【作者】 袁丽；

【导师】 孔令富；

【作者基本信息】 燕山大学 ， 计算机科学与技术， 2021， 博士

【摘要】 海上溢油污染是最常见的海洋污染之一。随着海洋运输业和海洋石油开采业的快速发展,溢油污染日益严重,给海洋环境和海洋生态平衡带来极大威胁。因此海洋溢油污染的治理、改善,成为海洋环境保护工程中刻不容缓的重要工作。而对溢油进行定位、识别和监测则是治理溢油污染的关键与前提。海洋激光诱导荧光(Laser Induced Fluorescence,LIF)探测技术因其具有的灵敏度高、选择性好、快速实时等特点,被认为是最重要和最有效的监测溢油的技术手段之一。目前基于激光荧光遥感的研究大多集中于未乳化溢油的监测,对海面溢油的重要存在形式-乳化溢油,尚未展开相关研究。而乳化溢油会给海洋环境带来更大危害,因此如何实现海面乳化溢油的定性描述和定量计算,即能够迅速准确给出乳化溢油的光谱分析数据,满足基于LIF的乳化溢油探测需要,具有重要的作用和意义。鉴于此,本文开展的主要研究工作如下:首先,阐述了LIF海面溢油探测的原理。主要从溢油的组成和荧光机理两个方面分别论述其优势所在,一方面溢油含有的大量芳香烃为监测识别提供物质基础,另一方面一定波长和能量的激光照射时,化合物能被激发出更长波长荧光的作用机理。最后确定采用荧光光谱来分析石油污染物的可行性,为后续研究奠定理论基础。其次,开展了海面乳化溢油的室内模拟工作,对其光谱特征进行分析。通过制备不同油水比、不同深度油水乳化物样品及获取其光谱,建立完善的光谱实验数据,阐明乳化与未乳化溢油相比的光谱变化规律,详细探究不同类型油水乳化物的精细光谱响应特征,并从实验数据分析得出荧光强度、拉曼强度、荧光峰位分别与深度、油水比之间的变化趋势。然后,对海面乳化溢油的识别问题进行了研究。主要包括乳化与非乳化的区分,乳化不同阶段的区分以及和油种区分相结合三个方面。基于前面实验分析所得出的光谱变化规律进行乳化溢油识别方法的探讨:一是光谱特征参量化与聚类分析相结合的识别方法,二是主成分分析与BP神经网络相结合的识别方法。两种方法具有不同的识别侧重点。最后,对海面乳化溢油的溢油量进行估算分析。一种方法是基于统计学理论,采用回归分析方法,进行具体数值统计,并建立相应的量值模型。另一种方法是基于等效思想,采用将乳化溢油等效为油膜的方法,分别建立油包水和水包油对应的等效模型。最终实现乳化溢油定量估算的研究。更多还原

【Abstract】 Oil spill pollution is one of the most common marine pollution.With the rapid development of marine transportation and oil exploitation,oil spill pollution is becoming more and more serious,which poses a great threat to the marine environment and ecological balance.Therefore,the treatment and improvement of oil spill pollution has become an urgent and important work in marine environmental protection engineering.The key and premise of oil spill pollution control is to locate,identify and monitor the oil spill.Marine laser induced fluorescence(LIF)detection technology is considered as one of the most important and effective methods for oil spill monitoring because of its high sensitivity,good selectivity,fast real-time and other characteristics.At present,most of the researches based on laser fluorescence remote sensing are focused on the monitoring of non emulsified oil spill,but there is no relevant research on emulsified oil spill,which is an important form of oil spill on the sea.Emulsion oil spill will bring more harm to the marine environment,so how to realize the qualitative description and degree identification of emulsion oil spill on the sea,that is,to quickly and accurately give the spectral analysis data of emulsion oil spill,to meet the needs of the detection of emulsion oil spill based on LIF,has an important role and significance.In view of this,the main research work of this dissertation is as follows:Firstly,the principle of LIF detecting oil spill on the sea surface is described.This dissertation mainly discusses its advantages from the composition and fluorescence mechanism of the spilled oil.On the one hand,a large number of aromatic hydrocarbons in the spilled oil provide material basis for monitoring and identification;on the other hand,when the laser with certain wavelength and energy irradiates,the compounds can be excited with longer wavelength fluorescence.Finally,the feasibility of using fluorescence spectrum to identify and analyze oil pollutants is determined,which provides the basis for the follow-up research.Secondly,the indoor simulation of emulsified oil spill on the sea surface is carried out,and its spectral characteristics are analyzed.The oil-water emulsion samples with different oil-water ratio and depth and their spectra were obtained through experiments.Establish perfect spectral experimental data,clarify the spectral change rule of emulsified and non emulsified spills,and analyze the fine spectral response characteristics of different types of oil-water emulsions in detail.And the variation trends of fluorescence intensity,Raman intensity and fluorescence peak position with depth and oil-water ratio are obtained from the analysis of experimental data.Then,the identification study of the emulsified oil spill on the sea surface was carried out.It includes the distinction between emulsification and non emulsification,the distinction between different stages of emulsification and the combination of emulsification and oil classification.According to the spectral characteristics and change rules of emulsification and non-emulsification obtained in the previous study,the emulsified oil spill identification method is studied: one is the identification method combining spectral feature parameter and cluster analysis,and the other is principal component analysis.Recognition method combined with BP neural network.And the two methods have different recognition focuses.Finally,estimate and analyze the oil spill volume of emulsified oil on the sea surface.One method is based on statistical learning theory,using regression analysis methods to carry out specific numerical statistics and establish a corresponding value model.The other method is based on the idea of equivalence,adopting the method of equivalent emulsified spilled oil to oil film,and establishing equivalent models corresponding to water-in-oil and oil-in-water respectively.Finally,the research on quantitative estimation of emulsified oil spill is realized.更多还原