近红外光谱法快速测定工作场所莠去津浓度的研究

发布时间：2018-03-07 09:13

本文选题：近红外光谱　切入点：莠去津　出处：《光谱学与光谱分析》2017年03期 　论文类型：期刊论文

【摘要】：莠去津是一种广泛使用的除草剂,我国是其原药的主要生产国家。为加强工作场所莠去津暴露浓度的检测力度,保障职业接触工人身体健康,研究开发工作场所莠去津浓度的现场快速检测方法具有重要现实意义。利用自行组装的便携式近红外光谱仪,采集了实验室配置的浓度为10~1 000 mg·L~(-1)的莠去津溶液样本光谱,并比较了多元散射校正、变量标准化、一阶导数方法、二阶导数方法及其组合等光谱预处理方法,竞争自适应重加权采样变量选择法和遗传算法等变量选择方法,偏最小二乘算法和支持向量机等回归方法对近红外光谱模型分析精度的影响。研究发现一阶导数是最佳光谱预处理方法;遗传算法优选的光谱变量表现优于竞争自适应重加权采样变量选择法;支持向量机模型表现优于偏最小二乘模型。基于遗传算法选择的16个光谱变量建立的支持向量机模型分析精度最高,其定标决定系数、验证决定系数、定标均方差、预测均方差和相对分析误差(成分浓度的标准偏差与预测均方差的比值)分别为1,0.99,17.54 mg·L~(-1),25.42 mg·L~(-1)和11.43,有望应用于工作场所莠去津浓度的实际检测中。该研究探讨了近红外光谱法检测工作场所莠去津浓度的可行性,相关结果对于未来类似工作的开展具有重要参考价值。
[Abstract]:Atrazine is a widely used herbicide, and our country is the main producing country of its raw medicine. In order to strengthen the detection of atrazine exposure concentration in the workplace and protect the health of the workers exposed to atrazine in the workplace, It is of great practical significance to study and develop a rapid detection method for atrazine concentration in workplace. By using a portable near infrared spectrometer, the sample spectra of atrazine solution with a concentration of 10 ~ (1 000 mg 路L ~ (-1)) were collected in the laboratory. The spectral pretreatment methods, such as multivariate scattering correction, variable standardization, first derivative method, second derivative method and their combination, are compared, and variable selection methods such as competitive re-weighted sampling variable selection method and genetic algorithm are compared. The effect of partial least square algorithm and support vector machine on the accuracy of near infrared spectral model analysis is studied. It is found that the first derivative is the best spectral pretreatment method. The performance of spectral variables of genetic algorithm is better than that of competitive adaptive re-weighted sampling variable selection method. The SVM model is superior to the partial least-squares model. The SVM model based on 16 spectral variables selected by genetic algorithm has the highest analytical accuracy, its calibration decision coefficient, validation decision coefficient, calibration mean square deviation, etc. The prediction mean square error and relative analysis error (the ratio of standard deviation of component concentration to predicted mean deviation) were 1 ~ 0.99t 17.54 mg 路L ~ (-1) ~ (-1) and 11.43, respectively. It is expected to be applied to the practical detection of atrazine concentration in workplace. This study has studied the near infrared spectrum (NIR). The feasibility of measuring atrazine concentration in workplace by spectral method. The relevant results have important reference value for the development of similar work in the future.
【作者单位】：职业安全健康北京市重点实验室北京市劳动保护科学研究所;中国农业大学理学院;
【基金】：北京市科学技术研究院2015年创新工程项目Ⅲ-1(PXM2015-178304-000004) 北京市博士后工作经费项目(2015zz-110)资助
【分类号】：R134;O657.33

【相似文献】