Determinação da tensão interfacial de óleos isolantes usando análise de imagens e espectropia no infravermelho próximo

Abstract

Power transformers are of great importance in the transmission and distribution of electrical energy. One of their most important parts is the insulating system, consisting of Kraft paper immersed in insulating oil. One of the parameters used for evaluating the degradation of this system is the oil interfacial tension. The aim of this study was to determine the interfacial tension in samples of insulating oils by using Near Infrared Spectroscopy (NIRS) and Image Analysis, combined with firstorder (PLS, partial least squares) and second-order (PARAFAC, parallel factor analysis, and NPLS, N-way PLS) calibration methods. Forty eight oil samples were obtained, whose values of interfacial tension had already been determined. These samples were divided in calibration (38) and validation (10) sets, and NIR spectra were recorded from 780 to 2500 nm. The best PLS model was built using the spectral range from 1360 to 1430 nm and eight latent variables, providing a root mean square error of prediction (RMSEP) of 2.1 dyne/cm. Scanner images were also obtained for these same sets of samples, converted for gray-scale and stacked to three-way data arrays, before applying PARAFAC and NPLS. The best PARAFAC model was built with six factors and provided a RMSEP of 3.9 dyne/cm, while the best NPLS model was obtained with three latent variables and provided a RMSEP of 3.0 dyne/cm. NIRS-PLS was considered the best model, providing individual prevision errors between -7.9 and 10.2%, which were acceptable for electric energy companies. The proposed methods are rapid and non-destructive, showing great advantages over the traditional ones, which are slow and produce chemical waste. This work opens perspectives for developing an on-line monitoring of the insulating transformer systems.