Practical Application of Neural Networks in Measuring PD in MV Power Cables by Joel Yeo Wei Wen

Partial discharge diagnostics has become a fundamental process at modern electricity supply utilities. Moreover, research into the physics of PDs has enabled development of monitoring systems with high resolution and reliability to the point that spot testing and online diagnostic efforts generally yield favourable results. This presentation explains an algorithmic approach constructed by a convolutional recurrent neural network coupled with feature engineering for PD analysis measured from offline medium voltage cable tests. Two case are outined where results confirm that the methodology is able to identify and accurately localize discharge activity.