On-Site Quality Control of Processed Land Animal Proteins Using a Portable Micro-Electro-Mechanical-Systems near Infrared Spectrometer

Abstract

This study was carried out using a spectral database comprising 1164 samples of processed land animal proteins provided by a rendering plant located in southern Spain (Seville). A variety of cloning algorithms and strategies were evaluated for the transfer of spectroscopic databases obtained on a FOSS NIRSystems monochromator (FNS), currently in use for at-line NIR analysis in the rendering-industry laboratory, to a MicroPhazir Rx portable micro-electro-mechanical-systems (MEMS)-based NIR spectrometer. Application of the spectral difference by wavelengths algorithm for spectral adjustment using a 10-sample standardisation set enabled standardisation of the two instruments and effective transfer of spectral data: spectral concordance values, as estimated by the RMS(c) statistic, were very similar at 10,829 and 10,388 μlog (1/ R) for the master (MEMS-NIR) and satellite instruments (FNS), respectively. Once the database had been transferred to the master, calibration equations were developed to predict moisture, crude protein, ash, crude fat, fat after hydrolysis, digestible protein, Ca and P, with a view to comparing the predictive ability of these equations with that of equations obtained using the satellite. Equations developed in the two instruments displayed similar accuracy and precision.