基于监测数据挖掘的库区滑坡表面变形模式及时空规律研究

项目名称： 基于监测数据挖掘的库区滑坡表面变形模式及时空规律研究

项目编号： No.41302260

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 天文学、地球科学

项目作者： 黄海峰

作者单位： 三峡大学

项目金额： 23万元

中文摘要： 基于监测数据定量化研究滑坡表面变形模式及时空规律，对深化水库滑坡稳定性评价、预测预报研究及提升防灾减灾水平具有重要意义。针对目前缺少区域性大量滑坡的完整监测数据以及有效的数据挖掘方法，导致相关研究在系统性和全面性上仍显不足的现状，本项目拟基于申请者长期承担三峡库区滑坡专业监测任务所积累的大量监测数据，以三峡库首秭归、兴山段为研究区，以专业监测滑坡的地表位移、降雨、库水位变动等监测数据为核心，综合采用包括曲线聚类、运动相关性、遗传算法、支持向量机、地理信息系统等数据挖掘方法，定量化分析并系统建立库区滑坡表面变形模式分类，进而从宏观区域上挖掘不同模式滑坡的变形时间和空间分布规律，同时针对典型单体滑坡在内外因综合作用下的表面变形开展时空耦合预测及规律研究。项目点面结合，将系统而全面地揭示库区滑坡变形的主要特征过程以及内外因驱动下的时空规律，具有重要的现实意义和应用价值。

中文关键词： 滑坡监测；数据挖掘；变形模式；时空规律；三峡库区

英文摘要： Landslides monitoring data objectively reflects the process of deformation and changes of main influencing factors, by use of which to mining and analyze surface deformation patterns and spatial-temporal characteristics for reservoir landslides, which is an important means to deepen related research such as the stability evaluation and prediction of reservoir landslides, as well as progress the level of geo-hazards prevention and mitigation in reservoir area. However, nowadays the systematic and comprehensive studies are still deficient because of lacking integrated monitoring data of a large number of landslides and lacking effective data mining methods.Taking into account the rich monitoring data have been accumulated in the process of undertaking landslides professional monitoring works in Three Gorges reservoir area, this project aims to base on the core landslide monitoring data include surface displacement, reservoir water level changes and rainfall,etc.,and use a variety of data mining methods include curve clustering, dynamic interaction analysis, genetic algorithm, support vector machine and geographic information system, etc.,to quantitative analyze and establish the surface deformation patterns classification of reservoir landslides, and then to find the spatial-temporal characteristics for the macro-

英文关键词： Landslide monitoring；data mining；deformation patterns；spatial-temporal characteristics；the Three Gorges Reservoir area