Research On Classification Of Contaminants And Detection Of Contamination Events In Drinking Water With UV Spectroscopy

Water is the most important resource for human survival and the quality of drinking water directly determines people’s lives.With the development of molecular spectroscopy and chemical metrology,the detection of organic contaminants in water distribution systems with ultraviolet(UV)spectroscopy has the advantages of rapid detection,low cost,no need for reagents and high sensitivity to organic contaminants.Its speed,accuracy and comprehensive analysis meet the requirements for water quality monitoring well,which has become a new method of drinking water quality monitoring.However,the current methods for detection of contamination events with UV has some limitations to classify contaminants and adapt to spectral fluctuations.Besides,the current methods are not able to detect some low concentration events.This thesis carries out the research on the classification of contaminants and detection of low concentration contamination events in water distribution systems with UV.The main research process and innovation are illustrated as follows:(1)The principle of water quality detection with UV is studied,and the problems in the process of water quality detection by UV are analyzed.Aiming at eliminating the interference of the random noise,local noise and scattering,this work undertakes the special pretreat according to the factors of different interference.Aiming at extracting the information of contaminants,the method utilizes mean centering and orthogonal signal correction.(2)Considering the current method has some limitations to classify contaminants,a method to classify contaminants is formed.It utilized the successive projections algorithm(SPA)and SVM to form a multi-classification model.SPA eliminates the interference of multi-collinearity and amplifies the difference among the UV spectra of different contaminants.And SVM classifies the contaminants accordingly afterwards.The experimental results prove the effectiveness of the multi-classification model which is used to classify contaminants.(3)In order to enhance the ability to detect the low concentration contamination,an adaptive detection method based on a semi-supervised learning model is proposed.More specifically,the proposed method comprehensively utilizes the UV spectra of contaminants and the latest normal water quality to dynamically update the random forest model.Experimental results validate the feasibility of the proposed method and demonstrate that the proposed method shows higher adaptability and better performance in the detection of low concentration contamination.(4)In order to improve the practicability of the detection method for contamination with UV,we design and develop a drinking water quality monitoring system based on Spring MVC.A water quality anomaly monitoring platform is formed,using Java,JavaScript and MySQL jointly,hybrid technology is used to package the above-mentioned spectral preprocessing algorithm,classification algorithm and contamination detection algorithm written by MATLAB,and apply these algorithms in the system to analyze UV spectra and to detect contamination online.Focusing on the contaminant classification with UV,the method combining SPA and SVM is studied.In order to improve the ability to detect the low concentration contamination events,the detection method based on semi-supervised learning model is formed.Based on the research of this thesis,a water quality monitoring platform with UV spectroscopy is developed and applied.