地质灾害风险评价阈回归联合聚类分析

Threshold Regressive Model and Cluster Analysis of Sediment-Related Disaster

根据吉林省近10年的地质灾害监测数据,采用统计方法讨论吉林省地质灾害风险评价问题,对地质灾害易发区进行划分。首先对系统聚类分析方法进行了改进,根据各种要素对地质灾害影响的程度引入了权重系数,分类后应用判别分析的方法检验出其精确度达到94.87%。为了进一步了解各种因素引致灾害的形式,掌握其在不同规模数值下形成灾害的特点及差异性,运用阈回归模型进行了数据拟合,因素的解释程度达到99.996 2%。加权聚类和阈回归方法对数据拟合程度的高精度,说明其在地质灾害风险评价问题中的适用性及灾害易发区划分的准确性。

By using 10-year monitoring data of geological disaster in Jilin Province,we focused on risk assessment and partition of susceptible area of geological disaster with statistics methods.Firstly,hierachical cluster analysis was improved by introducing into weight coefficient based on the importance of different factors,and the accuracy of partition was tested to achieve 94.87% by discriminant analysis.Secondly,a threshold regressive model whose accuracy got to 99.996 2% was used to learn the forms of disaster caused by all kinds of factors and master their features and differences of disaster formation under numerical values of different sizes of different factors.The high level of accuracy of the combination application of weighted hierachical cluster analysis with threshold regressive model showed it was proper methods to resolve the issue of risk assessment and our partition of susceptible area of geological disaster was precise.