【作者】 刘涛；

【导师】 刘燕德； 杨信廷；

【作者基本信息】 华东交通大学 ， 机械工程， 2011， 硕士

【摘要】 随着社会的进步和人民生活水平的提高,消费者在强调水果的外观和营养品质的同时,对其食用安全性提出了越来越高的要求。目前,常规的水果表面农药残留检测方法存在破坏样品、前处理繁琐、耗时长、成本高、对环境造成污染等不足,因此,探索一种简便、快速、绿色、环保的检测方法具有重要的实际应用价值。这对于扩大我国高档优质水果的出口,增加果农的收入,提高我国水果产业的国际竞争力有着重大的意义。拉曼光谱技术是一门基于拉曼散射效应而发展起来的光谱分析技术,体现的是分子的振动或转动信息。其拥有样品无需前处理、操作简便、耗时短、绿色、环保等优点,已广泛应用于石油化工、生物医学、地质考古、刑事司法、宝石鉴定等领域。论文以赣南脐橙作为研究对象,以毒死蜱和马拉硫磷作为分析测试指标,应用拉曼光谱技术对水果表面农药残留进行了快速检测方法与定量模型研究,取得的主要研究成果有:1.通过采集毒死蜱和马拉硫磷标样和赣南脐橙表皮的拉曼光谱,获得了毒死蜱和马拉硫磷以及赣南脐橙表皮的拉曼谱带归属和拉曼特征峰。2.分别对农药毒死蜱和马拉硫磷标准溶液和赣南脐橙表皮农药残留的拉曼光谱预处理方法进行了实验研究,获得了理想的拉曼光谱预处理方法。实验采集了不同基底（镜片和硅胶片）的毒死蜱和马拉硫磷溶液以及赣南脐橙表皮农药残留的拉曼光谱,对比不同的预处理方法,建立了基于偏最小二乘法的定量数学模型。研究表明:当以镜片作为基底时,线性基线校正预处理后毒死蜱溶液校正模型最为理想,位移校正预处理后马拉硫磷溶液校正模型最为理想;当以硅胶片作为基底时,二阶导数预处理后毒死蜱溶液校正模型最为理想,一阶导数预处理后马拉硫磷溶液校正模型最为理想。对于赣南脐橙表皮的毒死蜱残留,理想的预处理方法为一阶导数;而对于马拉硫磷残留,理想的预处理方法则为标准正态变量校正。3.建立了两种农药标准溶液的拉曼光谱快速定量数学模型。采用传统的峰强比值法和化学计量学算法对不同基底的毒死蜱和马拉硫磷溶液拉曼光谱数据建立校正模型,探讨了以镜片和硅胶片作为基底时采集农药毒死蜱和马拉硫磷的建模效果。研究表明:当以镜片作为基底时,利用多元线性回归建立的毒死蜱拉曼光谱检测模型最为理想,校正和交互验证结果为:r_c=0.960,RMSEC=3.166,r_cv=0.877,RMSECV=5.421;利用偏最小二乘法建立的马拉硫磷拉曼光谱检测模型最为理想,校正和交互验证结果为:r_c=0.998,RMSEC=0.748,r_cv=0.985,RMSECV=2.236。当以硅胶片作为基底时,利用偏最小二乘法建立的毒死蜱拉曼光谱检测模型最为理想,校正和交互验证结果为:r_c=0.986,RMSEC=1.628,r_cv=0.944,RMSECV=3.341;利用偏最小二乘法建立的马拉硫磷拉曼光谱检测模型最为理想,校正和交互验证结果为:r_c=0.996,RMSEC=1.175,r_cv=0.992,RMSECV=1.639。结果表明以硅胶片作为基底采集农药溶液的拉曼光谱效果比使用镜片时效果更好。4.建立了水果在两种农药残留情况下的拉曼光谱快速检测数学模型。对残留在赣南脐橙表皮上的毒死蜱和马拉硫磷的拉曼光谱数据分别采用峰强比值法和化学计量学方法建立了定量分析数学模型,探讨了拉曼光谱技术定量快速检测水果表面农药残留的可行性。实验结果为:利用最小二乘法—支持向量回归建立的毒死蜱残留拉曼光谱快速检测数学模型最为理想,r_c=0.974,RMSEC=2.592,r_cv=0.973,RMSECV=2.566;利用最小二乘法—支持向量回归建立的马拉硫磷残留拉曼光谱快速检测数学模型最为理想,r_c=0.948,RMSEC=4.101,r_cv=0.948,RMSECV=4.111。结果表明:利用拉曼光谱技术可以定量检测残留在水果表皮的农药残留含量,这为水果表面的农药残留快速检测提供了一种快速、绿色、环保的检测方法。更多还原

【Abstract】 With the progress of society and the improvement of people’s life, the higher demand of fruits security is needed as well as its appearance and nutritional quality. At present, routine analytical methods used for the determination of pesticide residue on the surface of fruits are destructive, complex, time-consuming, high cost and not environmentally friendly. Hence, it will be very worthful for pratical applications to explore a determination technique, which is simple, rapid, green and environmentally friendly. This technique is very meaningful to expand China’s exports of high-grade quality fruit, increase revenue of the fruit growers, and enhance the international competitiveness for the country’s fruit industry.Raman spectroscopy technique is a kind of spectral analysis technique based on the development of the Raman scattering effects, which show informations of the molecule’s vibration and rotating. With the merits of non-preparative sample, easy operation, short response time, green and environmentally friendly, Raman spectroscopy has been widely applied in petrol chemical, biomedicine, geoarchaeology, criminal justice and gem identification, etc. In the thesis, Gannan navel orange was chosen as the study objective, chlorpyrifos and malathion were chosen as analysis testing indexes. The rapid determination methods and quantitative models for pesiticide residues on the surface of fruits were studied using Raman spectroscopy technique. The main results of the thesis were involved:1. Raman spectra of chlorpyrifos and malathion standards and Gannan navel orange’s pericarp samples were acquired. Then the Raman spectrum’s ownerships and characteristic peaks of chlorpyrifos, malathion and Gannan navel orange’s pericarp were obtained.2. Raman spectrum preprocessing methods were studied and ideal methods were obtained for chlorpyrifos and malathion standard solutions and residues on the surface of Gannan navel orange. Raman spectra of the chlorpyrifos and malathion solutions at different substrates （ophthalmic lens and silicon sheet） and on the surface of Gannan navel orange’s pericarps were acquired. Then different spectral preprocessing methods were compared with and partial least square （PLS） regression models were established. The results showed that when the substrate was ophthalmic lens, the calibration model with linear baseline correcting preprocessing was the ideal model for chlorpyrifos solutions, and the calibration model with offset correcting preprocessing was the ideal model for malathion solutions. And when the substrate was silicon sheet, the calibration model with second derivative preprocessing was the ideal model for chlorpyrifos solutions, and the calibration model with first derivative preprocessing was the ideal model for malathion solutions. While the ideal preprocessing methods were first derivative and standard normal variate correction （SNV） for chlorpyrifos and malathion residues on the surface of Gannan navel orange, respectively.3. Rapid and quantitative mathematical models were established for the Rman spectra of the two kinds of pesticide standard solutions. Calibration models were eatablished using traditional peak-to-intensity ratio method and chemometric algorithms for Raman spectra data of chlorpyrifos and malathion solutions. Modeling effects between ophthalmic lens and silicon sheet for the acquisition of chlorpyrifos and malathion solutions were discussed. The results showed that when the substrate was ophthalmic lens, the model established by multiple linear regression （MLR） was the ideal model for chlorpyrifos solutions. The calibration and cross-validation results were: r_c=0.960, RMSEC=3.166, r_cv=0.877, RMSECV=5.421. And the PLS model was the ideal model for malathion solutions. The calibration and cross-validation results were: r_c=0.998, RMSEC=0.748, r_cv=0.985, RMSECV=2.236. However, when the substrate was silicon sheet, the PLS model was the ideal model for chlorpyrifos solutions. The calibration and cross-validation results were: r_c=0.986, RMSEC=1.628, r_cv=0.944, RMSECV=3.341. And the PLS model was the ideal model for malathion solutions. The calibration and cross-validation results were: r_c=0.996, RMSEC=1.175, r_cv=0.992, RMSECV=1.639. The results showed that it is better to acquire Raman spectrum of pesticide solutions by using silicon sheet than by using ophthalmic lens.4. Rapid and quantitative mathematical models were established for the Rman spectra of the two kinds of pesticide residues on the surface of fruit. Quantitative analysis mathematical models were established using traditional peak-to-intensity ratio method and chemometric algorithms for chlorpyrifos and malathion residues on the surface of Gannan navel orange. The feasibility of the rapid and quantitative determination of pesticide residue on the surface of fruit by Raman spectroscopy was discussed. The results showed that the least squares support vector regression （LS-SVR） model was the ideal model for the chlorpyrifos residue. The r_c, RMSEC, r_cv and RMSECV were 0.974, 2.592, 0.973 and 2.566. And the LS-SVR model was the ideal model for the malathion residue. The r_c, RMSEC, r_cv and RMSECV were 0.948, 4.101, 0.948 and 4.111. The results showed that pesticide residue on the surface of fruit could be determined by Raman spectroscopy, which supplied a kind of rapid, green and environmentally friendly method.更多还原