IRNA: AI Identifies Partial-Discharge Defects in Generator Windings with 84% Accuracy
Iranian researchers use Transformer and CNN models to locate defects and estimate discharge magnitude; study published in IEEE Transactions on Dielectrics and Electrical Insulation
IRNA reports Tehran University researchers used AI to locate partial-discharge defects in generator windings with ~84% accuracy (Transformer) and to estimate discharge magnitude (CNN), in a paper published in IEEE Transactions on Dielectrics and Electrical Insulation.
IRNA reports that researchers from Tehran University’s Electrical and Computer Engineering Department used artificial intelligence to locate insulation defects and estimate partial discharge in generator stator windings. They evaluated five AI models on two goals—identifying defect location and predicting discharge magnitude—using data generated from a finite-element simulation of a generator. The Transformer model achieved about 84% accuracy in locating the defect, while a CNN showed the lowest error in estimating discharge magnitude. The study, titled Deep Learning Approaches for Partial Discharge Localization and Apparent Charge Prediction in Generator Stator Windings, appears in IEEE Transactions on Dielectrics and Electrical Insulation, with a link to IEEE Xplore: https://ieeexplore.ieee.org/abstract/document/11268523