The Application Of Chemometrics In The Toxicity Analysis And Source Apportionment Of Atmospheric Particulate Pollutants

Along with the acceleration of urbanization,industrialization and regional economy integration,the complex air pollution characteristics have emerged in our country,which have a significant influence on ecosystem,human health,atmospheric visibility and climate change.Atmospheric particulate matter is one of the most complex and harmful pollutants in atmospheric environment,the concentration and ratio of major components in atmospheric particulate matter can be used to determine the the major sources of pollution.Polycyclic aromatic hydrocarbons(PAHs)is an important part of atmospheric particulate matter,and by studying the carcinogenicity and source apportionment of PAHs,the toxicity and sources of pollution can be judged.In this paper,according to the practical needs for the accurate prediction and source apportionment,the morphology characteristics and content of main components in atmospheric particulate matter were studied,and the RF model was constructed for the carcinogenicity studies of PAHs,and Raman spectroscopy combined with PLS method was used for the source apportionment of PAHs,thus achieving the source apportionment of atmospheric particulate matter.This study provides new thought and method for the real-time monitoring and control of complex air pollution.This paper is divided into four chapters,the main content are as follows:Firstly,taking the atmospheric particulate matter collected from campus dormitory and school gate area as study subject,the morphology characteristics and content of main components were study,then according to these information,the category,composition distribution and concentrations information of atmospheric particulate matter can be obtained.Secondly,taking the carcinogenicity of PAHs in atmospheric particulate matter as study subject,RF combined with QSAR was used in the prediction of carcinogenicity of PAHs.First,the molecular descriptors were calculated and the variable importance was used to select the input variables,then the RF-QSAR model was constructed for the prediction of carcinogenicity,and the result shows RF model have a better prediction performance compared with PLS-DA model,and it is also a potential method for the prediction of carcinogenicity of PAHs.Finally,taking the PAHs content and source apportionment as study subject,the feasibility of using PLS combined with Raman spectroscopy for the quantitative analysis and source apportionment of PAHs was studied.First,first-order derivative method was used for baseline calibration and noise filtration.Then the Raman spectra data were used to construct PLSR model for quantitative analysis of PAHs,and on this basis the PLS-DA model was constructed for the source apportionment of PAHs.The result shows PLS and PLS-DA model have a better prediction performance compared with RF model,and it is also an effective method for the accurate quantitative and pollutants sources analysis of PAHs in atmospheric particulate matte.