Anomaly Detection in Insulation of Switch-Disconnectors in Diesel Power Station Using ABNET, PSOM and MLP

Ricardo Henrique Fonseca Alves¹; Getúlio Antero de Deus Júnior¹; Carlos Eduardo Alves da Costa¹; Rodrigo Pinto Lemos¹

¹ Universidade Federal de Goiás (UFG)

doi:10.20906/CPS/SBSE2016-0127

Resumo

This paper main goal is to detect insulation abnormality in switch-disconnectors installed in diesel power stations using thermographic images. It was used image processing techniques and pattern recognition by means of artificial neural network with supervised and adaptive training. In order to implement the anomaly detection system it was necessary to capture thermographic images of switch-disconnectors with and without dry band issues. A thermographer certified by ITC took the collected images used in this paper. The language of technical computing, MATLAB©, was used to make the preprocessing of all the collected images and it was also used for the training of an ABNET network, a PSOM network and a MLP network. Through the cross-validation method, the hit rate of the ABNET's network was 81.5%, the hit rate of the PSOM was 87.4% and for the MLP the hit rate was 92.6%. The simulation shows evidence of the use of the ABNET network, since the MLP network has an architecture more complex. However, as new information such as temperature and humidity appears a MLP may become more attractive.

Palavras-chave: Artificial neural networks; power systems; electromechanical devices; maintenance engineering; thermography