A new non-linear model for transformer oil residual operating time

The purpose of this paper is to implement non-linear multivariable regression technique (nlm) for the prediction of the residual operating time (ROT) of the transformer oil characteristics. This technique depends on non-linear models for each transformer oil characteristic and linear combination of these non-linear models. The characteristics under study are the oil breakdown voltage, total acidity and water content. The proposed technique has proved its accuracy and applicability by justifying the oil characteristics which have a strong impact on ROT and by giving more accurate predicted results than those obtained by previous modeling techniques.