面向城市空气质量参与式感知的关联规则挖掘研究.docx

1、 硕 士 专 业 学 位 论 文面向城市空气质量参与式感知的关联规则挖掘研究专业名称 ： 电子与通信工程面向城市空气质量参与式感知的关联规则挖掘研究I摘要随着经济的高速发展，城市化及化石能源消耗不断增长，空气质量的实时监测和潜在成因及规律发现得到了广泛关注。当今智能手机应用的普及和强大的传感器功能，使得参与式感知（Participatory sensing）作为一种新的感知模式应用于城市空气质量监测成为可能，人们能够参与到城市感知中，更好地了解周围的空气状况。与传统空气质量监测方式不同的是，参与式感知以人为中心，提供了一种灵活、低成本代价、覆盖范围广且细粒度的空气质量监测方式。然而在感知城市空

2、气质量过程中，数据采样是影响智能手机端传感器性能的关键问题之一，采样策略需结合当前空气质量状况适当的进行调整，才能起到智能手机端节能的目的。因此，本文基于数据挖掘中的Apriori 关联挖掘算法，根据参与者贡献的历史空气质量数据，挖掘空气质量蕴含的规则模式，为智能手机端的自适应采样参数调整及城市空气污染防治提供依据，本文的具体工作如下：1）阐述了城市空气质量参与式感知系统的基本架构及其应用于城市空气质量监测的优势；针对节省智能手机端的能源消耗问题，强调了挖掘空气质量隐藏的规则模式对手机端自适应采样策略调整的现实意义。2）深入分析了传统 Apriori 算法的基本原理、执行流程；针对传统 Apr

3、iori 算法在运行过程产生大量候选项及扫描事务数据库过多的缺陷，调研了现有的常见 Apriori 算法优化思路。3）研究了一种基于预判筛选多叉树的改进 Apriori 算法，通过将事务数据库转换为布尔型数据库，采用预判筛选的方式，结合矩阵运算和多叉树的优点，改善传统Apriori 算法的缺陷，并在运行过程中不断压缩数据库，极大地提升算法运行时间效率；运用 Python 语言实现了传统 Apriori 算法及改进 Apriori 算法，并通过对比实验，验证了改进的 Apriori 算法相较于原算法具有更高的运行时间效率及稳定性。4）使用微软亚洲研究院的公开数据集“Air quality dat

4、a, meteorological data and weather forecasts of 43 cities in China”模拟参与者历史贡献的空气质量感知数据，运用改进的 Apriori 算法挖掘空气质量感知数据中隐含的规则模式，并深入分析时间、空间及气象因素对空气质量的影响；基于挖掘的空气质量关联规则，完成在线空气质量预测工作，并取得较高的预测准确率。关键字：城市空气质量；参与式感知；自适应采样；Apriori 算法；关联规则硕士学位论文IIAbstractThe real-time monitoring of air quality and its potential caus

5、es and regularity have received widespread attention with the high-speed development of economy and the fast growth of urbanization and fossil fuel consumptions. The popularization and powerful sensor function of smart phone applications now make it possible for participatory sensing to be used as a

6、 new sensing model for urban air quality monitoring. People can participate in urban sensing to better know about the air conditions surrounding them. Different from the traditional air quality monitoring way, participatory sensing is human-centric and provides a flexible, low-cost, wide-coverage an

7、d fine-grained air quality monitoring method. However, in urban air quality participatory sensing, data sampling is one of the most important problems that affect the performance of the smart phone sensors. The sampling strategy needs to be adjusted in accordance with the current air quality conditi

8、ons in order to achieve the goal of energy saving of smart phone. Therefore, based on the Apriori association mining algorithm in data mining, this paper finds the rules of air quality based on historical air quality data contributed by participants, and provides a basis for adaptive sampling parame

9、ter adjustment in smart phones and air pollution control. The specific work is as follows:1) Elaborates the basic structure of urban air quality participatory sensing system and its advantages in urban air quality monitoring; And in order to save energy consumption of smart phones, this paper emphas

10、izes the practical significance of mining hidden rules of air quality to adjust the adaptive sampling strategy of mobile phones.2) Deeply analyzes the basic principle, implementation process of the traditional Apriori algorithm; Aiming at the its defects of generating a large number of candidate ite

11、ms and scanning the database many times during the running process, investigates its existing common improvement ideas.3) Researches an improved Apriori algorithm based on prejudging, screening and multi-branch tree. By converting the transaction database into a Boolean database, adopting a prejudgi

12、ng screening method and combining the advantages of matrix operations and multi-branch trees, the defects of the traditional Apriori algorithm are improved. Simultaneously the database is continuously compressed, greatly improving the efficiency of the algorithm running time; The traditional Apriori

13、 algorithm and the improved Apriori algorithm are implemented in Python language. And the improved Apriori algorithm is proved to have higher running time efficiency and stability than the original algorithm through comparison experiments.面向城市空气质量参与式感知的关联规则挖掘研究III4) Uses the public dataset “Air qual

14、ity data, meteorological data and weather forecasts of 43 cities in China“ provided by Microsoft Asia research institute to simulate air quality sensing data of historical contributions by participants, and uses improved Apriori algorithm to mine the implied rule patterns in air quality sensing data

15、. Then deeply analyzes the influence of time, space and meteorological factors on air quality; The online air quality prediction work is completed and achieve high prediction accuracy.Keywords: Urban air quality; Participatory Sensing; Adaptive sampling; Apriori algorithm; Association rules;硕士学位论文IV

16、目录摘要 .IAbstract.II目录 .IV第 1 章 绪论 .11.1 研究目的与意义.11.2 国内外研究现状.21.2.1 参与式感知研究现状 .21.2.2 空气质量研究现状 .51.3 主要研究内容.61.4 论文结构安排.6第 2 章 城市空气质量参与式感知与关联规则 .82.1 城市空气质量参与式感知系统.82.1.1 系统架构 .82.1.2 关联规则在系统中的作用 .92.2 关联规则概述.102.2.1 数据挖掘相关知识 .102.2.2 关联规则基本概念 .122.2.3 关联规则挖掘过程 .132.2.4 关联挖掘常见算法 .142.3 本章小结.15第 3 章 Apriori 关联挖掘算法的研究与改进设计 .163.1 Apriori 算法研究 .163.1.1 Apriori 算法思想 .163.1.2 Apriori 算法缺陷 .173.1.3 Apriori 算法常见改进策略 .183.2 改进的 Apriori 算法设计 .193.3 改进 Apriori 算法性能分析 .243.4 本章小结.28第 4 章 空气质量参与式感知数据的预处理 .294.1 实验数据集介绍.294.1.1 空气质量监测站信息 .294.1.2 空气质量及气象数据 .304.2 空气质量指数计算.30