基于证据理论的机场噪声监测数据可信度分析

发布时间：2018-05-09 00:20

本文选题：证据理论 + 评估模型　；参考：《南京航空航天大学》2014年硕士论文

【摘要】：随着机场运输规模不断扩大，机场噪声污染日益恶化，安装机场噪声监测系统已成为众多机场监测周边噪声环境的重要手段。但是，这类监测系统中的固定监测点成本高、环境要求高、稳定性差，加之机场噪声的监测数据中夹杂有航空器之外的其他噪声源产生的环境噪声（如风噪、施工噪声等）数据，因此评估各监测点所监测到的噪声数据的可信度就变得尤为重要。本文在充分分析机场噪声监测数据特点的基础上，提出一种基于证据理论的机场噪声监测数据可信度评估模型。该模型利用数据挖掘的方法生成监测点间噪声数据的关联规则，然后利用关联规则进行基本概率赋值函数的获取，将获得的证据用Dempster组合规则进行融合并作出最终的决策。针对Dempster组合规则在冲突证据融合方面的缺陷，在充分研究了冲突证据度量标准的基础上，本文提出使用pignistic概率函数来衡量证据之间的冲突程度，，并在此基础上构建了伪证据识别方法和冲突证据融合方法。将这些方法应用于机场噪声监测数据可信度评估模型中，使该模型在有冲突证据的情况下不但能识别出冲突证据，还能利用改进的合成方法得到正确的融合结果。实验表明本文提出的伪证据识别方法和冲突证据融合方法能有效地在机场噪声监测数据中得到较好的效果，同时改进的评估模型应用在真实的机场噪声监测数据中具有较好的准确性和实用性。
[Abstract]:With the expansion of airport transportation scale and the worsening of airport noise pollution, the installation of airport noise monitoring system has become an important means to monitor the surrounding noise environment of many airports. However, the fixed monitoring points in this type of monitoring system have high cost, high environmental requirements, poor stability and environmental noise (such as wind noise) caused by noise sources other than aircraft in the monitoring data of airport noise. Therefore, it is very important to evaluate the reliability of the noise data from the monitoring points. Based on the analysis of the characteristics of airport noise monitoring data, a reliability evaluation model of airport noise monitoring data based on evidence theory is proposed in this paper. The model uses the method of data mining to generate the association rules of noise data between monitoring points, and then uses the association rules to obtain the basic probability assignment function. The obtained evidence is fused with the Dempster combination rule and the final decision is made. Aiming at the defect of Dempster combination rule in conflict evidence fusion, this paper proposes to use pignistic probability function to measure the conflict degree of evidence. On this basis, the pseudo-evidence recognition method and conflict evidence fusion method are constructed. These methods are applied to the reliability evaluation model of airport noise monitoring data. The model can not only identify the conflicting evidence but also obtain the correct fusion results by using the improved synthesis method. The experimental results show that the methods of pseudo-evidence recognition and conflict evidence fusion proposed in this paper can effectively obtain good results in airport noise monitoring data. At the same time, the improved evaluation model has good accuracy and practicability in real airport noise monitoring data.
【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】：X839.1;TB53

【参考文献】