Picture of person typing on laptop with programming code visible on the laptop screen

World class computing and information science research at Strathclyde...

The Strathprints institutional repository is a digital archive of University of Strathclyde's Open Access research outputs. Strathprints provides access to thousands of Open Access research papers by University of Strathclyde researchers, including by researchers from the Department of Computer & Information Sciences involved in mathematically structured programming, similarity and metric search, computer security, software systems, combinatronics and digital health.

The Department also includes the iSchool Research Group, which performs leading research into socio-technical phenomena and topics such as information retrieval and information seeking behaviour.


Deep neural networks for understanding and diagnosing partial discharge data

Catterson, V. M. and Sheng, B. (2015) Deep neural networks for understanding and diagnosing partial discharge data. In: 2015 IEEE Electrical Insulation Conference (EIC). IEEE, Piscataway, NJ, USA, pp. 218-221. ISBN 9781479973521

Text (Catterson-Sheng-IEEE-EIC-2015-Deep-neural-networks-for-understanding-and-diagnosing-partial)
Catterson_Sheng_IEEE_EIC_2015_Deep_neural_networks_for_understanding_and_diagnosing_partial.pdf - Accepted Author Manuscript

Download (262kB) | Preview


Artificial neural networks have been investigated for many years as a technique for automated diagnosis of defects causing partial discharge (PD). While good levels of accuracy have been reported, disadvantages include the difficulty of explaining results, and the need to hand-craft appropriate features for standard two-layer networks. Recent advances in the design and training of deep neural networks, which contain more than two layers of hidden neurons, have resulted in improved results in speech and image recognition tasks. This paper investigates the use of deep neural networks for PD diagnosis. Defect samples constructed in mineral oil were used to generate data for training and testing. The paper demonstrates the improvements in accuracy and visualization of learning which can be gained from deep learning.