A Multi-Gene Symbolic Regression Approach of Determining Insulating Oil Interfacial Tension

This paper aims at developing the relationship between interfacial tension (IFT) and other oil parameters obtained through testing of insulating oil. Considering the importance of IFT in determining age of insulating oil, it is imperative to understand the inherent characteristics of IFT in assessing the healthiness of transformer and reactor in service. The unique study undertaken in this paper attempts to develop a model using multi gene symbolic regression to generate and evaluate the relationship between IFT and other oil parameters. Data utilized in this study involves actual oil testing parameters of transformer and reactor in service over more than a decade. A comparative study of the developed model reveals its superiority over support vector machine (SVM). This research shall help utility engineers in determining aging of the insulating oil and shall further assist in taking preventive action for replacement of the degraded insulating oil time ahead of failure of the equipment.