NREL researchers have been actively developing rapid calibration models to predict the composition of a variety of biomass feedstocks by correlating near-infrared spectroscopic data to compositional data produced using traditional wet chemical analysis techniques. The rapid calibration models are developed using multivariate statistical analysis of the spectroscopic and wet chemical data. This work will present an overview of the latest version of these calibration models for corn stover feedstock and dilute acid pretreated corn stover.

Specific topics to be addressed include a comparison of the effect(s) of mathematical pretreatments on the adequacy of the resulting models. We will demonstrate that calibration models developed using a variety of mathematical pretreatments provide essentially equivalent prediction models.  Also, we will present a comparison of partial least squares models predicting multiple dependent variables simultaneously (PLS-2)  and multiple PLS models, each prediction a single dependent variable (PLS-1) for groups of constituents. Finally, we will present a comparison of different measures of uncertainty among different multivariate analysis packages.