The use of three dimensional filters for on-line partial discharge localisation in large transformers

Partial discharge (PD) activity develops in high voltage transformers due to ageing processes in the insulation system which are totally unavoidable. Moreover, operational overstressing and defects introduced during manufacture can also cause the existence of PD activity. The presence of this activity needs to be identified and treated, otherwise it will lead to further ageing and degradation processes until eventually there may be catastrophic failure. Previous reported work by the authors used Wavelet Transforms and Principal Component Analysis (PCA) separately in order to develop an approach for PD localisation within a transformer winding which is more suitable for implementation in an off-line mode. In this paper a new technique has been developed for on-line application using filters designed based on wavelet functions in the frequency domain and PCA loading coefficients. The combination of wavelet and PCA produces a unique frequency spectrum for the three dimensional filters. A transfer function of corresponding frequency spectrum can be obtained using estimation based on an IIR filter approximation. This paper reports on selection of filter order and its influence on PD source localisation. Ultimately, this study can be used to determine the optimum selection of the three dimensional filters for a given discharging transformer that will facilitate on-line localisation of the PD source.