Optimization of Wheat Protein Near-Infrared Calibration Model Based on SPXY

It is very crucial that a representative training set can be extracted from a pool of real samples. In this paper, a representative set of correction samples of wheat protein content is selected by using SPXY algorithm firstly; Secondly, the spectral data is pretreated to enhance spectral features; Thirdly, the model of wheat grain protein is established by using partial least squares regression. The results show that the model established by the calibration set selected by SPXY is better than the model established by the calibration set selected randomly. Root mean square error of prediction(RMSEP) and prediction correlation coefficient(R) are 0.41094 and 0.97705 respectively, which are similar to the model established by the initial calibration set.