The Study On The Determination Of Grain Collection By Near-infrared Spectroscopy

In order to cut down the consuming time of detecting collection quality of wheat such as moisture content, protein content, volume weight, falling number, ash content, gluten content by near infrared spectroscopy, the thesis carried through such study and discuss as the follows: Study the effect of different pretreatments and regression techniques on the calibration results. Discuss the possibility of determination of wheat collection quality such as moisture content, protein content, volume weight, falling number, ash content, gluten content by XDS Near-Infrared Spectroscopy and Portable Near-Infrared Grain Spectroscopy. Establish their predicted mathematic models. Choose the samples which independent of the calibration samples to test these predicted models. The result showed that:(1) With the XDS Near-Infrared Spectroscopy, the modified partial least square (MPLS) were the best regression to establish moisture content, protein content, falling number, ash content, wet gluten content calibration models, the partial least square(PLS)were the best regression to establish volume weight and dry gluten content calibration models.(2) With the Portable Near-Infrared Grain Spectroscopy, the partial least square (PLS) were the best regression to establish moisture content, protein content, volume weight, falling number, ash content, gluten content calibration models.(3) With the calibration models of moisture content, protein content, volume weight, falling number, ash content, gluten content by XDS Near-Infrared Spectroscopy, the standard error of cross-validation (SECV) were0.162,0.152,7.267,47.233,0.078,2.021,0.959; the models determination coefficient of correlation (RSQ) were0.979,0.984,0.860,0.846,0.825,0.796,0.755; the square error of prediction (SEP) were0.222,0.188,20.322,70.211,0.119,2.743,1.211. T test value between the chemical standard methods and NIRS method were0.261,0.304,1.235,0.030,0.005,0.364,0.119(P<0.05), this showed that the two methods had not statistic difference. This NIRS method can be applied to the collection quality of wheat.(4) With the calibration models of moisture content, protein content, volume weight, falling number, ash content, gluten content by Portable Near-Infrared Grain Spectroscopy, the standard error of cross-validation (SECV) were0.282,0.480,16.943,60.282,0.110,2.956,1.339; the models determination coefficient of correlation (RSQ) were0.985,0.988,0.833,0.906,0.871,0.937,0.946; the square error of prediction (SEP) were0.217,0.136,8.642,41.332,0.097,2.343,1.121. T test value between the chemical standard methods and NIRS method were0.261,0.304,1.235,0.030,0.005,0.364,0.119(P<0.05), this showed that the two methods had not statistic difference. This NIRS method can be applied to the collection quality of wheat.(5) We can known that NIRS method can be applied to the collection quality of wheat such as moisture content, protein content, volume weight, falling number, ash content, gluten content from seven different models established by two instruments. The results of Portable Near-Infrared Grain Spectroscopy near to XDS Near-Infrared Spectroscopy. In consideration of cost and volume, Portable Near-Infrared Grain Spectroscopy can be applied to the collection quality of wheat.