A signal processing technique for PD detection and localization in power transformer
The activity of partial discharge (PD) is major problem that causes degradation to the insulation system. Due to continuous stresses of high voltage it may lead to complete breakdown of insulation system. Due to the high cost of transformer windings, especially in the case of high kVA capacity, it i...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2014
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/48046/25/MohammedAbdAliAzizMFKE2014.pdf http://eprints.utm.my/id/eprint/48046/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The activity of partial discharge (PD) is major problem that causes degradation to the insulation system. Due to continuous stresses of high voltage it may lead to complete breakdown of insulation system. Due to the high cost of transformer windings, especially in the case of high kVA capacity, it is necessary to find a way to localize the PD along the winding length so that only faulty part is replaced instead of the whole winding. In this project a new technique of a digital signal processing method based on the Fast Fourier Transformation (FFT) is applied to locate the PD. The transformer winding was modelled in Matlab using an equivalent lumped transformer circuit for each winding section. The PD source is injected at different section, each section represent a different PD location. The output current was then analysed using FFT. For validation, comparisons had been made with the results of an experimental work on the same transformer winding configuration as well as with theoretical equations. The work has shown that the frequency spectra of the simulated output current have the same characteristic as that for the measured signal as well as for the theoretically derived transfer function. It is also shown that the crests and troughs in the frequency spectra could be used for locating the source of the discharge activity in the power transformer winding. The frequency of the zero in the simulated spectra increases as the discharge moves away from the measuring terminal. The poles are however not affected by the position of the PD source. An Artificial Neural Network (ANN) was successfully used to determine the section number where PD occurs based on just the frequency response of the output current. Further studies on the effects of the modelling parameters such as variations in the capacitance and inductance values due to factors such as ageing on the PD localization show that the capacitance is the most sensitive parameter in the model. In addition, it is also shown that the effect of noise on the PD localization can be eliminated using the Free Induction Decay (FID) technique. |
|---|