Neural Network Computing for the Electric Power Industry

I. Fault Diagnosis

IMPULSE TEST FAULT DIAGNOSIS ON POWER TRANSFORMERS USING KOHONEN’S SELF-ORGANIZING NEURAL NETWORK

Thomas Baumann

Alain Germond

Ecole Polytechnique Fédérale de Lausanne

Laboratoire de Réseaux Electriques

CH-1015 Lausanne, Switzerland

Fax: [phone]

Daniel Tschudi

ABB Sécheron Ltd.

CH-1211 Geneva, Switzerland

Fax: + 41 22 738 73 05

ABSTRACT

The paper presents a method based on Kohonen’s self-organizing neural network to perform an impulse test fault diagnosis on power transformers. Digital signals obtained during transformer impulse testing are to be classified. In fact, if the model of a transformer is well known, a classifier could be trained off-line with all possible signals of simulated faults that can occur in ...