Autonomous tracking of partial discharge pattern evolution based on optical flow

Measurement of partial discharges (PD) is one of the key indicators of high voltage insulation integrity. They long term effect may lead to insulation deterioration and breakdown. This paper presents a novel technique for autonomous tracking of partial discharge image sequence evolution. The approach is based on motion detection utilizing optical flow algorithm. The goal of the presented research was to automate pattern analysis, leading towards intelligent PD diagnostics. Experiments have been performed on various classes of PD images representing monitoring sequence, partial discharges in stator bars representing electrical machine insulation, effect of harmonics and development of corona phases. Optical flow based detection was successfully applied to the sequence of PD images. Presented approach of autonomous tracking of partial discharge pattern evolution based on optical flow may be especially suitable for monitoring applications, however it can be also applied to off-line diagnostics. Motivation to apply motion analysis to a PD pattern is based on the conviction and direction of future autonomous AI-supported partial discharge diagnostic systems.

Automated summary

The measurement of partial discharges is crucial for high voltage insulation integrity, as their long term effects can lead to insulation deterioration and breakdown. This paper presents a novel technique for autonomous tracking of partial discharge image sequence evolution using the optical flow algorithm. The goal of the research was to automate pattern analysis towards intelligent PD diagnostics, with successful application to various classes of PD images.