Classification study of composite insulator chemical formulations based on laser-induced breakdown spectroscopy

High-temperature vulcanized silicone rubber insulators are widely used in high-voltage transmission lines and substation equipment because of their excellent fouling resistance. The aging of silicone rubber occurs due to environmental factors as its operating life advances and is closely related to its production formulation. Establishing a classification method for insulator chemical formulations is beneficial regarding operational performance and proposing a more reasonable operating scheme for composite insulators. However, the silicone rubber production formulations of various manufacturers are usually confidential, and there is no technology that can quickly identify insulator chemical formulations. In this study, a rapid classification method for composite insulator chemical formulations was investigated based on laser-induced breakdown spectroscopy (LIBS). Returned insulators from seven different manufacturers are considered as the research object. Spectral data were collected separately, and the insulators were initially classified using the characteristic spectral lines of the elements corresponding to different fillers doped in silicone rubber. The spectral data were characterized using a recursive feature elimination algorithm for feature selection and a linear discriminant algorithm for data dimensionality reduction in the characteristic spectral data. Accordingly, the insulators were classified using a backpropagation neural network algorithm, achieving the best experimental results with an accuracy of about 95%. Thus, LIBS technology can achieve rapid classification of insulator chemical formulations, which can provide a basis for formulation correlation for the operation and maintenance of composite insulators, as well as guarantee the safety of transmission lines.