基于数据挖掘技术的白水河滑坡多场信息关联准则分析

Analysis on association rules of multi-field information of Baishuihe landslide based on the data mining

为探究滑坡多场监测数据间的关联准则,采用数据挖掘技术中的两步聚类法与Apriori算法,开展滑坡多场信息关联准则研究。以三峡库区白水河滑坡为例,分析ZG93监测点于2003年6月—2016年12月期间的监测数据,选取影响滑坡变形的主要诱发因子,采用两步聚类法对不同的影响因子进行预聚类和聚类,将数值型变量转化为离散型变量后,应用Apriori算法进行处理,生成满足最小置信度的关联准则,建立白水河滑坡多场耦合作用模式下的影响因子与滑坡位移变形关联准则判据。研究表明,关联准则对于滑坡灾害的变形分析具有重要的意义,数据挖掘技术可较好地应用于三峡库区地质灾害位移预测预报中。

In order to explore the association criteria of landslide multi-field monitoring data,we have adopted the two-step clustering method and Apriori algorithm,which belong to the classical data mining method,and we have also proposed the process of landslide monitoring data mining.Based on the Baishuihe landslide in the Three Gorges Reservoir Area,we analyzed the monitoring data of ZG93 from June 2003 to June 2016.The main inducing factors of the landslide displacement were selected,and the two-step clustering method was used to pre-cluster and cluster the different influence factors.We used Apriori algorithm to deal with the classified variables to generate frequent item sets that satisfy the minimum support degree.The association rules between the precipitating factors and the landslide deformation are established under the multi-field coupling mode of Baishuihe landslide.The results show that the correlation criterion is of great significance to the deformation analysis of landslide hazards and the data mining technology can be applied to the displacement prediction of geological hazards in the Three Gorges Reservoir Area.