LAUSR

Significant advances achieved in different sensor technologies and computer processing data have made possible to respond the needs of livestock sector, providing precise and rapid information on feed composition, being an alternative to real time quality control on compound feed the use of handheld NIRS sensors. This work aimed to evaluate two hand-held portable NIR spectrophotometers for on-site and real time analysis of nutritive parameters in raw compound feed: Phazir 1624 Polychromix Inc (PhIR) and MicroNIRTM 1700 by JDSU (MICRO). For computing data, different combinations of pre-treatments and multivariate statistical methods have been assayed to extract the valuable information of spectra data and to develop appropriate calibrations. The calibration models displayed greatest predictive capacity for Crude Protein (CP), Crude Fiber (CF) and Starch (STCH) and the determination coefficients of cross validation were 0.90-0.88 for CP, 0.85-0.91 for CF, 0.89-0.88 and 0.89-0.91 for STCH using PhIR and MICRO instruments respectively. Dry Matter showed the lowest determination coefficients of cross validation 0.67-0.73. Accuracy achieved 99-101% for both NIRS instruments and no differences were found when applying tstudent-test comparing reference and predicted data. Results obtained with both instruments were compared by using standard deviation and not significant differences were observed at the 5% level. Results so far have demonstrated the potential of these handheld NIRS instruments proposed here to estimate the individual compound feeds composition changes at farms level instantly, time avoiding the disadvantage of moving the samples to the lab.