海量数据挖掘大气颗粒物成分分析系统的设计与实现

发布时间：2018-05-21 04:01

本文选题：大气颗粒物 + 质谱仪　；参考：《中国科学院研究生院（沈阳计算技术研究所）》2015年硕士论文

【摘要】：大气颗粒物的成分对人类的身体以及生活环境、大气的能见度、城市交通以及全球环境问题都具有很大的影响,尤其是随着近几年我国空气污染情况日益严重,因而越来越受到人们的广泛重视。传统的大气颗粒物分析手段主要是依靠整体颗粒物分析的技术以及人工识别颗粒物种类及来源的方法,但是这些手段耗时长、人工成本高、准确率低,没有办法满足目前人们的需求。本文的目的是依托目前已有的颗粒物质谱仪,开发出一套可以满足应用需求的、自动的、实时的大气颗粒物成分分析系统。质谱仪具有强大的数据收集能力,每天可以收集到海量的大气颗粒物。通过分析和研究,我们使用了聚类算法将具有相似性质的颗粒物聚集到一个分组中,那么后面的处理就可以以分组为单位进行,达到了系统实时性的要求。为了达到系统自动性的要求,我们使用了逻辑回归算法对颗粒物分组进行自动命名,通过人工对训练样本的命名训练了模型参数,使得模型可以对测试样本进行分类,减少了人工对系统的干预程度。颗粒物命名完成后,我们可以对大气中的颗粒物组成成分进行分析,全面的了解监测时间、监测地点的空气质量状况。
[Abstract]:The composition of atmospheric particulates has great influence on human body and living environment, visibility of atmosphere, urban traffic and global environmental problems, especially with the increasingly serious air pollution in China in recent years. As a result, people pay more and more attention to it. The traditional analysis methods of atmospheric particulate matter mainly rely on the technology of whole particulate matter analysis and the methods of manually identifying the kinds and sources of particulate matter, but these methods take a long time, have high labor cost, and have low accuracy. There is no way to meet the current needs of the people. The purpose of this paper is to develop an automatic, real-time analysis system of atmospheric particulate matter composition, which can meet the needs of application based on the existing particle mass spectrometer. Mass spectrometer has powerful data collection ability, can collect massive atmospheric particulate matter every day. Through analysis and research, we use clustering algorithm to cluster particles with similar properties into a group, and then the subsequent processing can be carried out in units of the group, which meets the real-time requirements of the system. In order to achieve the requirement of automatic system, we use the logical regression algorithm to name the particle groups automatically, and train the model parameters by artificial naming of the training samples, so that the model can classify the test samples. The degree of manual intervention to the system is reduced. After the designation of particulate matter, we can analyze the composition of particulate matter in the atmosphere, understand the monitoring time and monitor the air quality of the site.
【学位授予单位】：中国科学院研究生院（沈阳计算技术研究所）
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：X513

【参考文献】