基于可见/近红外光谱与SIMCA和PLS-DA的脐橙品种识别

Application of visible and near infrared spectroscopy to identification of navel orange varieties using SIMCA and PLS-DA methods

脐橙的优良品种选育是通过芽变选种获得,为了选择一些具有特殊性状的脐橙芽变进行培植,需要对脐橙的品种进行鉴别。该研究应用可见/近红外光谱分析方法结合软独立模式分类(SIMCA)和偏最小二乘判别分析(PLS-DA)模式识别方法对赣南脐橙的品种进行识别。研究结果表明,采用原始近红外光谱结合SIMCA方法,实现了纽贺尔、奈弗宁娜、华脐以及朋娜4种脐橙的100%的识别;应用近红外光谱结合PLS-DA方法对校正样本建立判别模型,其校正及验证结果与实际分类变量的相关系数均大于0.970,交叉验证均方根误差(RMSECV)和预测均方根误差(RMSEP)都小于0.100,利用模型对验证集中纽贺尔、奈弗宁娜、华脐以及朋娜4种脐橙的识别率均为100%。为脐橙优良品种的选育提供快速鉴别分析方法。

The superior varieties breeding for navel orange is completed by the selection of bud mutation,and it is necessary to identify navel orange varieties in order to select some specialized character varieties for the cultivation. In this study,near infrared spectroscopy （NIRS） coupled with Soft Independent modeling of class analogy （SIMCA） and Partial Least Squares -Discriminant Analysis （PLS-DA） pattern recognition methods were used to identify the navel orange varieties. The results showed that four navel orange varieties were well identified by raw NIR spectra and SIMCA models,the identification rates were all 100%. For PLS-DA models,the correlations between the predicted category variables of calibration or validation and the measured category variables were all remarkable with a correlation coefficient （r） over 0.970 and low RMSECV and RMSEP （0.100）; The discrimination accuracy for the navel orange varieties was 100% by PLS-DA model based on the validation set of samples. It is suggested that Vis/NIR spectroscopy coupled with SIMCA and PLS-DA methods can be used for rapid detection of navel orange in superior varieties breeding.