The Model And Experimental Research Of Water COD Detection Based On Multi-spectral Information Fusion

Water is a fountain of human life,and a basic necessity for production and life.Recent years,water pollution has become more and more serious due to rapid economic development and city urbanization in China,and frequent water safety accidents or water pollution incidents have endangered people's health and social stability and harmony.Accurate and rapid monitoring of water quality has become special important and significant.In this thesis,a great deal of investigation has been carried out on various methods to detect water quality chemical oxygen demand?COD?detection.It is found that the traditional techniques of water quality testing are based on chemical analysis in the laboratory,which have complex testing process with large equipment maintenance and always cause secondary pollution.In order to solve this issue,a new technique is explored for detection based on multi spectral fusion analysis,which focus on water quality COD and is based on the method of the theory of information fusion.The main contents in this thesis are summarized as follows:?1?The background of detection of water COD by spectroscopic method is reviewed.The basic principle of ultraviolet?UV?spectrum and three-dimensional fluorescence spectra?excitation-emission matrix spectra,EEMs?are introduced in details,the characteristics of UV and EEMs for COD are analyzed,chemometrics methods are introduced.The general framework and the method to establish models are proposed,and the quantitative theoretical model is induced.?2?The chemometric algorithm theory and model establishment method based on spectrometry for detecting water COD are demonstrated.The concentrations of potassium acid phthalate standard solution and the COD value of the actual water samples which consist of sea water and surface water are taken as the object of study,and the method of water quality COD detection based on UV absorption spectrometry,3D fluorescence spectrometry,UV/fluorescence multi-spectral fusion and fluorescence/fluorescence multi-spectral fusion are analyzed in this paper.Different feature extraction algorithms?SiPLS,MWPLS,ACO-iPLS?and different chemometric methods?PLS,PSO-LSSVM?are used to establish the models,the prediction results of different spectral data models in this paper are compared with each other,which is the essential to explore the final model of this paper.The optimal detection model is established based on fluorescence multi-spectrum mid-level data fusion?MLDF?which combines the ACO-iPLS feature spectrum extraction method and PSO-LSSVM modeling algorithm.?3?The effect of different water parameters on the detection results of COD with spectral method is analyzed.Taking the potassium acid phthalate standard solution and the actual water samples as the research object,the influence of temperature,turbidity,pH,nitrite nitrogen?NO₂-N?,nitrate nitrogen?NO₃-N?and salinity on water COD detected by UV absorbance spectral technique and EEMs is discussed respectively.The effect of these various parameter on UV absorbance,the fluorescence intensity and the central position of fluorescence peaks are analyzed in theory.The corresponding compensation models of temperature and turbidity are established in experiment,and the predicted value of compensation is in good agreement with the real value.The function between COD and TOC is demonstrated by theoretical deduction and experimental verification.The method of detecting water COD based on UV and EEMs spectrum has been studied thoroughly with information fusion theory,a large number of experiments have been conducted to verify the theoretical analysis,and the validity of the experimental results is also verified by the spectral data of the actual water samples.The research results of this paper establish a theoretical foundation for developing optical in-situ sensors based on multispectral information fusion,and bring a very broad application prospect.