水体COD的光谱学在线测量方法-紫外和近红外光谱比较分析

发布时间：2018-11-27 12:29

【摘要】：水体COD的光谱学传感技术是现代环境监测的一个重要发展方向,与传统的分析方法相比,光谱分析技术更具有可连续监测、可在线监测和检测快速的明显优势,适合对环境水样COD的定点实时监测。分别获取水样的紫外吸收光谱和近红外光谱,通过不同的光谱预处理方法结合偏最小二乘法、多元线性回归法建立水样的COD定量预测模型,对水体COD的紫外和近红外光谱的定量预测及相关模型参数进行分析,发现用S-G平滑处理后的紫外光谱和近红外光谱建立的PLS模型均得到最佳预测效果,预测集R~2分别为0.992 1和0.987 7,RMSEP分别为10.438 6和5.972 0。紫外和近红外光谱法的MLR模型预测效果较差,预测集R~2分别为0.928 0和0.957 3。通过实验结果综合对比分析,紫外吸收光谱在280~310nm谱区建模预测性能较好,近红外光谱在7 250~6 870cm~(-1)谱区建模预测性能较好,紫外光谱对应定量预测模型的决定系数较高,而近红外光谱的稳定性和重复性更好。研究表明光谱传感技术可用于环境实际水体COD的定量预测分析,为开发便携式水体检测设备奠定了理论基础。
[Abstract]:Spectroscopic sensing technology of water body COD is an important development direction of modern environmental monitoring. Compared with traditional analytical methods, spectral analysis technology has obvious advantages of continuous monitoring, on-line monitoring and rapid detection. It is suitable for spot real-time monitoring of environmental water sample COD. The ultraviolet absorption spectra and near infrared spectra of water samples were obtained respectively. The COD quantitative prediction model of water samples was established by using different spectral pretreatment methods combined with partial least square method and multivariate linear regression method. The quantitative prediction of UV and NIR spectra of water COD and the related model parameters were analyzed. It was found that the best prediction effect was obtained by using the UV spectrum and the NIR PLS model established by S-G smoothing. The prediction set RMS-EP was 10.438 6 and 5.972 0, respectively, which were 0.992 1 and 0.987 7 m ~ (-1), respectively. The prediction results of MLR model with UV and NIR spectra are poor, and the prediction set RP2 is 0.928 0 and 0.957 3 respectively. Through the comprehensive comparison and analysis of the experimental results, the modeling and prediction performance of UV absorption spectrum in 280~310nm spectrum region is better, that of near infrared spectrum is better in 7 250 ~ (6 870) cm ~ (-1) region, and the determinant coefficient of UV spectrum corresponding quantitative prediction model is higher. The stability and repeatability of NIR spectra are better. The results show that the spectral sensing technology can be used for quantitative prediction and analysis of COD in environmental water bodies, which lays a theoretical foundation for the development of portable water detection equipment.
【作者单位】：北京农业智能装备技术研究中心北京市农林科学院;重庆邮电大学;
【基金】：国家(863)计划项目(2013AA10230202) 国家自然科学基金项目(31271614)资助
【分类号】：O657.3;X832

【相似文献】