Avaliation of figures of merit in multivariate calibration in the determination of quality control parameters in alcohol industry by near infrared spectroscopy / Avaliação de figuras de merito em calibração multivariada na determinação de parametros de controle de qualidade em industria alcooleira por espectroscopia no infravermelho proximo

Patricia Valderrama

Soluble solids (Brix), sucrose (Pol) and Reducing Sugar (RS) are important properties in the quality control of alcohol industry to determine grower payment. Thus, the validation was achieved through determination of figures of merit for multivariate calibration models using near infrared spectroscopy (NIR) in region of 1100 - 2500 nm by partial least square regression (PLS). The region of 1600 - 1850 nm corresponds to variables selection by interval partial least square regression (iPLS). A total of 1003 and 378 samples constitute the calibration and validation sets, respectively, divided by Kennard-Stone algorithm. The calibration set was optimized by outliers elimination based on data with extreme leverage, unmodelled residuals in spectral data and unmodelled residuals in the dependent variable. For validation, besides the leverage and unmodelled residuals in spectral data was also evaluated residuals based on in spectral repeatability. The figures of merit such as accuracy, precision, sensitivity, analytical sensitivity, selectivity, adjust, signal-to-noise ratio, limits of detection and of quantification and confidence limit were calculated. The results obtained ondicate that the models developed can be used in the alcohol industry as an alternative to refractometry and lead clarification to polarization measurements (standard methods for Brix and Pol, respectively). For RS it is necessary to have evaluation by the industry and growers, to model NIR performed in relation to the standard method of titration and the estimate currently made by the industry.

AUTOR(ES)

DATA DE PUBLICAÇÃO

RESUMO

ASSUNTO(S)

ACESSO AO ARTIGO

Documentos Relacionados