水体细菌微生物多波长透射光谱快速准确识别方法研究

Bacteria microbial pollution is the top issue facing the drinking water safety in China. Spectroscopic technique is a flexible tool for bacterial microbes determination in comparison with conventional field sampling and laboratory assay, and it provide a rapid and in real time analysis method. However, the present research methods based on part of the structure or component features of microbes,there are some disadvantages in real application at present time, such as the data of the feature information is single and the characteristic data is fewer, which may cause the difficulty of accurately identify analysis for different bacterial microbes. This project proposes multi-wavelength transmission spectroscopy solution based on the difference of micro structure & composition of bacteria microbial,and solve the target bacteria accurately identify under complex environment..The research work will start with the study of multi-wavelength transmission spectrum detection method of highly sensitive and in real time.and will be more focus on the multi-wavelength transmission spectral characteristics of aquatic pathogenic bacterial microbes, such as Escherichia coli, Salmonella and Shigella. The method of spectral analysis & extraction spectrum characteristic parameter will be studied based on the micro structure component to solve the lack quantitative character in spectral recognition analysis. To solve coexistence components and environmental factors interference affecting in accurate identification at complex background, the method using comprehensive similarity index will be studied for accurate identification of target bacteria species and precise separation of multi-component coexisting spectra. An algorithm model will be established for microstructure component analysis and species accurate identification of bacterial microbes based on multi-wavelength transmission spectral characteristics. The coming out project will provide a novel method for basic scientific research of bacterial microbes and rapid accurate identification of typical pathogenic bacteria in water.

细菌微生物污染是我国饮用水安全面临的首要问题。与现场采样、实验室检测方法相比，光谱技术提供了一种快速、实时分析的手段，但现有研究方法多基于细菌微生物部分结构或组分特征，存在特征信息数据单一、数据信息量少以及准确识别困难。本项目基于细菌微生物微结构成分差异提出多波长透射光谱研究方案，并重点解决复杂背景下目标细菌微生物的准确识别。.项目将从水体细菌微生物的高灵敏光谱快速获取方法入手，重点研究水体致病菌（大肠杆菌、沙门氏菌、志贺氏菌等）的光谱结构特性与特征；研究基于微结构成分的光谱解析与特征参数提取方法，解决识别分析中光谱量化特征缺少问题；研究采用综合相似度指数准确确定目标细菌微生物与精确分离共存组分光谱的方法，解决复杂背景共存组份干扰及环境因素影响难题；建立水体细菌微生物多波长透射光谱微结构成分解析与种类准确识别算法模型，为水体典型致病菌的快速识别预警和细菌微生物的基础科学研究提供方法基础。

针对水体典型致病菌快速在线识别预警需求，研究了水体细菌微生物多波长透射光谱特性解析与快速识别方法，建立了水体细菌微生物多波长透射光谱微结构成分解析与识别算法模型并进行了验证。取得主要研究成果如下：.（1）研究了水体细菌微生物多波长透射光谱快速高灵敏检测方法，建立了以氘灯、溴钨灯及多通道光栅光谱仪为光源和分光探测器件的多波长透射光谱快速测量实验系统，实现了0.08nm采样间隔下15秒30次光谱信号的自动采集。.（2）研究了水体细菌微生物微结构成分特征光谱解析方法，建立了基于Mie散射理论结合细菌微生物生理学结构的多波长透射光谱解析模型，对大肠杆菌、肺炎克雷伯氏菌和金黄色葡萄球菌分析结果表明，光谱解析最大相对误差小于8%，平均相对误差小于0.3%。.（3）研究了水体细菌微生物微结构成分特征参数提取方法，获取了细菌结构参数、平均核酸及蛋白质含量，重构了多波长透射光谱特征指纹。结果表明：大肠杆菌体积相对误差为7.9%，内部结构占比、平均核酸及蛋白质含量与文献具有较好一致性，不同特征参数相对标准偏差均小于5%。.（4）研究了光谱相似性分析、光谱特征值提取结合支持向量机以及人工神经网络的水体细菌微生物识别方法，建立了基于后向传播、广义回归以及概率神经网络的识别模型。结果表明：神经网络模型识别准确率均大于95%，光谱特征值提取结合支持向量机提高了识别准确率，均优于相似性分析识别结果。.（5）研究了水体细菌微生物混合光谱解析及浓度反演方法，建立了基于蒙特卡洛算法的细菌微生物混合光谱组分解析以及基于反向传播神经网络的浓度反演模型。结果表明：光谱归一化结合主成分降维后组分解析及浓度反演结果平均相对误差均小于7%。.项目完成了实际水体样品识别验证分析，初步研发了水体细菌微生物多波长透射光谱快速在线监测仪样机。研究成果为进一步开展水体典型致病菌的快速识别应用研究奠定了方法基础。