Development and validation of NIRS calibrations for the analysis of additives in animal nutrition products

Abstract

This final degree project focuses on the application of Near-Infrared Reflectance Spectroscopy (NIRS) as an alternative analytical technique to traditional methods for the analysis of additives intended for animal nutrition. It is a fast and non-destructive analytical technique that is becoming increasingly relevant in this industrial sector. The main objective of this project is to design multivariate calibration models using NIRS, applying the mathematical methods mPLS and PCA, for the determination and quantification of various parameters, such as moisture and antioxidant content. To achieve this objective, a robust database was built by analyzing the maximum number of available samples and assigning to each sample the reference values obtained using traditional analytical techniques, such as coulometric titration for moisture determination using Karl Fischer and gas chromatography (GC) for the quantification of different types of antioxidants. Once all data were entered into the system, predictive models were developed using specific software, with the aim of correlating the NIRS spectra obtained from the analysis of the samples with their corresponding reference values. Subsequently, the models were validated through the analysis of new samples, assessing their accuracy, precision, and robustness. Deviations were also identified and corrected to improve the prediction model. The results obtained showed a high correlation (R² > 0.95) and low prediction errors, lower than those associated with traditional analytical methods, both in the determination of moisture and in the quantification of the two types of antioxidants, demonstrate the viability of NIRS as an alternative analytical method