Long-term Online Partial Discharge Trend Monitoring for Cast Resin Transformers in Switchboard

As an insulation deterioration diagnosis technology for distribution equipment, establishment of an online diagnosis technology based on partial discharge (PD) measurement is required. This research aims to establish a method for safely diagnosing the insulation status of distribution transformers from outside of a switchboard. An on-line PD monitoring system was built for 22 kV cast resin transformers installed at a high voltage substation. Six inexpensive self-made TEV sensors were used to detect PD signal with the same sensitivity as commercially available products. Namely, TEV1 and TEV2 to TEV6 were placed on the inner and outer wall of the switchboard, respectively. A waveform classification algorithm was developed, which automatically classifies the waveform by comparing the signal magnitude detected by each sensor with output by a 1 MHz high-pass filter. Huge amount of data obtained by the automatic measurement was used to distinguish between PD signals and external noises. Typical phase-resolved partial discharge (PRPD) patterns were obtained before and after applying the signal detected with TEV1 to the automatic classification process. As a result, PD was found to be generated randomly regardless of the phase angle before the algorithm is applied, while only the signal in a specific phase region appears after the algorithm application, i.e. the latter PRPD pattern matches typical pattern usually observed with void discharge. Furthermore, it was found that signals detected with the multiple TEV sensors on the switchboard wall can identify whether signal sources come from inside or outside the switchboard. In addition, the trend of the maximum amplitude (V_max) of each signal, the maximum value V_mday in a day, the average value V_mave, and the phase characteristic (PRPD pattern) of V_max were recorded over a long period of time so as to evaluate the state of the equipment.