Dynamic wavelet filter design for denoising partial discharge signal from insulated power cables

Due to the partial breakdown of insulation present in insulated power cable resulting from ageing, fast rising pulses are injected into cable core conductor and earth screen. Due to noise profile of substation resulted from conducted and radiated emissions, it is often challenging to interpret partial discharge events since classical digital filtering will not be efficient to extract such a wide band frequency partial discharge event. Multi resolution analysis using wavelet filter banks is widely used to extract partial discharge events when hampered by various noise sources. However, the success rate of multi-resolution analysis using wavelet filters relies on degree of correlation between the impulse response of wavelet filter and the discharge pulse itself. There is a clear requirement to have a filter bank having impulse response aligned with propagation characteristics of cables. This paper proposes novel algorithm in designing wavelet basis function using the measured impulse response of the cable under test. Impulse response of cable required to characterize insulated power cable is done by injecting current pulse into cable screen using high frequency current transformer with data acquisition systems. Analysis and synthesis filters perfect reconstruction filter banks are constructed using measured impulse response.