Uncertainty Level Clustering Landslide Hazard Assessment Method And Application Research

China is one of the worst landslide disasters country in the world.As a major geological disasters,landslide was endangering villages,industrial mining areas and towns.with the characteristics of region,mass and multiple and the situation of the national economy overall level is not high,decided that our country is impossible to conduct comprehensive project management for all potentially dangerous landslide points in a short period of time with enough economic strength and technological strength.thus,the accuracy,reliability and validity of landslide hazard evaluation model have become the major difficulties of geological disaster forecasting and one of the issue that urgently need to address.Regional landslide hazard assessment system is an open nonlinear dissipative structure,therefore need to establish evaluation math model with nonlinear science.Data mining clustering analysis is a commonly used non-supervised classification that one of the multivariate statistical analysis,and classified the data with similar characteristics in the form of statistics,prediction and analysis algorithms are meaning clear.But because of the particularity of the landslide,there are usually many hidden and uncertain factors,such as rainfall and artificial damage,traditional clustering analysis methods cannot portrayed effectively,thus resulting in decreased accuracy of landslide hazard assessment,it also increases the difficulty of landslide hazard prediction and analysis.Therefore,the research of a new evaluation method of landslide disaster is no time to delay.This Paper's date is according to geological disaster detailed survey information of YanAn BaoTa district in North of ShanBei loess plateau,supported by the theory of loess geological disaster formed mechanism that effected in natural role,human activities and double impact of natural role and human activities,through the depth research of research district geological environment background and landslide profile,created regional landslide risk evaluation index system that effected by mutual role of evaluation factor and decision factor,introducing the efficient algorithm data of date mining clustering analysis,Summarizing the drawbacks of other researches on landslides and other geological disasters,and improve it,proposed a regional landslide hazard assessment method based on uncertainty hierarchical clustering split algorithm(U-DIANA).The algorithm is aim at the characters of uncertainty and strong participation of rainfall evaluation factor in the landslide geological disaster,through the uncertainty class clustering split algorithm that depend on data mining level clustering theory,addressed the description insufficient problem of uncertainty evaluation factor of rainfall;using ArcGIS for overlay analysis,and space statistics analysis,Visual analysis,series function mutual collaboration on the research district landslide risk evaluation factor and decision factor,reproduce geographic entity;Design landslide hazard level guidelines that combined with "search method" and "the expert assessment method",establish level evaluation model of landslide hazard,and verify the model in BaoTa district.Experimental results showed that the evaluation model has been made high precision on cluster validity analysis and accuracy is better than the traditional hierarchical clustering evaluation methods,achieve regional landslide assessment accuracy standard.Provide technical guarantee to disaster prevention and mitigation.