Power Transformer Insulation System Health Index with Missing Data Prediction using Random Forest

Health Index approach is currently one of the most common ways to assess the overall condition of power transformers. Data unavailability is still a problem in Health Index assessment. This paper discusses the prediction of transformer health conditions using five missing data replacement methods, which are removed parameter, average value, assume good, SLR, and Random Forest prediction. Seven scenarios based were simulated based on three missing parameters, namely 2FAL, IFT and Water Content. The accuracy is evaluated using the Health Index calculated with complete parameter. As much as 504 units of 150 kV power transformers were used in the analysis. The results show that Random Forest method produced the highest accuracy rate among the other methods with average value of 92%.