摘要
从实测数据中分析滑坡灾害的成因机理,对于准确识别潜在危险区与及时制定防治措施十分重要。由于现场监测数据的数量庞大、来源多样,常规的数据处理方法难以从海量监测数据中提取出有用的信息,进而对滑坡变形演化趋势作出正确评价和预测。本文基于经典数据挖掘方法中的两步聚类法与关联规则分析,提出了滑坡变形行为的关联分析挖掘技术,并以长江三峡库区新铺滑坡为例,对库水位波动及降雨影响下的特大滑坡位移速率进行了关联分析。结果表明:该滑坡的变形受库水位高程水平、库水位波动速率与降雨强度等因素的多重影响,水位下降、强降雨与滑坡变形密切相关;滑坡不同空间位置处的变形影响因素存在差异,由坡脚至坡顶,库水位波动的影响水平依次降低,降雨强度的影响水平逐渐增强。本文提出的数据挖掘方法可定量分析滑坡变形的控制因素,并通过与实测数据的对比验证了相关规则的可靠性,这对于海量监测数据条件下滑坡灾害的成因分析有重要意义。
Understanding the causes of landslides from real-time monitoring data is important for identifying triggering mechanisms and potential hazard areas and formulating mitigation measures in a timely manner.However,due to the large amount and diverse sources of monitoring data,the conventional data processing methods can hardly extract useful information from huge monitoring data so as to make a correct evaluation of landslide deformation behaviors and evolution trends.This paper introduces the two-step clustering and association rule analysis methods in the classical data mining methods,proposes the data mining process of association analysis of slope deformation behavior,takes the Xinpu landslide in the Three Gorges reservoir region of Yangtze River as an example,and carries out the association analysis of slope displacement velocity under the influence of reservoir water level and rainfall.The results show that the landslide deformation in the reservoir area is influenced by multiple factors such as the elevation of reservoir water level,reservoir water level fluctuation velocity and rainfall intensity.Water level decline and strong rainfall are closely related to landslide deformation.There are differences in deformation influencing factors at different spatial locations of the landslide.The influence level of reservoir water level fluctuation decreases and the influence level of rainfall intensity increases from front part to rear part.The data mining method can be used to analyze the influencing factors of landslide deformation.It can be used to analyze the influencing factors of landslide deformation,and the comparison with the measured data verifies the reliability of the rules.These results are important for the analysis of the causes of landslide disasters under the massive monitoring data.
作者
朱鸿鹄
王佳
李厚芝
叶霄
施斌
张勤
ZHU Honghu;WANG Jia;LI Houzhi;YE Xiao;SHI Bin;ZHANG Qin(School of Earth Sciences and Engineering,Nanjing University,Nanjing 210023,China;Nanjing University High-Tech Institute at Suzhou,Suzhou 215123,China;Institute of Exploration Technology,China Geological Survey,Chengdu 611734,China;College of Geological Engineering and Geomatics,Chang an University,Xi'an 710054,China)
出处
《工程地质学报》
CSCD
北大核心
2022年第5期1517-1527,共11页
Journal of Engineering Geology
基金
国家自然科学基金(资助号:42225702,42077235)
国家重点研发计划课题(资助号:2018YFC1505104).
关键词
关联规则
滑坡变形
数据挖掘
库区滑坡
Association rule
Slope deformation
Data mining
Reservoir landslide