中国科学院生态环境研究中心机构知识库
Advanced  
RCEES OpenIR  > 环境水质学国家重点实验室  > 学位论文
题名: 基于非负矩阵分解的源解析模型优化及其在子牙河沉积物PAHs溯源研究中的应用
作者: 高泽晋1
学位类别: 硕士
答辩日期: 2017-04
授予单位: 中国科学院大学
授予地点: 北京
导师: 单保庆
关键词: 非负矩阵分解,源解析,多环芳烃,沉积物,子牙河 ; Non-negative matrix factorization, source apportionment, polycyclicaromatic hydrocarbons, sediments, Ziya River Basin
其他题名: Optimization of source pportionment model based on non negative matrix factorization and its application in PAHs tracing study of sediments in Ziya River
学位专业: 环境工程
中文摘要: 对环境中的污染物开展定量溯源研究不仅是环境质量评估的重要依据,也是 有效控源、保障环境安全的基本前提。非负矩阵分解( Non-Negative Matrix Factorization,NMF)因其能够确保所分解矩阵的非负特性而被应用在环境数据 的溯源研究中。作为因子分析模型的改进手段,NMF不仅能够依据数学推导求 得污染源的指纹图谱,还能够对因子载荷矩阵和得分矩阵作非负性约束,进而解 析出具有实际物理意义的污染源贡献率及因子得分。尽管 NMF在众多定量源解 析方法中是一种相对先进的源解析求解模型,但在其实际运用过程中,仍然要使 用参数估值法来确定源贡献率的数值。且 NMF在依据随机矩阵的梯度下降方向 做多次迭代优化后,常常会陷入到模型的局部极值。鉴于此,本论文试图克服现 阶段 NMF源解析技术的缺陷,提出一种新的基于 NMF的源解析方法(Non- Negative Matrix Factorization NEW,NMF-NEW)。通过改进 NMF模型的初始化 方法及其运算模式,获得全局最优的解算数值,使源解析结果更为可靠。在此基 础上,对 NMF-NEW模型进行稳定性测试,并将其应用到海河流域最重要的五 大水系之一——子牙河的多环芳烃溯源研究之中,阐明该水系的主要污染来源类 型及其贡献率。本论文的主要贡献及研究结论如下: (1)针对现有定量源解析技术的缺陷,提出了一种新的基于 NMF的源解析 方法,该方法具有以下优点:将因子分析模型和 NMF模型之优点相结合,通过 对初始浓度矩阵进行归一化处理,解决了不同量纲之间的匹配问题,确保了矩阵 分解结果的可靠性;将矩阵分解的无约束优优化问题转化为约束优化问题,同时 对目标函数进行非负约束,以确保分解而得的因子载荷矩阵和污染源贡献率矩阵 均为非负;通过计算协方差矩阵的特征值和累计方差贡献率,合理的估计出显著 污染源数目;对归一化后的浓度数据协方差矩阵进行 SVD分解,并将该分解结 果作为运算初始值,使初始结果接近目标函数的极值结果,保证了本文所提出的 新方法在确保取得全局最优解的同时,也能够加速迭代收敛速率。 (2)以 2014年黄浦江表层沉积物多环芳烃浓度数据为输入源,开展了上述 源解析新方法的稳定性测试。将本方法的源解析结果与 EPA-PMF 5.0模型的运 行结果进行比对分析,以验证其可靠程度。由本文所提新方法解析而得的主要源 及其贡献率依次为交通源 61.3%、燃煤及生物质燃烧源 26.7%、炼焦源 12.0%。 经 EPA-PMF 5.0运行而得的三位主要源及其贡献率依次为交通源 57.6%、燃煤和 生物质燃烧源 32.1%、炼焦源 10.3%。两种源解析模型的解析而得的污染源类别 及主次顺序较为一致,这表明本文所提新方法相具有很好的稳定性。两种源解析 方法均表明黄浦江表层沉积物中的多环芳烃主要来源自于交通源排放以及燃煤 / 生物质的燃烧。 (3)对子牙河流域沉积物中多环芳烃的源解析研究发现,NMF-NEW和 PMF 得到的污染源指纹谱基本一致。其中,NMF-NEW对沉积物中多环芳烃主要来源 的判定结果依次为石油类燃烧源 35.28%,煤炭类燃烧源 31.5%,交通类排放源 21.95%,生物质燃烧源 11.27%;与之对应 PMF得到的各污染源及贡献率依次为 石油类燃烧源 39.42%,煤炭类燃烧源 30.78%,交通类排放源 23.05%,生物质燃 烧源 6.75%。此溯源结果与河北省能源结构数据相一致,揭示了化石类燃烧源和 交通排放源是导致子牙河流域多环芳烃污染的主要原因。控煤并提倡清洁能源的 使用或许是控制该地区水环境中有机毒物污染进一步恶化的重要措施。 NMF- NEW在应用过程中,无需人为计算数据集的不确定度,而将数据预处理与源解 析流程融合为同一个算法框架,剔除了冗余内存。NMF-NEW极大的提升了现有 源解析方案的效率和精度,且其操作简便,更易于推广应用。
英文摘要: Quantitative source apportionment of environmental pollutants is not only an important basis for the assessment of environmental quality, but also the basic premise for the effective control of pollution source. Non-negative matrix factorization (NMF) can be applied in the study of source apportionment of environmental data because of its non-negative characteristic. As an improved means of factor analysis model, NMF can not only work out the fingerprint of pollution source by mathematical derivation, but also make nonnegative constraints on the loading matrix and score matrix. Furthermore, the contribution rate of pollution source and the factor score with physical significance can be analyzed. Although NMF is a relatively advanced source apportionment model in many quantitative source apportionment methods, it still needs to use some parameters to estimate the contribution rate of pollution source. Besides, NMF model often falls into the local extremum after it iterates many times based on the gradient descent direction of the random matrix. In view of this, this paper attempts to overcome the shortcomings of the current NMF source apportionment technology, and proposes a new source apportionment method based on NMF algorithm. By improving the initialization method and the operation mode of the NMF model, the global optimal solution is obtained. In this way, the source apportionment results can be more reliable.Test the stability of the NMF-NEW model and then apply it to the study of source apportionment of PAHs in the Ziya River, one of the five most important rivers in Hai River basin, to expound the main pollution sources and their contribution rate. The main contributions and conclusions of this paper are as follows: (1) This paper proposed a new source apportionment method based on NMF alg- orithm. The new method has the following advantages: It can not only combine the superiorities of the factor analysis model and the NMF model, but also solve the matching problem between different dimensions by normalizing the initial concentration matrix. The scheme above ensures the reliability of the matrix decomposition results; Transformed the unconstrained optimization problem into a constrained optimization problem. And then a non-negative constrain to objective function is used to ensure that the factor loading matrix and the pollution contribution rate matrix are both non-negative; Estimated the number of significant pollution sources reasonably by calculating the eigenvalues of the covariance matrix and the cumulative variance contribution rate; The covariance matrix of the normalized concentration data is decomposed by SVD, and the decomposition result is used as the initial value of the operation. So that the initial results are close to the extreme value of the objective function, which guarantees that the new method proposed in this paper can accelerate the iterative convergence rate while ensuring the global optimal solution. (2) Test the stability of the new source apportionment method proposed in this pa- per. In order to verify the reliability of the new method, use the concentration datas of PAHs in surface sediments of Huangpu River as inputs to compare the source apportionment results of the new method with that of the EPA-PMF 5.0 model. The main pollution sources and its contribution rates calculated by the new method were traffic source(61.3%), coal combustion and biomass burning source(26.7%) and coke refining source(12%). The three main sources and its contribution rates were obtained by EPA-PMF 5.0, which were traffic source(57.6%), coal combustion and biomass burning source(32.1%) and coking refining source(10.3%). There were slight differences in the final contribution rate of the two source apportionment models, however, the category and the primary and secondary order of pollution sources are more consistent. The test results showed that the new method proposed in this paper was very stable. The source apportionment results of the two methods both indicated that PAHs in the surface sediments of Huangpu River mainly come from the emission of the traffic source and the combustion of coal and biomass. (3) The source apportionment results of PAHs in sediments of the Ziya River Bas- in showed that the fingerprints of pollution source identified by NMF-NEW and PMF were consistent. The main pollution sources and its contribution rates analyzed by NMF-NEW model were petroleum combustion source(35.28%), coal combustion source (31.5%), traffic emission sources(21.95%) and biomass combustion source(11.27%). Correspondingly, the pollution sources and its contribution rates analyzed by PMF were petroleum combustion source (39.42%), coal combustion source(30.78%), traffic emission sources(23.05%) and biomass combustion source(6.75%). The traceability results were highly consistent with the datas of energy structure in Hebei Province, which revealed that the main source of PAHs pollution in the Ziya River Basin were the fossil burning sources and traffic emission sources. To control the use of coal and promote the use of clean energy may be an very important measure for preventing further deterioration of organic pollutants in the water environment in this region. In the application process,the NMF-NEW model needn’t to calculate the uncertainty of the data set. The data preprocessing and source analysis process were fused into the same algorithm framework, which eliminates redundant memory. NMF-NEW model greatly improved the efficiency and accuracy of existing source resolution schemes, and was easier to operate and popularize.
内容类型: 学位论文
URI标识: http://ir.rcees.ac.cn/handle/311016/38621
Appears in Collections:环境水质学国家重点实验室_学位论文

Files in This Item:
File Name/ File Size Content Type Version Access License
高泽晋--基于非负矩阵分解的源解析模型优化及其在子牙河沉积物PAHs溯源研究中的应用.pdf(3077KB)学位论文--限制开放 联系获取全文

作者单位: 1.中国科学院生态环境研究中心

Recommended Citation:
高泽晋. 基于非负矩阵分解的源解析模型优化及其在子牙河沉积物PAHs溯源研究中的应用[D]. 北京. 中国科学院大学. 2017.
Service
Recommend this item
Sava as my favorate item
Show this item's statistics
Export Endnote File
Google Scholar
Similar articles in Google Scholar
[高泽晋]'s Articles
CSDL cross search
Similar articles in CSDL Cross Search
[高泽晋]‘s Articles
Related Copyright Policies
Null
Social Bookmarking
Add to CiteULike Add to Connotea Add to Del.icio.us Add to Digg Add to Reddit
所有评论 (0)
暂无评论
 
评注功能仅针对注册用户开放,请您登录
您对该条目有什么异议,请填写以下表单,管理员会尽快联系您。
内 容:
Email:  *
单位:
验证码:   刷新
您在IR的使用过程中有什么好的想法或者建议可以反馈给我们。
标 题:
 *
内 容:
Email:  *
验证码:   刷新

Items in IR are protected by copyright, with all rights reserved, unless otherwise indicated.

 

 

Valid XHTML 1.0!
Copyright © 2007-2018  中国科学院生态环境研究中心 - Feedback
Powered by CSpace