籼稻品质分析的近红外光谱模型建立及其应用研究

Study on the Establishment and Application of Near-Infrared Spectrum Model for Indica Rice Quality Analysis

为了满足籼稻品质快速分析的需求,本研究利用籼稻精米粉近红外光谱建立了直链淀粉含量、蛋白质含量、碱消值、垩白度的回归预测模型.结果表明,本研究提供的预测模型具有良好的测定效果,用偏最小二乘法(PLS)获得的籼稻精米粉直链淀粉含量、蛋白质含量、碱消值、垩白度的回归模型和交叉验证显示最优校正决定系数(R^2)和交叉检验均方误差(RMSECV)分别为0.9561、1.55,0.9510、0.258,0.9076、0.283,0.9014、4.14.说明所建的近红外光谱预测模型具有实用价值.

In order to keep the demand of indica rice. quality analysis rapidly, the regression prediction models of rice amylose content, protein content, allkali spreading value and chalkiness were established from near-infrared spectral scanning data of milled rice powder. The results showed that the prediction models obtained in this study were of real determination effect. The prediction models derived from the partial least squares （PLS） and cross-certification for rice amylose content, protein contents, alkali spreading value and chalkiness indicated that the optimal calibration determination coefficient （R2） and cross-examination mean square errors （RMSECV） were 0.9561, 1.55; 0.9510, 0.258; 0.9076, 0.283; 0.9014, 4.14. That approved the near-infrared spectrum prediction models are of easy usage in practice.