Our focus in this thesis was evaluation of nutritional value of mixed rations for dairy cows using near infrared reflectance spectroscopy. Our goal was to establish if this physical method has a potential to evaluate composition of mixed rations. The mixed rations were prepared in a laboratory using dried feed samples. The nutritional values of mixed ratios were presented in tables. The predictive equations were developed by the use of WinISI computer program. We have tested the accuracy of predictive equations using validation sample set. The results were presented in tabels and gaphically. Parameters of calibration equations for feed ingredients (crude protein, crude fiber, raw ash, raw fats) and energy value (metabolizable energy, net energy for lactation) were excellent (R2> 0.99), however, validation with an independent validation set of samples showed that equations are not robust enough. It was suggested that the predictive equations should be improved with a greater number of more variable mixed ratios.
|
