贝叶斯信息标准在滑坡因子敏感性分析中的应用  

Application of Bayesian information criterion to sensitivity analysis of landslide factor

作　　者：李雪平[1] 唐辉明[1] 

机构地区：[1]中国地质大学工程学院,武汉430074

出　　处：《岩土力学》2006年第8期1393-1397,共5页Rock and Soil Mechanics

基　　金：国家自然科学基金项目(No.40072085)

摘　　要：滑坡因子敏感性分析是滑坡预测和治理的重要前提。以巫山县新址西区作为试验区,运用滑坡影响因素与历史滑坡之间建立的Logistic回归模型,通过贝叶斯信息标准进行模型优劣程度的比较,以期得出本区滑坡因子的敏感程度。设计了逐个加入影响因子进行非嵌套模型的优劣程度对比的试验方法。试验区滑坡因子敏感程度计算结果排队依次为:岩性、高程、距有影响构造线距离、坡度、坡向、坡形。试验为区域斜坡稳定性评价提供了一种新的、可靠的方法。It is important to analyze sensitivity of landslide factor. The west area in new Wushan County was taken as an experimental area. Logistic model was set up for which independent variables are landform, physiognomy, lithological character and geological structure conditions and dependent variable is historical landslide condition. The sensitivity degree is expected according to Bayesian information criterion. The authors design test method which taken in factor one by one and compared with superior degree on not nested model. The results of sensitivity degree are lithological character, height, distance to effected structure line, gradient, slope direction, slope shape. The experiment provides a new and credible method for evaluating slope stability.

关 键 词：贝叶斯信息标准 BIC 滑坡 因子 敏感性 

分 类 号：TU443[建筑科学—岩土工程]