Partial discharge (PD) techniques have been used for many years for detecting incipient insulation faults in power transformers. However, reliable electrical techniques for locating the PDs still need to be developed. This lack of development is caused by the difficulty in extracting information from PD terminal signal measurements and relating this information to PD location. The main objective of this thesis is to develop an electrical PD location technique for use in power transformers. The overall approach consists of three elements. First, a computer simulation model is developed. Simulated PD pulses are injected into the winding at all locations at which a PD might occur. A file of results is created of all the terminal responses to the injected PDs. Second, a PD is injected into the transformer winding undergoing experimentation at some specific site, and the PD signals measured at the terminals are then analysed and the characteristics relating to the location of the PD extracted. Third, algorithms are constructed for matching the file of calculated results with the experimental results for the purpose of identifying the location of the PD in the transformers. A computer simulation model was specifically developed for studying the characteristics of PD propagation in power transformer windings. The model includes a lumped-element, multi-winding transformer model and a PD waveform model. Simulation analyses are carried out to define the transfer functions from the PD locations to the measuring terminals. Features contained in the spectra of the PD terminal signals relating to PD location are extracted. The effect of different winding construction, and terminal connections, on PD propagation are studied. Experimental tests were performed on full-size transformer windings to investigate the differences between the PD terminal signals produced when a PD pulse was injected at different locations. Experimental results were obtained for a multi-winding 220kV transformer winding and for a single 110kV transformer winding. Finally, PD location algorithms are developed based on comparing the characteristics of the PD signals measured at the transformer terminals experimentally, with those obtained from simulation analysis. Feature template matching methods and artificial neural network techniques are developed. It is shown that for the 110kV transformer winding, the feature template matching PD location algorithm predicts the location of the PD source with an accuracy better than 5% of the winding length. However, for high power, high voltage, multi-winding transformers, further refinements of the feature template matching PD location algorithm are required. Likewise the artificial neural network PD location algorithm, whilst showing promise, also needs further research to be applied to such transformer windings. In brief, the work presented indicates that the overall approach used to locate PDs in power transformers is sound. Methods, techniques and algorithms have been developed which with further refinement will allow substantial advances to be made in this field.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:594385 |
| Date | January 1999 |
| Creators | Wang, Zhongdong |
| Publisher | University of Manchester |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
Page generated in 0.0018 seconds