Time-frequency characteristics of abnormal waveshapes in PD pulse sequence using ultra-wideband detection

In the present study, a HVDC PD test platform with bandwidth of 10 kHz-50 MHz and 250 MS/s sampling rate is built for the ultra-wideband (UWB) detection and investigation of time-frequency (T-F) characteristics of abnormal waveshapes or waveforms in PD pulse sequence. It points out that PD UWB detection systems currently developed resorting to the pulse sequence grouping technology (PSGT) to achieve the PD and noise sources separation were realized by feature extraction of pulse waveforms using T-F map or equivalent T-F method (ETFM), which are based that the single recorded pulse waveforms in time and corresponding frequency domain generated by PD or noise source are regular and stable. Considering the phenomenon that the single recorded waveform in time domain presenting some abnormal patterns such as multi-peaks, continuous discharges and even multi-pulses during 1-2 mu s in the measured PD pulse sequence, an algorithm for real-time discrimination of abnormal pulse waveform in time domain using threshold moving window (TMW) is designed and approved by the HVDC tests of three typical defect discharge models (corona discharge, inner discharge and surface discharge), which can be used to enhanced applicability for the PSGT mentioned above. While, T-F characteristics of those abnormal waveforms in PD pulse sequences of HVDC tests are shown, which reconfirm that the marked single recorded waveforms in pulse sequence should be excluded for using the PSGT with T-F map or ETFM. (c) 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).