论文部分内容阅读
以100份中国核心大豆种质资源为材料,建立高效液相色谱法(HPLC)测定大豆中异黄酮5组分和总异黄酮。扫描大豆的近红外光谱,以傅里叶近红外光谱法(FT-NIRS)与HPLC技术相结合,采用偏最小二乘(PLS)回归和交叉验证法,探讨利用FT-NIRS技术预测异黄酮含量的可行性。总异黄酮近红外预测模型的内部交叉验证其校正决定系数(R2)和交叉检验均方误差(RMSECV)分别为0.8763和0.515,外部验证其校正决定系数(R2)和预测均方误差(RMSEP)则分别是0.9492和0.599。结果表明,利用FT-NIRS预测大豆中总异黄酮含量是可行的,但是各异黄酮组分的近红外模型不能达到准确预测要求。大豆异黄酮近红外模型的建立对今后大豆的异黄酮选育工作可以提供帮助。
Using 100 core soybean germplasm resources in China as materials, HPLC was used to determine isoflavone 5 and total isoflavone in soybean. The FT-NIRS was used to detect the isoflavone content by scanning near infrared spectroscopy (FT-NIRS) and the combination of HPLC and FT-NIRS. Partial least squares regression (PLS) and cross validation Feasibility. The calibration coefficients (R2) and cross-validation mean square error (RMSECV) were 0.8763 and 0.515 respectively for internal cross-validation of the total isoflavone NIR prediction model. External validation of the calibration determination coefficient (R2) and the prediction root mean square error (RMSEP) They are 0.9492 and 0.599 respectively. The results showed that it is feasible to predict the content of total isoflavones in soybean by FT-NIRS. However, the near infrared model of each isoflavone component can not achieve the accurate prediction requirement. The establishment of soybean isoflavone near-infrared model can help the future isoflavone breeding of soybean.