PREDICTING FURFURAL CHANGES IN POWER TRANSFORMERS DUE TO SHORT-CIRCUIT CURRENTS

Abstract

Power transformers are among the main equipment for transferring electrical energy from production fields to consumption fields. To ensure the health of the cellulose insulation of these equipment, oil sampling was performed and the amount of furfural in the oil was determined by liquid chromatography.In this study, the prediction of the amount of furfural changes in power transformer oil due to the passage of short-circuit currents and the age of the power transformers at the time of the fault has been addressed. For this purpose, 18 power transformers were investigated over a period of 11 years and the number and intensity of short-circuit currents, the age of the power transformer during the fault, and the changes in furfural in each power transformer oil sampling period were collected. By statistical analysis and comparison of machine learning models, the best model for predicting furfural changes was obtained based on the highest coefficient of determination and the lowest mean absolute error and root mean square error. The accuracy of this model was proven by examining two 230 kV and 160 MVA power transformers with different lifetimes, as well as applying a short circuit and measuring the amount of furfural in the oil of a 400 kV and 315 MVA power autotransformer in a high-voltage substation. In addition to creating a new framework for assessing the condition of transformers, this study allows analysts to predict the amount of furfural in power transformer oil without the need for sampling and testing.