针对茶叶中的农药残留问题,利用表面增强拉曼光谱(SERS)技术结合二维相关光谱法快速检测茶叶中毒死蜱残留。以金纳米为增强基底,采集含不同浓度毒死蜱残留茶叶样本的SERS,利用标准正态变量变换(SNV)对原始拉曼光谱进行预处理,再以毒死蜱浓度为外扰,进行二维相关同步光谱和自相关谱分析,筛选出与毒死蜱浓度变化相关的特征谱峰,利用灰狼算法(GWO)优化支持向量机(SVM)参数,建立茶叶中毒死蜱残留分析模型,并与偏最小二乘(PLS)模型得到的结果进行比较。结果表明:利用二维相关光谱法优选出毒死蜱的14个特征谱峰,所建SVM模型对预测集样本的决定系数 Rp2为0.98,方均根误差为1.32,相对分析误差为6.32,能用于茶叶中毒死蜱残留的实际估测,模型性能优于采用1096 cm ^-1单个特征谱峰建立的SVM模型和PLS模型。研究结果表明:将二维相关光谱法用于筛选与茶叶中毒死蜱浓度相关的特征谱峰是可行的,为拉曼光谱中特征变量优选提供了新思路;同时也表明,SERS结合二维相关光谱法可以实现茶叶中毒死蜱残留的快速检测,为茶叶农药残留快速检测装置的开发提供了方法支持。

In this study, surface-enhanced Raman spectroscopy (SERS) combined with two-dimensional correlation spectroscopy is used to develop a quantitative analysis model for rapidly detecting chlorpyrifos pesticide residues in tea. First, using gold colloid as the enhanced substrate, the spectral data of chlorpyrifos residues in tea samples with different concentrations are collected via SERS. Then, the original Raman spectra are pretreated using standard normal variate transformation (SNV). The chlorpyrifos concentration is considered as the disturbance and the characteristic peaks of chlorpyrifos are screened out via two-dimensional correlation synchronous spectrum and autocorrelation spectrum analysis. Parameters of the support vector machine (SVM) algorithm are optimized using the gray wolf algorithm (GWO), and the optimized SVM model is used for analyzing the chlorpyrifos residues in tea. The performance of optimized SVM model is compared to that of the model based on partial least squares (PLS). Results show that 14 chlorpyrifos characteristic peaks are screened using the two-dimensional correlation spectroscopy and the determination coefficient ( Rp2) of the proposed SVM model in the prediction set is 0.98, the root mean square error of prediction (RMSEP) is 1.32, and the relative prediction deviation (RPD) is 6.32. These values indicate that the developed model can be used for the actual estimation of chlorpyrifos pesticide residues in tea and performs better than the SVM model based on the 1096-cm 1 feature peak and PLS model. Thus, two-dimensional correlation spectroscopy is suitable for screening characteristic peaks related to chlorpyrifos concentrations in tea. This finding leads to a new idea for optimizing the characteristic variables in Raman spectroscopy. Results also show that SERS combined with two-dimensional correlation spectroscopy can rapidly and accurately detect chlorpyrifos pesticide residues in tea. The proposed method will provide methodological support for the development of rapid detection devices for analyzing pesticide residues in tea.