Study On The Identification Method Based On Three Exciation-emssion Matrix Fluorescence Spectra For Hazardous Chemicals In Water And Its Application

In recent years, sudden water pollution accidents occur frequently in China, especially for water pollution accidents caused by hazardous chemicals which become a serious threat to drinking water safety. Real-time detection for water quality change timely warning for sudden water pollution accident has become an urgent task to be carried out for environmental monitoring. With many advantages such as good selectivity, high sensitivity, no needs for sample pretreatment and the ability of real-time online detcting, fluorescence spectroscopy technique become the most promising technique for real-time detecting of hazardous chemicals in water. However, due to the lack of appropriate water hazardous chemicals automatic identification method, real-time qualitative analysis for a variety of hazardous chemicals in water based on fluorescence spectroscopy cannot be carried out, which severely limits the application of the fluorescence analysis technique in drinking water quality monitoring and early warning.In this paper, aiming at the technical requirement of real-time monitoring and early-warning for hazardous chemicals in water, a new method named similarity index method step by step for automatic identification of hazardous chemicals in water was proposed. According to the new method, the feature information of three-dimensional fluorescence spectral for hazardous chemicals in water is extracted in numerical format firstly, and then some similar indices based on the numeric feature spectral information are built, the indentification of components under test is carried out with the comparison of similar index step by step. The main results are summarized as follows:(1) Taking four kinds of typical fluorescent hazardous chemicals-phenol, styrene, p-aminophenol and aniline as objects, the extraction method of three dimensional excitation-emission matrix(3DEEM) fluorescence spectral feature informaiton for hazardous chemicals in water was studied. Fluorescent center and spectral shape were taken as two kinds of main feature information respectively. The excitation-emission wavelength represents the fluorescent center feature, the total excitation/emission spectrum and the degree of spectra dispersion in excitation/emission direction represent the spectral shape feature. Numerical extraction for the position of fluorescent center excitation-emission wavelength, the total excitation/emission spectrum and the degree of spectra dispersion in excitation/emission direction were carried out and then the feature information database of3DEEM fluorescence spectral for water hazardous chemicals in water was set up.(2)A new kind of automatic identification method for hazardous chemicals in water named similarity index identification method step by step was proposed. Based on the3DEEM fluorescence spectral feature information of for hazardous chemicals in water, Fluorescent center similarity index Re (0≤Rc≤1) and Integrated spectral shape similarity index Rs (0≤Rs≤1) were constructed. For mixed-component system, trilinear decomposition for mixed fluorescence spectra was carried out using parallel factor (PARAFAC) analysis model and the caculated fluorescence spectra for each component were obtained. With appropriate threshold value of Rc and Rs, the corresponding chemical species for calculation spectra can be indentified by comparing Rc and Rs of standard reference spectrum and calculation spectrum step by step. Verification test for the automaic identification method was conducted using phenol, styrene, p-aminophenol and aniline mixed samples, when Rc and Rs thresholds were both set to0.9, the identification accuracy for four hazardous chemicals is100%.(3) The application performance of similarity index identification method step by step in actual water body was studied. By using actual Yangtze River to compound to get simulate water samples which are polluted by styrene, ammonia, phenol, aniline,4kinds of hazardous chemicals in different levels. Under the premise of deducting the actual water’s auto-fluorescence background, combined with PARAFAC model and gradually similarity index identification method to identify the analysis obtained component in types, when Rc and Rs thresholds are set to0.9, the recognition accuracy reach to100%. Quantitative analysis of concentration showed that the relative error of concentration of the four kinds of substances between0.8to3.1percent.