PD pattern identification using acoustic emission measurement and neural networks

Partial discharges (PD) cause the degradation of insulation and may lead to the breakdown of power apparatus and cables, therefore, it is very important to recognize and locate such partial discharges. Acoustic emission (AE) techniques can be applied to PD detection and have proved to be effective. AE relies on the detection of waves produced by a sudden deformation in stressed materials. These waves travel from the source to the sensor(s) where they are converted to electrical signals. This approach is free from electrical interference and suitable for online monitoring. This paper describes current research at the University of Southampton (UK) which is concerned with the development of a neural net-based online monitoring system using AE method that is capable of determining not only the accurate location but also the type of partial discharge within a high voltage power cable joint.