【作者】 刘光艳；

【导师】 王洪鉴；

【作者基本信息】 山东师范大学 ， 分析化学， 2003， 硕士

【摘要】 开展本研究的主要目的是为了适应日益深化的城市环境监测与管理的需要，寻求建立最优化的大气环境监测网络和大气环境质量分析与评价模型，以使环境工作者能够做到正确的评价环境质量，科学的制定环境评价标准；准确地提供环境监测报告、预报，对环境进行现代化的管理等。 本文主要讨论环境监测点位优化布设的若干问题，并针对滨州市的环境监测网络的实际情况确立了本市例行监测点位优化布设的理论基础、原则和方法，提出了空气监测优化布点的实施方案。监测点位布设最优方案的选择，其目的在于以最少的监测点位和工作量来获得具有最大代表性的和完整的、系统的可靠的大气环境监测信息。目前用于大气环境监测点的优化方法较多，各有特色，而模糊聚类法作为多目标决策的一种优选方法，有着独到之处。本文尝试着将该方法进行适当改进，运用最优分割法对样品进行分类，采用模糊矩阵最大元法对各个监测点位的原始数据进行检验、置信水平排序、最优化分级、模糊相关分析以及优化结果检验，从而最终确定优化点位。该方法使用简便，模型直观，分级自然，避免了诸多人为因素，符合空气监测网格化布点的要求，能反映空气污染的最高点、均值点、趋势点和对照点，同时也能反映优化后点位的整体代表性，这是本文的研究内容之一。 同时，本文分析了大量的业已存在的研究成果，首次提出了一种空气质量评价的新方法，即空气质量降维B-P神经网络评价法，在传统B-P神经网络的基础上，应用主成分分析将多维空间的样本数据降维到低维空间上，然后建立B-P网络评价模型。通过对比分析，证明这种基于降维技术和B-P神经网络相结合的方法用于解决非线性多维体系的环境质量评价问题是行之有效的。该研究是基于现实的问题提出来的，其解决方法进一步丰富和发展了大气环境质量分析与评价模型，同时，也进一步的丰富了B-P神经网络在大气环境质量分析与评价中的应用。这是本文的研究内容之二。更多还原

【Abstract】 The main purpose of this investigation is for the development of environmental monitoring and the requirement of city administration, to establish the optimized environmental monitoring network and the model of environmental quality assessment, to help the environmental researcher to make a correct environmental quality assessment, scientifically define the environmental standard, accurately provide environmental monitoring report and prediction, and administrate the environment by modern method.This thesis was centered on the optimization of environmental monitoring site. According to the environmental monitoring network of Binzhou city, the theoretical basis, principle and method of the optimized routine monitoring site in our city were established; the project of air monitoring optimization site was also proposed. The aim of selecting the optimized monitoring site is to obtain the most representative, complete, systematic and responsible air environmental monitoring information by the least monitoring sites and workload. At present there are many methods on the site optimization of air environmental monitoring and each has its characteristics. Fuzzy classification, as an optimized multi-goal decision-making method, has its specific characteristics. The article tries to improve the method properly, classifies the samples by using optimized separating method, adopt to test each site’ s original measurements, arrange reliability , optimizedly classify, analyze obscure relations and examine the optimized results. Thus the optimized sites are finally determined. The method is easy to use, and has directly perceived model and natural classification. It avoids many subjective factors, accords with the demands of air environmental monitoring site, and the optimized site can represent the whole. This is one of thestudying contents of the article.The thesis also analyzed a lot of existing research results and first proposed a new method of assessing air quality, i. e. the reducing dimension B-P neural network assessment method. Based on traditional B-P neural network, this method reduced the sample data of multi-dimensional space to low-dimensional one by main component analysis, and then established the B-P network assessment model. The contrast analysis proved that this method of combining reducing dimension technology and B-P neural network was effective for the solution of environmental monitoring assessment problem in non-linear multi-dimensional system. The project based on the realistic problem, has very important application and promotion for the development of environmental monitoring and intensification of environmental administration. Moreover, the method enriches and developes the analysis and assessment model of air environment quality; and also extends the application of B-P neural network in analysis and assessment of air environment quality. This is the other studying contents of the article.更多还原