摘要
基于随机森林分类算法,提出一种边坡稳定预测模型。应用收集的边坡稳定状态实例资料,选取边坡的岩石重度、黏聚力、内摩擦角、边坡角、边坡高度、孔隙水压力等参数,对边坡稳定性进行预测,并与SVM模型和RBF神经网络模型的预测精度相比较;研究了随机森林模型中决策树的数量对模型预测精度的影响,并最终确定在样本数目为100左右时,模型中树的棵数设为500较合适。模型预测结果表明:选取的边坡参数比较合理,随机森林预测模型较其他模型更为精准,能有效预测边坡的稳定状态。
A slope stability forecasting model based on Random Forest classification was proposed. With the collected data of slope stability status,six parameters,including gravity density of rocks,cohesive force,internal friction angle, slope angle, slope height and pore water pressure were selected as factors influencing the slope stability to predict the status of slope. The prediction accuracy of the model was verified by Holdout test and 9 fold cross validation test and it was compared with the SVM algorithm and RBF network algorithm. The number of the trees in the predicting model was also studied and 500 or so was appropriate while the number of the examples was about 100. The results of the forecasting model show that the Random Forest model is more accurate than other models, which is in accord with engineering practice.
出处
《人民黄河》
CAS
北大核心
2017年第5期115-118,共4页
Yellow River
基金
国家自然科学基金重点项目(41323001)
高等学校博士学科点专项科研基金项目(20120094110005)
江苏省杰出青年基金项目(BK20140039)
关键词
随机森林
分类算法
边坡稳定
预测模型
精度比较
Random Forest
classification algorithm
slope stability
forecasting model
precision comparison
