Impulse fault diagnosis in power transformers using self-organising map and learning vector quantisation

An artificial intelligence approach is proposed to an impulse fault diagnos is problem in oil-filled power transformers. The experiment focuses on the distinction between the effects caused by faults of a different nature and the different physical location of occurrences in a transformer winding. Th e proposed method involves an artificial neural network-based pattern recog nition technique, to recognise the frequency responses of the winding admit tance of atypical high-voltage transformer under healthy and different faul ty conditions of winding insulation. It attempts to establish a correlation between the nature and site of the internal insulation fault and its assoc iated frequency response. A self-organising neural network model has been e mployed as the basic pattern recogniser, to discover the significant patter ns and to extract the hidden information from a set of frequency response p atterns obtained from an EMTP model of the transformer with artificially si mulated faults. A learning vector quantisation-based classification techniq ue has been applied to efficiently classify visually indistinguishable resp onse patterns. The method applied to a winding model of a high-voltage tran sformer, with tap changer winding, exhibited high diagnostic accuracy by su ccessful detection and discrimination of faults of a different nature and s ite of occurrence.