南京林业大学[017].doc

PAGE \* MERGEFORMATI 南京林业大学 本科毕业设计（论文） 题 目：沉降观测数据分析及可视化 方法探讨 学 院： 土木工程学院 专 业： 测绘工程 学 号： 070604221 学生姓名： 张永磊 指导教师： 陈健 职 称： 讲师 二〇一一年五月二十九日 摘要 沉降监测是建筑物变形监测中一项重要的监测内容。沉降监测对保证建筑物的安全有重要意义，通过对沉降数据的分析可获知沉降原因及对未来沉降趋势进行预报。针对沉降数据序列的特征，本文分析、研究了数据处理及信息挖掘的基本方法，结合GIS软件，通过实例讨论沉降数据分析的可视化问题，为决策提供支持。 在沉降数据序列预处理方面，本文研究了数据的插补、重采样、滤波与奇异值剔除等的处理方法。在沉降数据信息挖掘方面，分析对比了小波降噪前后多项式回归方法对沉降序列趋势项的提取效果,统计指标对比表明: 在趋势项提取过程中，小波降噪后多项式回归方法提取趋势项优于直接用多项式回归方法；研究了频谱分析法在信息挖掘中的作用，对残差和小波分解的细节分量应用频谱分析表明了文中例证沉降序列中含有周期项信息；研究应用了AR(P)模型对沉降数据进行预报，实验表明预报结果能满足工程需要。初步研究了沉降可视化方法,应用GIS优秀的管理和分析空间数据功能，实现了沉降数据的可视化，达到了研究的目的。 关键词:沉降数据；预处理；信息挖掘；可视化  Abstract Subsidence monitoring is an important monitoring in the content of deformation monitoring. Subsidence?monitoring?is significant to ensure?the safety of buildings, we can be informed of the reason of settlement through the analysis of settlement data, and also we can make prediction of the?future?trends of the?settlement. Based on the characteristics of data series of the?settlement, the paper researches the basic methods of data processing and information mining. With the help of GIS software, the paper also discusses the method of settlement data visualization. In the term of data preprocessing, this paper researches method of data interpolation, method of resample and so on. In the term of information mining, this paper compares the effect?of the polynomial fitting??in the trend?removal?of settlement before and after wavelet de-noising ; Indicators of statistical comparison show that:??using polynomial regression?method after wavelet de-noising is much better. This paper also uses Fourier transform?methods to identify?periodic data sequence, then uses AR (P)?model?to predict the?settlement, and the forecast results can meet the engineering?needs.?Finally with the help of GIS?software, using its excellent?spatial data?management and analysis?capabilities, this paper ac