摘要
针对城市供水中各类有机污染事件难以通过简单阈值法进行超标报警等问题,本研究提出基于三维荧光光谱和平行因子分析法(parallel factor analysis,PARAFAC)的饮用水水质定性判别方法。在三维荧光光谱检测数据中含有的瑞利和拉曼散射会影响光谱信息的分析,因此采用Delaunay三角形内插值法进行数据预处理;然后,应用对荧光激发-发射矩阵(excitation-emission matrix,EEMs)进行分解,获得被测样品的荧光峰位置分布及各组分相对质量浓度,用于判断水体中是否受到有机污染物侵入;接着,应用支持向量机(support vector machine,SVM)对检测样本根据相对质量浓度值特征信息进行分类,将自来水水样和有机物污染的水样区分开来。最后,设计了不同质量浓度苯酚水溶液等检测实验,验证分析了所研究方法对于有机污染物的检出与判别能力。
The issue of potable water security is directly related to people＇s health and social stability.It is of great significance to detect the abnormality of potable water effectively and quickly to timely warn the occurrence of water pollution incident.With the technology means of obtaining water quality information increasingly diversified,domestic and foreign researchers have proposed different anomaly detection methods to extract and analyze the characteristic information of abnormal water samples.However,organic pollution in urban water supply is difficult to detect by simple threshold methods.At present,the application of spectral methods on the anomaly detection of urban water supply is still in the ascendant.According to the relevant literatures,the detection limit of potable water pollution based on ultraviolet and visible spectrum is relatively high.Therefore this paper proposed a method based on three-dimensionalfluorescence spectroscopy and parallel factor analysis（PARAFAC）for potable water quality inspection,which shows a better performance than ultraviolet and visible spectrum.Detection experiments of different concentrations of phenol solution et al.were designed to verify and analyze the detection and discrimination ability for organic pollution abnormality of this method.Data preprocessing was carried out using Delaunay interpolation because Rayleigh and Raman scattering contained in detection data will affect the analysis of spectral information.And then the fluorescence excitation-emission matrix（EEM）was decomposed using parallel factor analysis,and the distribution of fluorescence peaks and the relative concentration of factors were obtained,which can be used to judge whether the water is polluted.Support vector machine was applied to classify measured samples based on feature information of relative concentration value,which can distinguish normal water samples from the ones polluted by organics.Through the analysis of the experimental results,we could find that the method effectively detects the water samples contaminated by organics.In this experiment,phenol-contaminated water samples whose concentration were higher than 2μg/L,salicylic acid-contaminated water samples whose concentration were higher than 2μg/L,and rhodamine B-contaminated water samples whose concentration were higher than 1μg/L were all detected.It is proved that in the case of being close to the national standard,the use of three-dimensional fluorescence spectroscopy and PARAFAC can distinguish the pollution concentration of micrograms level.Aiming to solve the problems that potable water is easily contaminated by organics,this paper designed an experiment to detect the fluorescence spectroscopy of five days running water and three kinds of organic solution of different concentrations.Three-dimensional fluorescence spectroscopy combined with PARAFAC method is applied to detect the abnormality.The results shows that the proposed method can distinguish water samples contaminated by organics of microgram level.It could be concluded that the abnormal detection of potable water based on threedimensional fluorescence spectroscopy is feasible and has practical application value.
出处
《浙江大学学报（农业与生命科学版）》
CAS
CSCD
北大核心
2016年第3期368-377,共10页
Journal of Zhejiang University:Agriculture and Life Sciences
基金
国家自然科学基金(61573313
U1509208)
浙江省科技厅公益资助项目(2014C33025)
浙江省重点研发计划项目(2015C03G2010034)
中央高校基本科研业务费专项资金(2015FZA5012)
关键词
三维荧光光谱
平行因子分析
饮用水
定性判别
three-dimensional fluorescence spectroscopy
parallel factor analysis（PARAFAC）
potable water
qualitative discrimination
作者简介
第一作者联系方式：陈方（http：／／orcid．org／0000-0003-0799-1401），E-mail：cathy_flying@qq．com
通信作者（Correspondingauthor）：黄平捷（http：／／oreid．org／0000-0002-5487-6097），E-mail：huangpingjie@zju．edu．cn
