Losses Prediction in the Distribution Transformer Using Hierarchy Neural Networks

Abstract

This study presents Hierarchy Neural Networks (HNN) for losses prediction in 3-phase distribution transformer sized 100 kVA 22kV/400-230V. The transformer parameters as the follows: core loss, copper loss and windings resistance at the several temperature levels are collected from the transformer testing process in the factory. These parameters are used as training sets for the HNN. The primary and secondary winding resistances and varied temperature levels are applied as inputs of the HNN in order to obtain the transformer losses as the HNN output. The testing results show that HNN can provide the power losses with accurate in some degree when compared with the measuring results.