摘要
针对影响边坡稳定多个相关因素的复杂性和不确定性,以及数据样本的不平衡,现有的方法无法提供精确的边坡稳定性结果等原因,需要迫切开发复杂的数据处理算法。本文通过详细调查获取了1994—2011年221个圆弧形滑坡案例中边坡高度、倾角和边坡坡体容重、黏聚力、内摩擦角、孔隙水压系数等特征参数及其相对应的稳定性状态(稳定、失稳)组成了模型样本库。然后,提出了一种基于支持向量机(Support Vector Machine,SVM)的方法进行边坡稳定性的预测分析,并采用精准率、AUC值和F1-Score评估模型预测性能。通过具体工程实例表明:SVM模型分类精确率、AUC以及Ce分别为0.970、0.898和0.925,明显优于GB、RF、KNN等模型,其预测结果具有很高的可信度。
Aiming at the complexities and uncertainties of several relevant factors affecting slope stability and the imbalance of the related data samples as well as the cause of that the accurate slope stability result cannot be provided by the existing methods concerned,an algorithm for processing the complicated data is necessary to be urgently developed.Through the detailed investigation,the characteristic parameters of the height,dip angle,cohesion,internal friction angle,pore pressure coefficient,etc.of the slopes in the cases of 221 circular landslides occurred in the period of 1994—2011 as well as the corresponding stability statuses(stabilized or destabilized)are obtained and then a model sample database is made up herein.Afterwards,a SVM(Support Vector Machine)-based method is put forward to carry out the prediction analysis of the slope stability,while the prediction performance of the prediction model is evaluated by means of accuracy,AUC value and F1-Score.Through the specific engineering cases,it is indicated that the classification accuracy rate,AUC and F1-Score of the SVM model are 0.970,0.898 and 0.925 respectively,which are obviously better than those from the models of GB、RF、KNN,etc.,thus its predicting result has a very high reliability.
作者
张云雁
ZHANG Yunyan(Yuxi Water Conservancy and Electric Power Survey and Design Institute of Yunnan Province, Yuxi 653100,Yunnan, China)
出处
《水利水电技术》
北大核心
2020年第11期205-209,共5页
Water Resources and Hydropower Engineering
基金
国家重点研发计划项目(2018YFC0407102)。
关键词
边坡稳定性
支持向量机
预测分析
边坡设计
slope stability
support vector machine
prediction analysis
slope design
