基于不同评价因子组合的铁路沿线滑坡危险性评价

Risk Assessment of Landslide along Railway Based on Different Combination of Evaluation Factors

滑坡危险性评价是滑坡风险评估的重要组成部分,对滑坡的预测和防治意义重大.如何科学合理的选取评价因子是现有研究中的薄弱环节.以雅安至巴塘段铁路为研究区域,首先,预选出12个滑坡评价因子;然后,采用主成分分析法、粗糙集和灰色关联分析3种因子筛选法得到3种评价因子组合,将其分别输入到随机森林、支持向量机和逻辑回归模型进行滑坡危险性评价;最后,采用受试者工作特征曲线对比12种组合的模型预测精度.实验结果表明:采用灰色关联分析筛选出的评价因子组合更准确,将其输入到随机森林模型中,所得预测精度最佳,模型曲线下面积为0.880 6,将滑坡危险性评价结果与滑坡隐患点进行验证,检验精度为86.4%.采用不同因子筛选方法选取最佳的评价因子组合,使得评价结果更为准确,为雅安至巴塘段铁路沿线滑坡灾害的风险控制提供参考依据.

Landslide risk assessment is an important part of landslide risk assessment, which is of great significance to the prediction and prevention of landslides.How to select evaluation factors scientifically and reasonably is a weak link in the existing research.Taking the Ya’an Batang railway as the study area, 12 landslide evaluation factors are pre-selected, and then the combination of three evaluation factors is obtained by using the three-factor screening methods of principal component analysis, rough set, and grey correlation analysis.Then they are input into the random forest, support vector machine, and logistic regression model for landslide risk evaluation.Finally, the subject working characteristic curve is used to compare the prediction accuracy of 12 combinations of models.The experimental results show that the combination of evaluation factors screened by grey correlation analysis is more accurate.When it is input into the random forest model, the prediction accuracy is the best, and the area under the model curve is 0.880 6.The landslide risk evaluation results are verified by the landslide hidden danger points, and the inspection accuracy is 86.4%.Different factor screening methods are used to select the best combination of evaluation factors, which makes the evaluation results more accurate, and provides a reference basis for the risk control of landslide disaster along Ya’an Batang railway.