Study On Simplizing Calibration Method Of Multiple Liner Regression In Near Infrared Analysis

Near infrared (NIR) analytical technique is a fast non-invasive analytical technique. In this dissertation, the analytical principles of NIR diffusion and its regular mathematical models are introduced, and some key factors influencing NIR analysis are discussed. According to the problem of lacking calibration samples practically, a calibration method based on combinating spectra is put forward. Such method is used to reduce the degree of freedom of regression equations. When using such method to predict the protein content of maize and barley samples, under the condition of lacking calibration samples, an ideal result is accomplished.