基于基因表达式编程的大坝位移强度聚类分析模型研究

发布时间：2018-03-11 17:27

本文选题：大坝变形　切入点：基因表达式编程　出处：《江西理工大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着大坝变形监测技术的发展，大坝变形监测数据的获取也越来越多样化。但是其数据分析方法仍然是针对单点的预测，无法从整体的角度对大坝的变形进行分析。人们迫切需要一种可以从宏观的角度快速分析大坝整体变形规律的方法。由此可见，对大坝进行位移强度聚类分析，找出大坝的整体变形规律具有重要的意义。针对这种情况，本文将改进的基因表达式自动聚类算法应用于大坝位移强度的聚类分析中，主要研究工作如下：（1）对引入的位移强度进行了深入的分析，结合大坝变形监测数据的特点，选择大坝所有变形监测点在监测时段内的平均位移速率作为临界速率，并以此为基准确定大坝的位移强度值；（2）针对大坝变形监测数据所含的突变的噪声可能对聚类分析产生不利的影响，采用小波阈值去噪算法对原始的大坝变形监测数据进行噪声的去除；（3）为了消除大坝监测点分布不均及监测点之间较大的空间间隔所导致的大坝整体变形规律分析中存在分析的盲点，在坝体表面生成1m*1m的空间格网均匀的覆盖坝体，采用径向基插值算法对格网点进行位移强度插值，以此反映大坝的整体变形情况；（4）针对现有的基因表达式自动聚类算法在高维度空间数据聚类方面的不足，对现有的算法进行了改进。提出基于主成分的基因表达式自动聚类算法，并将高维度的大坝变形监测数据映射到低维度空间，通过降维实现大坝变形监测数据的聚类操作；（5）采用.NET与Matlab混合编程技术，实现所提出的基于基因表达编程的大坝位移强度聚类分析模型；（6）将基于基因表达式编程的大坝位移强度聚类分析模型应用于江西省赣州市上犹江水库大坝的变形监测工程中，并对该大坝的整体变形规律进行过分析。通过分析可以清楚地看出，，上犹江水库大坝坝体下游的位置形变活动比较稳定，上游的位置变形活动比较剧烈，中间位置形变活动居于二者之间；大坝的左岸变形活动比较稳定，右岸变形活动比较频繁；且水流方向的形变活动要强于大坝轴线方向和垂直位移方向。将位移强度聚类可视化效果图与各个方向的位移强度值三维效效图进行比较分析，可以发现聚类的结果与大坝的变形情况基本上是相符的，由此也证明了所提出的基于基因表达式编程的大坝位移强度聚类分析模型的聚类分析结果基本上是可靠的。
[Abstract]:With the development of dam deformation monitoring technology, the acquisition of dam deformation monitoring data is becoming more and more diversified. It is not possible to analyze the deformation of the dam from the whole angle. People urgently need a method to analyze the deformation law of the dam quickly from the macro angle. Thus, the displacement intensity cluster analysis of the dam is carried out. It is very important to find out the global deformation law of the dam. In this paper, the improved gene expression automatic clustering algorithm is applied to the clustering analysis of the dam displacement intensity. The main research work is as follows:. 1) based on the analysis of the displacement intensity introduced, combined with the characteristics of the dam deformation monitoring data, the average displacement rate of all the dam deformation monitoring points during the monitoring period is selected as the critical rate. The displacement strength of the dam is determined by this criterion. 2) aiming at the sudden noise in dam deformation monitoring data, the wavelet threshold de-noising algorithm is used to remove the noise from the original dam deformation monitoring data. 3) in order to eliminate the blind spot in the analysis of dam deformation law caused by the uneven distribution of monitoring points and the large space interval between the monitoring points, the dam body is uniformly covered by a space grid of 1mmm or 1m on the surface of the dam. The radial basis interpolation algorithm is used to interpolate the displacement intensity of the lattice dot to reflect the whole deformation of the dam. 4) aiming at the deficiency of the existing automatic clustering algorithm of gene expression in high-dimensional spatial data clustering, this paper improves the existing algorithm, and proposes an automatic clustering algorithm of gene expression based on principal component. The high-dimensional dam deformation monitoring data are mapped to the low-dimensional space, and the clustering operation of dam deformation monitoring data is realized by reducing the dimension. In this paper, the hybrid programming technology of .NET and Matlab is used to realize the dam displacement intensity cluster analysis model based on gene expression programming. The model of dam displacement intensity cluster analysis based on genetic expression programming is applied to the dam deformation monitoring project of Shangyou River Reservoir in Ganzhou City Jiangxi Province and the whole deformation law of the dam is analyzed. It can be clearly seen from the analysis that the deformation activity in the lower reaches of the dam body of Upper Jujiang Reservoir is relatively stable, the deformation activity in the upper reaches is more intense, and the deformation activity in the middle position is between the two. The deformation activity of the left bank of the dam is relatively stable, and the deformation activity of the right bank is more frequent. The deformation activity in the direction of flow is stronger than that in the direction of the axis and vertical displacement of the dam. The visual effect map of displacement intensity clustering is compared with the three dimensional effect chart of displacement intensity in each direction. It can be found that the clustering results are basically consistent with the deformation of the dam, which also proves that the clustering analysis results of the proposed clustering model of dam displacement strength based on genetic expression programming are basically reliable.
【学位授予单位】：江西理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TV698.11

【参考文献】