UV Spectrophotometer Based AI Techniques for Remnant Life Estimation of Power Transformers

Power transformers are critical equipment in the electrical power system. Diagnostic and condition monitoring is very important for testing of power transformer oil in order to find the remnant life and has to be done carefully. Mineral oil in transformer serves two purposes the insulating as well as cooling and it is inseparable material of dielectric insulation system. The oil gets contaminated mainly due to ageing. The contaminant which changes the oils chemical and physical properties are moisture, acids, metal particles, sludge, and other compounds which are present in cellulose and due to aging insulation of the cellulose. UV- Spectrophotometer response is a nonintrusive test used to determine the transformer integrity and the response can be measured instantly with relatively cheap equipment and there is no need for an expert person. The present paper introduces two approaches of AI techniques they viz. Fuzzy logic and Artificial neural network (ANN) to estimate the relationship between dissolve decay content and UV spectrophotometer response of transformer oil method to determine the remnant life of the transformer oil. These two methods use the UV spectrophotometer bandwidth and absorbance values of the transformer oil as the inputs which are in service at several locations.