基于Logistic回归算法的滑坡预报模型

Landslide Prediction Model Based on Logistic Regression Algorithm

为提高滑坡灾害的预报准确率、改善传统变量选取方法在成灾因子选取上的不足等问题,提出一种基于Logistic回归算法的滑坡预报模型。模型采用陕西省某地地质灾害监测点的监测数据作为模型数据集,用核主成分分析法对影响滑坡灾害的主要因子进行筛选,通过降维得到6个主要影响因子,并将筛选出的有效数据作为训练样本,训练得到滑坡预报模型。实验结果表明,该模型对滑坡发生概率的预报结果较为准确,相关指标符合要求。模型已成功应用于陕西省山阳县滑坡监测工程相关数据工作中。

In order to improve the accuracy of landslide disaster prediction and improve the shortcomings of traditional variable selection methods in disaster factor selection,a landslide prediction model based on Logistic Regression algorithm is proposed.In the model,the monitoring data of a geological disaster monitoring point in Shaanxi Province is used as the model data set,and the main factors affecting landslide disaster are screened by kernel principal component analysis.Six main influencing factors are obtained by dimension reduction,and the selected effective data are used as training samples to train the landslide prediction model.The experimental results show that the prediction results of landslide occurrence probability by the model are more accurate,and the related indexes meet the requirements.The model has been successfully applied to the relevant data of landslide monitoring project in Shanyang County,Shaanxi Province.