Classification of Insulating Liquids Thermal Treatment Using Infrared Spectroscopy and Multivariate Statistical Method

Abstract

Paper is focused on the investigation of the possibility of using statistical methods as discriminant analysis and principal component analysis for a classification of the aging of insulating material. The Statistical method was used in combination with Fourier Transform Infrared Spectroscopy results (FT-IR). The investigation was carried out on thermally aged mineral oil samples. The tested oil was exposed to thermal treatment at a temperature of 130°C for times ranging from 24 to 572 hours. Infrared spectra were measured at predefined times of aging using transmission technique in the middle area of the infrared electromagnetic spectrum. Program TQ Analyst was used for other evaluation of changes in measured infrared spectra and the possibility of their separation according to aging times. Based on the experiment performed, discriminability of individual levels of aging based on infrared spectra (IR) measurements was identified even in cases of minimum spectra differences. Combination of FT-IR measurement together with the application of multivariate statistical methods seems to be promising in possible application to aging level identification and automatization of the process of evaluation.