基于支持向量机的近海富营养化快速评价模型的建立
发布时间：2018-11-16 14:22
【摘要】：近年来，全球近海富营养化现象凸显，我国拥有世界上最大的陆架边缘海，粗放式经济发展模式造成我国近海富营养化现象尤为严重，，成为世界上近海富营养化最为严重的国家之一。我国近海富营养化的评价大多采用第一代富营养化评价模型，第一代富营养化评价模型所用参数的分析与检测周期长、工作量大，并且需要专业的仪器、专业的人员，不能实现对近海海域富营养化的快速监测和评价，而第二代富营养化评价模型并不完全适用于我国近海富营养化评价，且需要大量长期的调查资料作为支撑，所以极需发展一种可以对富营养化水体进行快速评价的方法体系。 本文以支持向量机为基础，用CDOM三维荧光组分、紫外吸收以及DO、Chla、浊度等现场可快速监测的水化因子替代DIN、DIP、COD等常规富营养化评价参数，建立新型快速的富营养化评价模型，并成功的运用此模型对东海海域的富营养化状况进行了评估。主要的研究成果如下： （1）利用第一代富营养化评价方法中的NQI方法对我国渤海、黄海和东海海域进行富营养化特征评价，发现我国东海的富营养化现象比较突出，其次是渤海、黄海，并在NQI的基础上对渤海、黄海以及东海富营养化的主要贡献因子进行分析，发现渤海富营养化的主要贡献因子为DIN、COD，黄海的主要贡献因子为COD、DIN、DIP，东海海域富营养化的主要贡献因子为DIN、COD。将各海域富营养化的主要贡献因子与三维荧光组分以及在波长为220nm、254nm、270nm、350nm的紫外吸收数据进行多元相关性分析，发现各海区富营养化的主要贡献因子与快速评价参数具有显著的相关性，为快速评价模型的建立提供了重要的依据。 （2）以支持向量机为载体，用CDOM三维荧光组分、在波长为220nm、254nm、270nm、350nm下的紫外吸收数据以及Chla替代NQI方法中的DIN、DIP、COD建立了我国近海富营养化快速评价模型，支持向量机近海富营养化快速评价模型主要是利用在MATLAB平台下的LIBSVM来实现，通过支持向量机建立的近海富营养化的快速评价模型与NQI、E值、模糊综合评价进行对比，发现以NQI和模糊综合评价作为参照的交叉验证分类准确率都在90%以上，以E值为参照的交叉验证的分类准确率分别在70%左右，并且交叉验证平均分类准确率并没有随着训练集数目的增加而增加，而是基本稳定在一个区间。选取以NQI为参照的163个训练集样本建立支持向量机近海富营养化快速评价模型，并得出了快速评价模型的判别函数。 （3）根据支持向量机所建立的富营养化快速评价模型对东海海域的富营养化状况进行评价，评价结果与模糊综合评价模型的评价结果以及前人对长江口富营养化的评估结果相类似，并对评价结果中出现的异常区域进行了分析，因此，建立在支持向量机基础上的近海富营养化评价模型在东海海区的富营养化评价中的应用结果是合理的。
[Abstract]:In recent years, the phenomenon of global offshore eutrophication has been highlighted. China has the largest marginal sea in the world, and the extensive economic development model has caused the phenomenon of eutrophication in China's coastal waters to be especially serious. It has become one of the most serious eutrophication countries in the world. The first generation eutrophication evaluation model is mostly used in the evaluation of offshore eutrophication in China. The analysis and detection period of the parameters used in the first generation eutrophication evaluation model is long, the workload is large, and professional instruments and personnel are needed. The rapid monitoring and evaluation of eutrophication in offshore waters can not be realized. However, the second-generation eutrophication evaluation model is not fully applicable to the evaluation of eutrophication in coastal waters of China, and it needs a lot of long-term investigation data as the support. Therefore, it is very necessary to develop a rapid evaluation method system for eutrophication water. Based on support vector machine (SVM), the hydration factors such as CDOM 3D fluorescence components, UV absorption and DO,Chla, turbidity were used to replace the conventional eutrophication evaluation parameters such as DIN,DIP,COD. A new rapid eutrophication evaluation model was established and successfully used to evaluate eutrophication in the East China Sea. The main research results are as follows: (1) the eutrophication characteristics of the Bohai Sea, Huang Hai and the East China Sea are evaluated by using the NQI method of the first generation eutrophication evaluation method, and it is found that the eutrophication phenomenon in the East China Sea is quite prominent. The second is Bohai Sea, Huang Hai, and on the basis of NQI, the main contribution factors of Bohai Sea, Huang Hai and East China Sea eutrophication are analyzed. It is found that the main contribution factor of Bohai Sea eutrophication is that of DIN,COD, Huang Hai, and that of COD,DIN, is the main contribution factor of Bohai Sea eutrophication. The main contributing factor of eutrophication in DIP, East China Sea is DIN,COD.. The main contribution factors of eutrophication in each sea area were analyzed by multivariate correlation analysis with three dimensional fluorescence components and UV absorption data at wavelength of 224 nm ~ 254 nm ~ 350 nm. It is found that the main contribution factors of eutrophication in each sea area are significantly correlated with the rapid evaluation parameters, which provides an important basis for the establishment of rapid evaluation model. (2) based on the support vector machine (SVM), the rapid evaluation model of eutrophication in offshore China was established by using CDOM three-dimensional fluorescence component, UV absorption data at wavelength of 220nmmO254nmmP270nm 350nm and Chla replacing DIN,DIP,COD in NQI method. The fast evaluation model of offshore eutrophication based on support vector machine (SVM) is mainly realized by LIBSVM based on MATLAB platform. The fast evaluation model of offshore eutrophication established by support vector machine is compared with NQI,E value and fuzzy comprehensive evaluation. It is found that the accuracy of cross-validation classification based on NQI and fuzzy comprehensive evaluation is more than 90%, and that of cross-validation with E value as reference is about 70%, respectively. And the average classification accuracy of cross-validation does not increase with the increase of the number of training sets, but is basically stable in an interval. A fast evaluation model of offshore eutrophication based on support vector machine (SVM) was established with 163 training set samples with reference to NQI, and the discriminant function of the fast evaluation model was obtained. (3) based on the rapid eutrophication evaluation model established by support vector machine, the eutrophication in the East China Sea is evaluated. The evaluation results are similar to those of the fuzzy comprehensive evaluation model and the results of the predecessors' assessment of eutrophication in the Yangtze Estuary, and the abnormal areas in the evaluation results are analyzed. The application of the offshore eutrophication evaluation model based on support vector machine in the eutrophication evaluation of the East China Sea is reasonable.
【学位授予单位】：中国海洋大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P734.44;X55;X824
本文编号：2335754
[Abstract]:In recent years, the phenomenon of global offshore eutrophication has been highlighted. China has the largest marginal sea in the world, and the extensive economic development model has caused the phenomenon of eutrophication in China's coastal waters to be especially serious. It has become one of the most serious eutrophication countries in the world. The first generation eutrophication evaluation model is mostly used in the evaluation of offshore eutrophication in China. The analysis and detection period of the parameters used in the first generation eutrophication evaluation model is long, the workload is large, and professional instruments and personnel are needed. The rapid monitoring and evaluation of eutrophication in offshore waters can not be realized. However, the second-generation eutrophication evaluation model is not fully applicable to the evaluation of eutrophication in coastal waters of China, and it needs a lot of long-term investigation data as the support. Therefore, it is very necessary to develop a rapid evaluation method system for eutrophication water. Based on support vector machine (SVM), the hydration factors such as CDOM 3D fluorescence components, UV absorption and DO,Chla, turbidity were used to replace the conventional eutrophication evaluation parameters such as DIN,DIP,COD. A new rapid eutrophication evaluation model was established and successfully used to evaluate eutrophication in the East China Sea. The main research results are as follows: (1) the eutrophication characteristics of the Bohai Sea, Huang Hai and the East China Sea are evaluated by using the NQI method of the first generation eutrophication evaluation method, and it is found that the eutrophication phenomenon in the East China Sea is quite prominent. The second is Bohai Sea, Huang Hai, and on the basis of NQI, the main contribution factors of Bohai Sea, Huang Hai and East China Sea eutrophication are analyzed. It is found that the main contribution factor of Bohai Sea eutrophication is that of DIN,COD, Huang Hai, and that of COD,DIN, is the main contribution factor of Bohai Sea eutrophication. The main contributing factor of eutrophication in DIP, East China Sea is DIN,COD.. The main contribution factors of eutrophication in each sea area were analyzed by multivariate correlation analysis with three dimensional fluorescence components and UV absorption data at wavelength of 224 nm ~ 254 nm ~ 350 nm. It is found that the main contribution factors of eutrophication in each sea area are significantly correlated with the rapid evaluation parameters, which provides an important basis for the establishment of rapid evaluation model. (2) based on the support vector machine (SVM), the rapid evaluation model of eutrophication in offshore China was established by using CDOM three-dimensional fluorescence component, UV absorption data at wavelength of 220nmmO254nmmP270nm 350nm and Chla replacing DIN,DIP,COD in NQI method. The fast evaluation model of offshore eutrophication based on support vector machine (SVM) is mainly realized by LIBSVM based on MATLAB platform. The fast evaluation model of offshore eutrophication established by support vector machine is compared with NQI,E value and fuzzy comprehensive evaluation. It is found that the accuracy of cross-validation classification based on NQI and fuzzy comprehensive evaluation is more than 90%, and that of cross-validation with E value as reference is about 70%, respectively. And the average classification accuracy of cross-validation does not increase with the increase of the number of training sets, but is basically stable in an interval. A fast evaluation model of offshore eutrophication based on support vector machine (SVM) was established with 163 training set samples with reference to NQI, and the discriminant function of the fast evaluation model was obtained. (3) based on the rapid eutrophication evaluation model established by support vector machine, the eutrophication in the East China Sea is evaluated. The evaluation results are similar to those of the fuzzy comprehensive evaluation model and the results of the predecessors' assessment of eutrophication in the Yangtze Estuary, and the abnormal areas in the evaluation results are analyzed. The application of the offshore eutrophication evaluation model based on support vector machine in the eutrophication evaluation of the East China Sea is reasonable.
【学位授予单位】：中国海洋大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P734.44;X55;X824
【参考文献】
相关期刊论文 前10条
1 王冉;杨道军;;基于支持向量机的巢湖富营养化程度评价研究[J];环境科学与管理;2011年05期
2 高定;陈同斌;刘斌;郑袁明;郑国砥;李艳霞;;我国畜禽养殖业粪便污染风险与控制策略[J];地理研究;2006年02期
3 黄明智;马邕文;万金泉;黄明护;;污水处理中人工神经网络应用研究的探讨[J];环境科学与技术;2008年03期
4 杨道军;王冉;沈刚;;SVM与ANN在湖泊富营养化评价中的对比研究[J];环境科学与技术;2012年01期
5 殷福才,张之源;巢湖富营养化研究进展[J];湖泊科学;2003年04期
6 蒋胜晖;文畅平;;陶瓷原料分类的Bayes判别分析法[J];硅酸盐通报;2008年02期
7 王家俊;汪帆;马玲;;SIMCA分类法与PLS算法结合近红外光谱应用于卷烟纸的质量控制[J];光谱学与光谱分析;2006年10期
8 程远月;郭卫东;;厦门湾有色溶解有机物光漂白的三维荧光光谱研究[J];光谱学与光谱分析;2009年04期
9 祝鹏;廖海清;华祖林;谢发之;汤智;张亮;;平行因子分析法在太湖水体三维荧光峰比值分析中的应用[J];光谱学与光谱分析;2012年01期
10 周楠;赵卫红;苗辉;;獐子岛附近海域溶解有机物的荧光特征[J];光谱学与光谱分析;2012年04期
相关博士学位论文 前3条
1 程远月;沉积物间隙水中CDOM光学特性与河口CDOM光化学反应研究[D];厦门大学;2007年
2 唐洪杰;长江口及邻近海域富营养化近30年变化趋势及其与赤潮发生的关系和控制策略研究[D];中国海洋大学;2009年
3 吴在兴;我国典型海域富营养化特征、评价方法及其应用[D];中国科学院研究生院（海洋研究所）;2013年
本文编号：2335754
本文链接：https://www.wllwen.com/kejilunwen/haiyang/2335754.html
