Application of principal components analysis and signal-to-noise ratio for calibration of spectrophotometric analysers of food

Spectrophotometric analysers of food, being instruments for determination of the composition of food products and ingredients, are today of growing importance for food industry, as well as for food distributors and consumers. Their metrological performance significantly depends on the numerical performance of available means for spectrophotometric data processing; in particular - the means for calibration of analysers. In this paper, a new algorithm for this purpose is proposed, viz. the algorithm using principal components analysis. It is almost as efficient as partial least squares algorithms of calibration, but much simpler. It is fully automatic, viz. the selection of the most informative components is based on the signal-to-noise ratio characterising processed measurement data. The practical effectiveness of the proposed algorithm is demonstrated on a test problem consisting in determination of the concentrations of components of trinary oil mixtures.