Pattern recognition of the factors affecting the reproducibility of FRA measurements in power transformers

J.L. Velásquez Contreras, M.A. Sanz-Bobi, W. Koltunowicz, A. Alexander , M. Heindl

Conference on Condition Monitoring and Diagnosis 2010 - CMD2010, Tokyo (Japan). 06-11 Septiembre 2010

Summary:
Factors affecting the reproducibility could lead to a misleading assessment of FRA results. The identification failure modes and reproducibility problems form part of the assessment tasks carried out by human experts. There is a necessity of representing human knowledge for supporting the assessments of FRA result by non-experts. This paper is focused on the factors affecting the reproducibility. First an overview to such factors is presented and exemplified with real case studies. Then the application of decision trees as pattern recognition algorithm for knowledge representation is illustrated. The obtained results are very promising and show the suitability of using Artificial Intelligence techniques in the field of FRA.

Keywords: FRA; reproducibility; assessment; transformers

Publication date: September 2010.

Citation:
Velásquez Contreras, J.L., Sanz-Bobi, M.A., Koltunowicz, W., Alexander , A., Heindl, M., Pattern recognition of the factors affecting the reproducibility of FRA measurements in power transformers, Conference on Condition Monitoring and Diagnosis 2010 - CMD2010, Tokyo (Japan). 06-11 September 2010.

IIT-11-033A