Diagnosing Faults in Power Transformers With Autoassociative Neural Networks and Mean Shift

• Type: Article in International Scientific Journal
• Date: 2012

Authors

Vladimiro Miranda (Author) FEUP; Adriana Castro (Author) FEUP; Shigeaki Lima (Author) FEUP

Journal

Title: IEEE Transactions on Power Delivery

Vol. 27 No. 3

Pages: 1350-1357

ISSN: 0885-8977

Publisher: IEEE

Indexing

ISI Web of Knowledge - 48 Citations

ISI Web of Science

Scopus - 81 Citations

Scientific classification

FOS: Engineering and technology > Electrical engineering, Electronic engineering, Information engineering

CORDIS: Technological sciences > Engineering > Electrical engineering

Other information

Authenticus ID: P-002-8F7

DOI: 10.1109/tpwrd.2012.2188143

Resumo (PT): This paper presents a new approach to incipient fault diagnosis in power transformers, based on the results of dissolved gas analysis. A set of autoassociative neural networks or autoencoders are trained, so that each becomes tuned with a particular fault mode or no fault condition. The scarce data available forms clusters that are densified using an Information Theoretic Mean Shift algorithm, so that training becomes efficient. Then, a parallel model is built where the autoencoders compete with one another when a new input vector is entered and the closest recognition is taken as the diagnosis sought. A remarkable accuracy of 100% is achieved with this architecture, in a validation data set using all real information available.

Abstract (EN): This paper presents a new approach to incipient fault diagnosis in power transformers, based on the results of dissolved gas analysis. A set of autoassociative neural networks or autoencoders is trained, so that each becomes tuned with a particular fault mode or no fault condition. The scarce data available forms clusters that are densified using an Information Theoretic Mean Shift algorithm, allowing all real data to be used in the validation process. Then, a parallel model is built where the autoencoders compete with one another when a new input vector is entered and the closest recognition is taken as the diagnosis sought. A remarkable accuracy of 100% is achieved with this architecture, in a validation data set using all real information available.

Language: English

Type (Professor's evaluation): Scientific

No. of pages: 8