Implementation of ANFIS and FIS-ID3 in Transformer Oil-Paper Assessment

Abstract

Power transformers play vital role in electrical power system, that makes it crucial to monitor its current condition. Measurements of oil-paper insulation health index through parameters like Furfural (2FAL) and Degree of Polymerization are keys to assess life of power transformer, despite the fact that those measurements are not included in routine test. This paper presents the use of Artificial Intelligence to do condition assessment of oil-paper insulation of power transformer through its dissolved gases and dielectric characteristics. As much as 182 in-service transformers are observed and were used as training set of the ANFIS and FIS-ID3 model, while 69 other transformers were used as testing dataset. Acidity, Interfacial tension, CO, and CO 2 are potential to become input parameters of the health index. The result shows that FIS-ID3 have some advantages over ANFIS that is worthily noted for the next development.