Detection of Inter-Turn Short-Circuits in Power Transformers Using a Neuronal Network Model

GUSTAVO LEAL; MARIANO DE PAULA; MATIAS MEIRA; GUILLERMO BOSSIO; CARLOS VERUCCHI

Oberá, Misiones

Congreso; XX Reunión de Trabajo en Procesamiento de la Información y Control (RPIC 2023); 2023

IEEE

Inter-turn short-circuits in windings are one of the most frequent faults in transformers. Although there are offline techniques for their detection, the current trend is to move towards techniques that can be applied online. This work presents a strategy for online detection of inter-turn short-circuits based on differential current monitoring. A neural network-based model is used to estimate the differential currents of each phase under different power supply and load conditions. The neural network is trained with a fault-free transformer. The differential currents estimated by the neural network are compared with the measured ones, and if the error exceeds a certain threshold, the presence of a fault is assumed. Experimental validation in the laboratory is also presented.