Detection and identification technologies of wide band partial discharge

A wide-band online partial discharge (PD) detection method is discussed to minimize the influence of the noise and interferences during the course of measurement and diagnosis followed. The traditional IEC pulse current detection method and the recently developed ultra high frequency detection method have limited their bandwidth to avoid the abundant ambient noise and interference, and loss of the information of the PD signals as well. The methods can not record the entire discharge waveforms of the PD current pulses, causing a fundamental lack of information to neither differentiate various signals and noises, nor distinguish various PD or interference patterns. The wide band detection method covers a bandwidth wide enough to record the entire partial discharge pulse waveforms. Based on the features of their pulse waveforms, such signals can be separated and recognized in a 'clean' condition. An experiment employing such method on a 110 kV cable with several discharge patterns and onsite noises is presented. The analysis shows that, with sufficient information of the entire waveform, those discharge patterns can be separated, and each of the 'clean and pure' sub-patterns, such as the internal discharge and background noises can be isolated and well recognized. For field PD detection and diagnosis, the method would increase the ability to remove the impact of onsite noises and interferences, as well as the ability to diagnose the insulation condition based on the PD information.