Identification of Power Transformer Insulating Paper's State Based on Principal Component Analysis.

The power transformer is a vital element in power system assets. When failure occurs in transformers, it leads to more revenue loss for the electrical utilities. Power transformer failures may occur due to high stresses on the insulating systems (oil and paper). These stresses categorize as electrical, thermal, and mechanical stresses. The main contribution of the current work is to increase the contribution of essential test parameters, which contribute to determining the insulating transformer state. In this work, principal component analysis (PCA) is not used as a prediction or classification tool. Still, it preserves only the necessary information and marginalizes or reduces the role of parameters with fewer contributions. PCA is a statistical method to compress an extensive data set of variables to a smaller number of orthogonal factors unrelated to interpreting the correlation matrix. One hundred forty-seven data samples were collected from the Saudi Electricity Company and used in this study. The data include the dissolved gas analysis (DGA) test, the breakdown voltage (BDV), moisture content (MC), Acidity (AC), interfacial tension (IFT), oil color (OC), dissipation factor (DF) or tan □, furan content (2-FAL). Based on PCA, some test variables can be selected as indicators for the insulating paper state from more test variables groups, reducing the cost of all tests. [ABSTRACT FROM AUTHOR]