribution cables. Little is known about the characteristics of the PD activity that is produced as cables of this design degrade under rated conditions. This thesis describes the development of a PD measurement experiment that records PD data from either defective or damaged three-phase MV PILC cables under rated voltage conditions. The experiment has been designed to replicate the environment experienced by cable circuits in the field. The aim was to investigate the potential transfer of knowledge generated by the experiment onto an on-line commercial operational system. An investigation into the PD produced by the various degradation mechanisms have been undertaken to evaluate the relationship between the PD source conditions and recorded signals. It has been found that the phase-resolved PD patterns produced by different degradation mechanisms are unique. Consequently, a PD source discrimination technique has been successfully applied to both experiment and field data. The algorithm relies on the finding that the wavelet energy (WE) distribution of a PD pulse is source dependent. A support vector machine (SVM) was used to accurately classify PD pulses from different sources that had been tested experimentally. The ability to accurately discriminate between different PD sources in both experiment and field data should lead to a significant step forward in the field of PD diagnostics.
Supervisor: Lewin, Paul Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: QA75 Electronic computers. Computer science