Machine Learning Based Prediction of Transformer Health Index and Remaining Life Using Transformer Oil Parameters
TL;DR
A machine learning-based approach for predicting the Transformer Health Index (HI) and the Remaining Life (RL) by means of the diagnostic parameters of transformer oil is described, allowing for timely planning of maintenance, reducing the occurrence of unexpected transformer failures, extending the service life of the assets and improving the overall efficiency of asset management for transformers in modern power systems.
Abstract
Power transformers are important components of electrical power systems and it is essential that they operate reliably if continuous power supply is to be maintained. Because they are constantly subjected to electrical, thermal, mechanical, and environmental stresses, the insulation suffers degradation, a fact that can be seen in the physicochemical properties and in the characteristics of the dissolved gases in the transformer oil. Traditional methods of assessing condition depend on periodic testing of the oil and on the judgement of experts, a procedure which is time-consuming and open to personal bias. The present paper describes a machine learning-based approach for predicting the Transformer Health Index (HI) and the Remaining Life (RL) by means of the diagnostic parameters of transformer oil. The model used in this study is based on a dataset consisting of 470 samples of transformer oil and including 14 diagnostic features, namely the concentrations of dissolved gases (H₂, O₂, N₂, CH₄, CO, CO₂, C₂H₄, C₂H₆ and C₂H₂), Dibenzyl Disulfide (DBDS), power factor, interfacial voltage, dielectric rigidity and water content. Following data preprocessing and feature engineering, Random Forest Regression is applied in order to identify the complex nonlinear relationships between the oil parameters and the condition of the transformer. The model's performance is assessed using the Mean Absolute Error (MAE), the Root Mean Square Error (RMSE) and the coefficient of determination (R²). The results of the experiments show that the proposed framework is able to make accurate predictions of both the Health Index and the Remaining Life, thus allowing a reliable evaluation of the transformer's condition. The method developed supports predictive maintenance by allowing for timely planning of maintenance, reducing the occurrence of unexpected transformer failures, extending the service life of the assets and improving the overall efficiency of asset management for transformers in modern power systems.