Predicting Transformer Health Indices Through Oil Quality Parameters with Artificial Neural Networks (ANN)

• Published in: 2023 IEEE International Conference on Applied Electronics and Engineering, ICAEE 2023
• Main Author: Abd Aziz A.M.; Azmi M.A.; Izwan Mohd Supian M.F.; Talib M.A.; Abidin A.F.; Al Junid S.A.M.
• Format: Conference paper
• Language: English
• Published: Institute of Electrical and Electronics Engineers Inc. 2023
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85180529863&doi=10.1109%2fICAEE58583.2023.10331501&partnerID=40&md5=8762a15b20c5a345cb8360cc1cf02c2c

Power transformers require attentive maintenance and monitoring due to their high cost and their crucial role in the stability of the electricity grid. Predicting the condition of these transformers is a multi-faceted challenge, mainly due to the complicated calculations based on rankings and expert predictions in the industry. Against this background, this study presents an innovative approach to predicting condition indices for power transformers based on oil quality assessment using artificial neural networks (ANN). The study is divided into three main phases: Identification of the gap, development of the model and the subsequent phases of data collection, testing and validation. Our results show that the ANN-Bayesian Regularisation Model (BR) with 10 and 20 neurons emerges as the optimal choice for predicting the health index of transformers based on oil quality variables. This research culminates in the confirmation that ANN can provide an optimised prediction procedure that reduces the complexity of current industry practises for determining transformer health indices. © 2023 IEEE.