树脂吸附结合近红外光谱同时检测小龙虾中铅、镉模型的建立

Establishment of a model for simultaneous determination of lead and cadmium in Procambarus clarkii by resin adsorption combined with near infrared spectroscopy

中文关键词:  树脂吸附  近红外光谱  小龙虾  铅  镉  预测模型

英文关键词:resin adsorption  near infrared spectroscopy  Procambarus clarkii  lead  cadmium  prediction model

中文摘要:

      目的 建立树脂吸附结合近红外光谱模型同时检测大批量小龙虾中铅、镉含量的方法。方法 小龙虾经微波消解后, 用D405大孔吸附树脂吸附小龙虾消解液中的铅、镉, 采集吸附树脂的近红外光谱, 并采用一阶导数、小波变换、标准正态变换和多元散射校正进行光谱预处理, 选取较佳预处理方法, 结合竞争自适应重加权采样法进行最优波段选择; 利用偏最小二乘法建立最优定量预测模型, 并对模型进行外部验证, 探究模型预测准确度; 收集6个地区的小龙虾对模型进行应用验证, 探究模型实际应用可靠性。结果 D405树脂对小龙虾消解液中铅、镉的吸附率均达98.5%以上。经小波变换光谱预处理, 结合波段选择, 建立的铅、镉定量模型预测准确度较高, 校正集交叉验证均方根误差和相关系数分别为0.08、0.12及0.98、0.95; 外部验证集的预测均方根误差和相关系数分别为0.07、0.10及0.98、0.98。模型实际应用可靠, 铅、镉含量参考值与预测值之间偏差的标准差和相关系数分别为0.01、0.01及0.99、0.98。结论 建立的小波变换-竞争自适应重加权采样-偏最小二乘定量模型对小龙虾样品中的铅、镉含量都具有更好的预测效果, 树脂吸附结合近红外光谱可以用于同时检测大批量小龙虾中的铅、镉。

英文摘要:

      Objective To establish a method for simultaneous determination of lead and cadmium in a large number of Procambarus clarkii by resin adsorption combined with near-infrared spectroscopy. Methods After digested by microwave, the lead and cadmium in the Procambarus clarkii digestion solution were adsorbed with D405 macroporous adsorption resin, and the near infrared spectra of the adsorbed resin were collected. The pretreatment method of spectra was optimized with first derivative, wavelet transform, standard normal variate, and multivariate scattering correction, respectively, while the suitable wavenumber ranges of spectra were selected by using competitive adaptive reweighted sampling. Then the optimal quantitative prediction model was established with partial least squares method, and the external verification of the model was carried out to explore the prediction accuracy of the model. Finally, the Procambarus clarkii from 6 regions were collected, while the application verification was carried out to explore the reliability of the practical application of the model. Results The adsorption rates of lead and cadmium in digestion solution of Procambarus clarkii with D405 resin were both more than 98.5%. After wavelet transform pretreatment and band selection, the quantitative models for lead and cadmium with high prediction accuracy were established, in which the root mean square error of cross validation of calibration set were 0.08 and 0.12, respectively, while the correlation coefficients of calibration were 0.98 and 0.95, respectively; the root mean square error of prediction of external validation set were 0.07 and 0.10, respectively, while the correlation coefficients of prediction were both 0.98. Moreover, the practical application of the established model had good degree of reliability since the standard error of deviation between the reference value and the predicted value of lead and cadmium were both 0.01, and correlation coefficient were 0.99 and 0.98, respectively. Conclusion The wavelet transform-competitive adaptive reweighted sampling-partial least squares quantitative model has better prediction effects on lead and cadmium content in Procambarus clarkii samples, the resin adsorption combined with near infrared spectroscopy may be used to simultaneously determine the content of lead and cadmium in large quantities of Procambarus clarkii.