Abstract
The aim of this study was to develop and validate a multivariate calibration method based on partial least squares (PLS) and near infrared spectroscopy for determining the interfacial tension of insulating oils used in electricity power transformers. Forty-eighty oil samples were obtained from an electricity power company and were divided into a calibration set of 38 leaving 10 for the validation set. The reference values were measured with a ring tensiometer and ranged from 15 dyn cm−1 to 45 dyn cm−1. The spectra were registered from 1280 nm to 2500 nm and the best PLS model was selected with 10 latent variables, providing a root mean square error of prediction of 2 dyn cm−1 and a coefficient of determination between reference and predicted values of 0.962. This method was validated in the context of multivariate calibration and was considered to have adequate accuracy (prediction errors between −9% and 12%), precision and linearity. A bias test verified the absence of systematic errors in the model. A pseudo-univariate calibration curve was constructed based on the concept of net analyte signal. The proposed method was simple, rapid, non-destructive and considered appropriate for the routine use in a quality control laboratory of an electricity power company.
© 2011 IM Publications LLP
PDF Article
More Like This
Cited By
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
Contact your librarian or system administrator
or
Login to access Optica Member Subscription