马氏综合权重距离判别法在中小型滑坡灾害评估中的应用

我国西南地区滑坡灾害具有点多面广、规模小、危害大的特点，群测群防是目前中小型滑坡灾害最为有效的预防手段。针对这种现状，为进一步完善群测群防体系，在马氏距离判别法和加权马氏距离判别法的基础上对权重值进行了修正改进，提出了一种新的判别分析方法——马氏综合权重距离判别法。该方法不仅保留了加权马氏距离判别中区分每一个指标重要性差异的优势，并且结合德尔菲法对权重值进一步修正，弥补了原来加权马氏距离法在判别过程中距离函数出现负值的不足。从本质上来说，这种方法是把由样本驱动的权重值和实际工程经验的权重值相结合，使权重值的取值更加符合实际。以该方法为基础，在一定的历史资料背景下，从滑坡致灾因素和孕灾因素考虑，选取适当的判别因子，建立判别模型，对未知潜在滑坡灾害进行判别归类，并且与马氏距离判别法和Fisher判别法比较。研究发现，马氏综合权重距离判别法具有更高的可靠度和判别精度，适于在群测群防体系中进一步推广使用。

Application of Mahalanobis Comprehensive Weighting Distance Discriminant Method in Stability Assessment of Small and Medium Landslide

The landslide hazards in southwest China are widely distributed, smallscale and of great harm. Its prediction de pends on the monitoring and prevention system, which is the most effective method at present. The paper tries to improve the value of weight based on the Mahalanobis distance discriminant method and weighted Mahalanobis distance diseriminant method, and pro poses a new diseriminant methodMahalanobis comprehensive weighting distance discriminant method, to better complete the mon itoring and prevention system. The method not only keeps the advantage of separating each parametre properly according to their im portance, but also combines with the Delphi method to improve the value of weight. Essentially, this kind of method combines theweights driven by samples and practical engineering experience, so it will fit the reality better. Based on this method, the paper se lects proper discriminant factors and establishes discriminant model to classify unknown potential landslide disasters under a cer tain historical background. At the same time, it has been compared with the Mahalanobis distance discriminant method, Fisher dis criminant method. The results show that the Mahalanobis comprehensive weighting distance discriminant method has a higher dis criminant ability and it is suitable to further promote the use in the monitoring and prevention system.