부산대학교 도서관

In this paper the application of pattern recognition techniques as a tool for diagnostics of ethylene-propylene diene monomer (EPDM) polymer insulators is discussed. Assessment of degradation is carried out by taking small samples of insulator shed material in-situ with subsequent analysis of the samples by Fourier transform infrared spectroscopy (FTIR) method. Chemical indicators of insulation condition obtained from FTIR are fed to a pattern recognition program, cluster analysis, which is shown to successfully discriminate between insulators from different locations, manufacturers and with varying degree of degradation. A database linking the numerical indicators of insulation condition and type, age and location of insulators has also been developed.