Estimation of the Ageing Condition of Oil-Filled Transformers Based on the Oil Parameters Using a Novel Fuzzy Logic Algorithm

The condition of the transformer insulation determines the remaining life of the transformer. Over the transformer service life, both liquid and solid insulation undergoes a continuous aging process under electrical, chemical, mechanical and thermal stresses. The insulating liquid of a transformer can be reconditioned or replaced; however, this is not the case for the cellulose insulation. Therefore, the condition of the paper insulation is the major factor for determining the aging status of a transformer. To assess the paper condition, the common method is to measure the degree of polymerization (DP) of the paper insulation as a significant parameter. This method is destructive as it requires a sample of paper from inside of the transformer. Therefore, it could not be applied for transformers in operation. There are several approaches to approximate the DP value without direct measurements of paper samples. This research presents an improved method based on a fuzzy logic system for the estimation of the DP based on dissolved gases and chemical parameters of the liquid insulation. The algorithm is developed to create the rules for the use of fuzzy sets based on the information gain extracted by the entropy of the laboratory measurement data. This algorithm employs entropy to examine the sample homogeneity. Entropy is a measure of information theory that can determine the dataset characteristics concerning impurity and homogeneity. The algorithm uses fuzzy sets of oil parameters like Acidity, Interfacial Tension (IFT), Carbon Dioxide (CO2) and Carbon Monoxide (CO) and the breakdown voltage (BDV) for determination of the DP value.