Evaluation of aging and degradation for silicone rubber composite insulator based on machine learning

Abstract Silicone rubbers (SIRs), as the main material of composite insulator sheds, have aging phenomenon for the long-term operation in the outdoor, which has an important impact on the performance degradation of composite insulator. Periodic examine and replacement of severely aged composite insulator is of great significance for the safe operation of power system. However, there is no clear standard to estimate the aging degree. In order to evaluate the degree of aging and degradation more accurately, the experimental test and analysis of the SIR sheds were carried out from three aspects: physical characteristics, chemical composition and electrical properties. The physical characteristic results show that the surface of the aging SIR will appear obvious holes and cracks, as well as hardening. Chemical composition results show that the internal composition of the continuous decomposition, while a large number of heavy metal elements accumulate in the pollution. The change of physical characteristics and chemical composition of SIR sheds may lead to the deterioration of electrical properties. According to the above results, the influence rule of measurement parameters and aging degree was obtained, and the key characteristic parameters of aging were summarized and extracted. Based on the characteristic parameters, a decision tree aging evaluation model is established for state evaluation. The accuracy of machine learning is 100% and 93.2% respectively. For the on-site application, early warning of moderately aged insulators and replacement of severely aged insulators are proposed in the on-site maintenance of SIR composite insulators.