【作者】 吕江涛；

【导师】 王玉田；

【作者基本信息】 燕山大学 ， 仪器科学与技术， 2010， 博士

【摘要】 随着我国陆地与海底石油开采规模不断扩大,海洋和内河航运日益繁荣以及工农业生产的迅猛发展,工业废水、生活污水、农业排水及其他废物排放量逐年增加。排放的废水中残留许多石油产品,矿物油类污染物泄漏排放到自然水体中,将破坏生态环境并严重危害人们的身体健康。因此,水体中的油类污染物种类的鉴别和浓度的测定对环境治理工作十分重要。水中油类含量的测量主要采用浊度法、超声法、光散射法、重量法、紫外吸收法、非色散红外吸收法、红外分光光度法、色谱法、荧光光度法等等。受其测量原理的限制,对于微量的溶解于水中的油类污染物,其测量精度不高,不能进行矿物油种类识别,并且使用不便。本文研究一种新的基于荧光技术的检测识别微量矿物油类污染物的方法,提出一种光纤传导、CCD光谱探测、荧光光谱模式识别技术相结合的水中油类污染物检测识别系统设计方案,实现对水中矿物油直接快速检测和种类鉴别。从荧光检测理论出发,通过实验研究几种常见矿物油的荧光特性,确定所要检测矿物油的激发光谱和荧光光谱的波长范围,确定了矿物油浓度与相对荧光强度的线性关系。研究激发光源的光谱特性、光纤的传输特性、色散元件的分光特性、光谱分析用高性能线阵CCD的光谱响应特性等。设计系统的激发光源、光纤探头、小型CCD光谱仪(多色仪)、高速数据采集和荧光信号处理系统等;研究弱荧光信号的高效收集与传输;研究基于复杂可编程逻辑器件CPLD的CCD光谱检测数据采集系统。在单片机控制下,由CPLD自动实现荧光光谱数据的高速采集。研究矿物油种类模式识别技术,从三维光谱数据中提取、选择能够反映油种本征特征的光谱数据。分别采用核主成分分析(KPCA)法,独立分量分析(ICA)法,对水中油类污染物的荧光光谱信号进行特征提取;研究朴素贝叶斯、K近邻、支持向量机以及多小波神经网络四种分类器分别与KPCA和ICA结合的分类效果,比较各种方案的分类识别率,最终确定系统矿物油种类识别方法。研究采用基于朗伯—比尔定律建立的荧光光谱法定量分析水中油含量的数学模型,研究线性校正原理实现单组分矿物油污染物的定量分析,结合导数荧光光谱法对多组分有机物体系不经前期分离而直接实现各组成成分的定量测定,评价系统的性能。更多还原

【Abstract】 With the development of economy, the exploitation scale of the fossil oil on land or on sea bed has been enlarged. The carrying trade by ship has been flourished and industry and agriculture have made great progress. The emission amount of the industrial effluent and sanitary waste and agricultural drain and other rubbish are increased on an annual basis. There is a lot of remained petroleum product in the waste water. It caused the pollution of the environment. The mineral oil is leaked out to the water. It has been seriously affected the health of the human. So, it is of great importance to identify the species of the mineral oil in the water and determinate the density of the mineral oil to the environment protection.At the present time, existing detection method of oil pollution ad mineral oils mainly includes the nephelometric and ultrasonic method, light scattering method, gravimetric method, ultraviolet absorption method, no dispersive infrared absorption method, infrared spectrophotometer method, and chromatography and fluorophotometric method and so on. The measurement accuracy of them is not good enough to the micro content measurement of the oil because of the measuring principle of them. They are not portable and can not identify the species of the mineral oil.In this paper, the method to identify the species of the micro content oil in water based on the fluorescence spectrum detection is studied. A design scheme of the identification of the mineral oil in water based on the combination of the optical fiber sensing and the spectral detection and the pattern recognition of the fluorescence has been proposed. It can realize the fast detection and the species identification of the mineral oil.Starting with the basic principle of fluorescence measurement, the fluorescent characteristic was researched by the experiment. The optimally detection parameter of the wavelength range of the excitation and the emission spectra has been determined. The linear relation between the density of the mineral oil and the fluorescence intensity has been determined. The spectral character of the optical source and the transfer characteristic of the fiber and the dichroism of the dispersion element and the spectral response characteristic of the CCD has been researched. The optical source, the fiber-optics probe, the little CCD spectrometer, the high speed data acquisition and the signal process of the system have been designed. The highly active collection and the transfer of the weak fluorescence signal have been researched. The CCD spectrum detection system based on the CPLD has been researched. The CPLD realized the auto data acquisition under the control of the chip.The pattern recognition technology of the mineral oil has been researched. It aims to extract the data which can reflect the intrinsic characteristic of the mineral oil from the spectral data. The KPCA and the ICA was used separately to extract the characteristic of the spectral signal of the mineral oil in water. The Naive Bayes and the KNN and the SVM and the WNN were used separately to classify the characteristic extracted by the KPCA and ICA. The discrimination of them was compared and the method of the identification of the mineral oil has been determined.The mathematic model to quantitative analyze the oil content in water based on the Lambert-Beer law has been researched. The linearity correction principle was used to realize the single constituent quantitative analysis of the mineral oil. It combined the derivative spectrum method to realize the quantitative determination of every component in the multi-component mixed oil without the prophase dissociate. The performance of the system was appreciated.更多还原