机场噪声多监测点噪声值关联分析

发布时间：2018-06-05 18:58

本文选题：机场噪声 + 关联规则　；参考：《南京航空航天大学》2014年硕士论文

【摘要】：近年来随着民航事业的迅速发展，机场噪声越来越成为人们关注的话题。借助机场噪声监测点数据集挖掘机场噪声分布情况和发展趋势，才能为有效地预测并采取有效措施防治飞机噪声污染和有计划地开发利用机场及周边区域的土地提供重要的科学依据。本文主要研究的是机场噪声监测点噪声值之间的关联规则。关联规则是目前数据挖掘与知识发现的主要研究内容之一，侧重于确定数据中不同属性之间的联系，找出满意约定支持度(Support)和信任度(Confidence)阈值的多个属性之间的依赖关系。自1993年R.Agrawal，R.Srikant首次提出该问题来，已经出现了许多关联规则挖掘算法。本文的工作主要分为以下几个方面： 1)研究当前主要的关联规则挖掘算法。研究当前主要的关联规则挖掘算法，并分析总结各种算法存在的优势和不足。 2)研究机场噪声监测点数据集的特点。根据监测点数据集的特点，本文采用DENCLUE算法对监测点数据集进行聚类，并根据改进的爬山算法找出每一类中的代表点。 3)分析机场噪声监测点数据集的影响因素。监测点噪声值主要与天气、温湿度、航迹、跑道、机型等信息密切相关。本文采用了灰色关联度分析法对同一航班不同机型的噪声影响进行分析，并进行了实验验证。结果表明，灰色关联度分析方法在对于影响噪声值条件的选择具有可行性。 4)建立关联规则的初始集。首先根据噪声值的分贝数，删除那些非噪声事件的噪声值，然后再根据航迹，删除非航迹上的监测点的噪声值，最后采用等分法将噪声值转变为噪声值区间。 5)挖掘监测点噪声值之间的关联规则。本文应用Apriori算法挖掘关联规则，并指出了Apriori算法的不足，在此基础上提出了ATNSOA-Apriori算法，最后将两者进行了实验对比。结果表明ATNSOA-Apriori算法更有效率。 6)关联规则的数量的回归分析。利用回归分析法设计几种回归方程，然后利用复相关系数检验各个回归方程的拟合效果，，最后用显著性检验来验证参数的系数是否显著为零。本文采用复相关系数最大的回归方程作为最优方程来预测给定参数下的关联规则的数量。
[Abstract]:In recent years, with the rapid development of civil aviation, airport noise has become a topic of concern. With the aid of the data set of airport noise monitoring points, the noise distribution and development trend of excavator field are analyzed. It can provide an important scientific basis for effectively predicting and taking effective measures to prevent and control aircraft noise pollution and to develop and utilize the land in the airport and its surrounding areas in a planned way. This paper mainly studies the association rules between noise values of airport noise monitoring points. Association rule is one of the main research contents of data mining and knowledge discovery at present. It focuses on determining the relationship between different attributes in data, and finds out the dependencies between several attributes of satisfactory agreement support and confidence degree Confidence. Since R. Agrawaln R. Srikant first proposed this problem in 1993, there have been many association rules mining algorithms. The work of this paper is divided into the following aspects: 1) the main algorithms for mining association rules are studied. The main algorithms for mining association rules are studied, and the advantages and disadvantages of these algorithms are analyzed and summarized. 2) the characteristics of airport noise monitoring data set are studied. According to the characteristics of monitoring data set, this paper uses DENCLUE algorithm to cluster the data set of monitoring points, and finds out the representative points in each class according to the improved mountain climbing algorithm. 3) analyze the influence factors of airport noise monitoring data set. The noise of monitoring points is closely related to weather, temperature and humidity, track, runway, type and so on. In this paper, grey correlation analysis method is used to analyze the noise effect of different types of aircraft on the same flight, and the experimental results are verified. The results show that the grey correlation degree analysis method is feasible for the selection of the conditions affecting the noise value. 4) establishing the initial set of association rules. The noise values of those non-noise events are deleted according to the decibels of the noise values, and then the noise values of the monitoring points on the non-track are deleted according to the track. Finally, the noise values are transformed into the noise value intervals by using the equipartition method. 5) mining association rules between noise values of monitoring points. In this paper, Apriori algorithm is applied to mining association rules, and the deficiency of Apriori algorithm is pointed out. On this basis, the ATNSOA-Apriori algorithm is proposed, and finally, the experimental comparison between the two algorithms is carried out. The results show that ATNSOA-Apriori algorithm is more efficient. 6) regression analysis of the number of association rules. Regression analysis is used to design several regression equations, and then the complex correlation coefficient is used to test the fitting effect of each regression equation. At last, the significance test is used to verify whether the coefficients of the parameters are significantly zero. In this paper, the regression equation with the largest complex correlation coefficient is used as the optimal equation to predict the number of association rules under given parameters.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：V351;TB53

【参考文献】