Prediction of the Insulating Paper State of Power TransformersUsing Artificial Neural Network

Abstract

Power transformers are considered the heart of power systems. The malfunction or undesirable outage of the power transformer will cause a tremendous revenue loss for the utilities. Therefore, a regular or preventive test must be accomplished on the transformer to check its state. Some standards, such as the American Transformer Diagnosis Guide and the American Society for Testing and Materials, have instructions for testing the transformers. The current works addressed which tests can be accomplished to predict the insulating paper state, which is the indicator of transformer aging. Furthermore, ANN model will be constructed to use it as a prediction tool of the paper state when the water content (WC), acidity (ACI), interfacial tension (IFT), oil color (OC), and 2-furfuraldehyde (2-FAL) were known. The ANN results indicated that the ANN's prediction accuracy was 93.87%.