陈波,詹明强,黄梓莘. 基于时空聚类挖掘的库岸边坡位移监测数据约简[J]. 水利水运工程学报,2022(5):94-101. DOI: 10.12170/20210728002
引用本文: 陈波,詹明强,黄梓莘. 基于时空聚类挖掘的库岸边坡位移监测数据约简[J]. 水利水运工程学报,2022(5):94-101. DOI: 10.12170/20210728002
(CHEN Bo, ZHAN Mingqiang, HUANG Zishen. Reduction of multi-point displacement monitoring data of reservoir bank slope based on spatio-temporal clustering mining[J]. Hydro-Science and Engineering, 2022(5): 94-101. (in Chinese)). DOI: 10.12170/20210728002
Citation: (CHEN Bo, ZHAN Mingqiang, HUANG Zishen. Reduction of multi-point displacement monitoring data of reservoir bank slope based on spatio-temporal clustering mining[J]. Hydro-Science and Engineering, 2022(5): 94-101. (in Chinese)). DOI: 10.12170/20210728002

基于时空聚类挖掘的库岸边坡位移监测数据约简

Reduction of multi-point displacement monitoring data of reservoir bank slope based on spatio-temporal clustering mining

  • 摘要: 库岸边坡失稳会对工程自身效益和周边安全造成巨大损失,而位移监测数据可以直接表征库岸边坡安全状况。传统变形位移分析一般仅考虑单个监测点,不同监测点之间位移的相似性和关联性有待挖掘。基于时空数据挖掘领域的聚类方法,综合考虑测点属性和空间特征,采用K-means算法度量测点间的相似程度,实现变形区域划分;在变形区域划分基础上,采用遗传算法优化的投影聚类算法,将高维数据向低维空间映射,通过提取测点数据特征,筛选得到重点关注的测点和压缩数据量。经实例工程数据验证,时空聚类挖掘方法便捷、有效,逐步实现了边坡位移监测数据约简,可用于类似库岸边坡的监测数据挖掘。

     

    Abstract: The instability disaster of reservoir bank slope will cause huge losses to the benefit of the project and the safety of surrounding life and property, and the displacement monitoring data can directly characterize the safety status of reservoir bank slope. In view of the traditional deformation and displacement analysis, only a single monitoring point is considered, and the similarity and relevance of displacement between different monitoring points still need to be excavated. Based on the clustering method in the field of spatio-temporal data mining, considering the attribute characteristics and spatial characteristics of measuring points, K-means algorithm is used to measure the similarity between measuring points and realize the division of measuring points; based on the division of measuring points, the projection clustering algorithm optimized by genetic algorithm is used to map the high-dimensional data to the low-dimensional space. By extracting the characteristics of measuring point data, the purpose of screening the measuring points needing attention and compressing the data order is achieved. Based on practical engineering data, it is shown that the spatio-temporal clustering mining method is convenient and effective, and gradually reduces the monitoring data of slope displacement. The method can be used for monitoring data mining of similar reservoirs.

     

/

返回文章
返回