近红外光谱法快速鉴定油菜杂交种的纯度

建立近红外光谱技术测定油菜杂交种纯度的方法。考察了样品杯类型、光谱预处理方法和波长范围对近红外模型预测性能的影响。结果发现,由不同样品杯采集近红外光谱所建立的校正模型,其预测性能存在较大的差异,旋转杯明显优于安瓿瓶;采用消除常数偏移量对光谱进行预处理能有效地提取光谱信息,选择5 000~8 000 cm–1波数范围作为建模谱区,其包含的有效信息率最高。在最佳条件下建立油菜杂交种纯度的校正模型,其决定系数(R2)为0.980 0,交互验证均方根误差(RMSECV)为0.008 59。利用该模型对预测集进行测定,预期均方根误差(RMSEP)为0.007 59,表明该模型具有很好的预测性能,近红外光谱法用于杂交种纯度的鉴定是可行的。

Determination of Seed Purity for Rape Hybrids by Near Infrared Spectroscopy

A new method was developed for determination of the purity of rape hybrids by near infrared spectroscopy(NIRS). The influence of sample cup types, spectrum pretreatment methods and frequency ranges on the quality of model was studied. The result was discovered that the performance of model was better built using rotating cup than ampoule. The spectra information was able to be exacted thoroughly by the pretreatment method of linear offset subtraction. The wave number range of 5000-8000 cm-1 as model spectrum region included the most effective information. The calibration model of the purity of rape hybrids was built under the optimal conditions. The internal cross coefficient(R2) of determination for the calibration set reached 0.9800, the root mean square error of cross validation(RMSECV) was 0.00859. The root mean square error of prediction(RMSEP) was 0.00759, which indicated that the method had a good prediction performance. Near infrared spectroscopy could be used in determination of the purity of rape hybrids.