【作者】 黄丽；

【导师】 陈瑜； 马瑞峻；

【作者基本信息】 华南农业大学 ， 农业硕士（专业学位）， 2023， 硕士

【摘要】 有机磷农药是农业生产中最常用的一类杀虫剂,其在田间喷洒时一般只有10%～20%附着在农作物上,其余大部分都残留在土壤和水体环境中,长期、不加限制地使用农药会对生态环境造成巨大危害。针对自然水体,目前国内外主要是采用色谱分析方法进行有机磷农药检测,但该方法前处理过程繁琐、具有破坏性,且检测仪器笨重、价格昂贵,无法实现农药残留的现场快速检测。近年来,紫外-可见光谱技术因其快速、无损的特点被广泛应用于有机磷农药的残留分析。在复杂的自然水体中,紫外-可见光谱技术结合传统的一阶校正方法无法实现多组分有机磷农药的检测。因此,本文提出将紫外-可见光谱结合二阶校正方法——平行因子分析法(Parallel factor analysis,PARAFAC)应用于水体中多组分有机磷农药的快速检测,主要研究内容和结论如下:纯净水背景下的基础理论研究。以毒死蜱、甲基对硫磷、丙溴磷为实验对象,采用纯净水为稀释剂配置不同组分的有机磷农药溶液样本,优选出50mm光程的比色皿,获取各样本在紫外-可见光范围内的二维吸收光谱数据。将不同的二维光谱数据构建为三维矩阵后采用PARAFAC算法进行分解,利用分解得到的得分矩阵与各组分的真实浓度构建线性回归模型,利用所构建的模型对不同的预测集进行预测。结果表明,算法分解得到各因子的吸收峰位置与各组分的实际光谱基本吻合,表明PARAFAC算法可以对多组分有机磷农药进行定性分析。所建立的2组分模型实现了对其中所有组分的定量检测,3组分模型实现了3组分混合溶液中毒死蜱和甲基对硫磷的定量检测。农田水背景下的实际应用研究。以毒死蜱、辛硫磷、噻唑磷为实验对象,采集华南农业大学农事训练中心的农田水为稀释剂,配置不同组分的有机磷农药溶液样本,优选出100mm光程的比色皿,获取各样本在紫外-可见光范围内的吸收光谱数据。将获取的多个二维光谱数据构建为三维矩阵后,采用PARAFAC算法进行分解,构建各组分的线性回归模型,利用所构建的模型对不同的预测集进行预测。结果表明,由4因子和5因子分解得到的三种有机磷农药的光谱与其各组分的真实光谱基本吻合,表明PARAFAC算法可以实现农田水中3组分有机磷农药的定性分析,其中由4因子构建的模型实现了对3组分混合溶液中毒死蜱和辛硫磷的定量检测。综上所述,本文采用紫外-可见光谱技术结合PARAFAC分析方法,构建了适用于纯净水和农田水中的多组分有机磷农药预测模型,满足了对其中各组分定性分析和部分组分定量检测的要求,实现了“数学分离”代替“化学分离”,为复杂自然水体中有机磷农药的快速检测技术提供了参考和借鉴。更多还原

【Abstract】 Organophosphorus pesticides are the most commonly used in agricultural of a class of pesticides,which are generally sprayed in the field only 10%to 20%attached to crops,most of the rest are left in the soil and water.Use of pesticides for an extended period of time without restriction can have a negative impact on the ecosystem.Organophosphorus pesticides are primarily detected for natural water by chromatographic analysis both domestically and internationally,but the pre-treatment process is time-consuming and destructive,and the detection equipment is large and expensive,making it impossible to detect pesticide residues quickly in the field.In recent years,UV-Vis spectroscopy has been widely used for the analysis of organophosphorus pesticide residues because of its rapid and non-destructive characteristics.In complex natural water,UV-Vis spectroscopy combined with traditional first-order correction methods cannot achieve the detection of multi-component organophosphorus pesticides.Therefore,this paper proposes to apply UV-Vis spectroscopy combined with a second-order correction method——parallel factor analysis(PARAFAC)to the rapid detection of multi-component organophosphorus pesticides in water,and the main research contents and conclusions are as follows:Firstly,chlorpyrifos,methyl parathion and profenofos were used as experimental objects,and pure water was used as diluent to configure samples of different components of organophosphorus pesticide solutions,and a 50 mm range cuvette was preferred to obtain the two-dimensional absorption spectral data of each sample in the UV-visible range.The various two-dimensional spectral data were built as three-dimensional matrices,decomposed using the PARAFAC algorithm,and then linear regression models with the true concentrations of each component were built using the score matrix obtained from the decomposition.The built models were then used to make predictions for various prediction sets.The results showed that the absorption peak positions of each factor obtained from the decomposition of the algorithm basically matched with the actual spectra of each component,indicating that the PARAFAC algorithm can perform qualitative analysis of multi-component organophosphorus pesticides.The 2-component model was developed to achieve quantitative detection of all the components,and the 3-component model achieved quantitative detection of chlorpyrifos and methyl parathion in the 3-component mixed solution.Secondly,chlorpyrifos,octreotide and thiazophos were used as experimental objects,and farmland water from the Agricultural Training Center of South China Agricultural University was collected as diluent to configure samples of different components of organophosphorus pesticide solutions,and a cuvette with 100 mm range was preferably selected to obtain the absorption spectral data of each sample in the UV-visible range.The PARAFAC algorithm was used to break down and build linear regression models for each component of the many obtained two-dimensional spectrum data into a three-dimensional matrix.The built models were then used to forecast various sets of predictions.The results showed that the spectra of the three organophosphorus pesticides obtained by the 4-factor and 5-factor decomposition basically matched with the real spectra of their respective components,indicating that the PARAFAC algorithm could achieve the qualitative analysis of the 3-component organophosphorus pesticides in farmland water,in which the model constructed by the 4-factor achieved the quantitative detection of chlorpyrifos and phoxim in the 3-component mixed solution.In summary,this thesis used UV-Vis spectroscopy combined with PARAFAC analysis to construct multi-component organophosphorus pesticide prediction model for pure water and farmland water,which meets the requirements of qualitative analysis of each component and quantitative detection of some components,realizing"mathematical separation"instead of"chemical separation",providing reference for rapid detection of organophosphorus pesticides in complex natural water.更多还原