基于神经网络和模糊评判的滑坡敏感性分析

滑坡的敏感性涉及到很多因素,如滑坡体的坡度、坡的朝向、坡度的类型、岩石特性、海拔高度、植被覆盖等特征。神经网络具有非线性映射能力,利用这些与滑坡发生紧密相关的因素作为网络的输入,构造一个具有反映滑坡敏感性的评价网络,输出端为敏感性分析的结果。本文针对某具体地区,提取相关因素,构造评价指标体系并量化,利用该地区样本集数据对滑坡敏感性评价神经网络进行训练,用训练后的网络对实例并结合模糊评判进行了相互验证,结果说明利用神经网络和模糊评判进行滑坡敏感性分析是可行的。

Susceptibility of landslide based on Artificial Neural Networks and fuzzy evaluating model

Interrelated characters with landslide such as slope gradient,slope aspect,lithology of slope,slope type,slope altitude,and vegetable etc,can be gotten from Geographical Information System(GIS).Because Artificial Neural Networks(ANN) has high nonlinear projecting ability,the evaluation network with reflecting dangerous index of landslide is constructed and the output of evaluation network is the result of susceptibility of landslide.In this paper,aiming at a certain zone,related characters were distilled and evaluating index system was gotten and quantified.The evaluating network could be trained by sampling data from the researched zone on the basis of GIS.Some examples were judged by the trained network and also tested by fuzzy evaluating model.The result revealed that it would be feasible to evaluate susceptibility of landslide by ANN or fuzzy evaluating model.The frame with decision-making for landslide prediction was formed,in which observation from all kinds of means and prediction models were integrated.