Diagnosing Faults in Power Transformers With Variational Autoencoder, Genetic Programming, and Neural Network

This work presents a new approach for the diagnosis of incipient faults in power transformers by considering dissolved gas analysis (DGA). A multilayer perceptron (MLP) neural network was trained to diagnose type transformer fault. For training and testing classifier, data were used from in-service obtained IEC TC 10 database other literature. To address imbalance adopted thus improve generalization augmentation technique based on variational autoencoder used. selection extraction characteristics inputs genetic programming (GP) is proposed, which allows creation n-dimensional space characteristics, providing greater ability increase interclass distances intraclass compaction. performance proposed comparisons made using classification results through 60599 conventional fault method MLPs without use extractor. The demonstrate applicability methodology diagnosis, with system obtaining an accuracy 95.18% test basis, higher than achieved methods perform comparison results.