Comparison of modeling effect among rice from different regions with near infrared technology

The representative rice samples were selected from four provinces of Anhui, Jiangsu, Hunan and Heilongjiang. With the determination of its moisture content, crude protein content, fatty acid values and amylose content,scanning the near infrared spectra,establishing the quantitative analysis model by using partial least squares( PLS), comparing the different modeling results of same index of different regions to explore whether establishing the model of near infrared technology need to distinguish areas. The results showed that if the samples were representative enough,the modeling effect of same index between different areas exist certain differences based on reliable model built of different indexes. Four provinces moisture content modeling determination coefficient( R²) were 0.91, 0.97, 0.94, 0.85, crude protein content modeling determination coefficient(R²) were 0.94,0.78,0.87,0.72,fatty acid value modeling determination coefficient(R²)were 0.77,0.60,0.83,0.65,amylose content modeling determination coefficient(R²) were 0.68,0.58,0.69,0.53,respectively.