基于随机森林的砂土地震液化预测模型  

Seismic Liquefaction Potential Assessed by Random Forest Method

作　　者：黄浩[1] 薛新华[1] 樊旭[1] HUANG Hao;XUE Xin-hua;FAN Xu(College of Water Resource and Hydropower,Sichuan University,Chengdu 610065,China)

机构地区：[1]四川大学水利水电学院

出　　处：《中国农村水利水电》2019年第8期158-161,173,共5页China Rural Water and Hydropower

基　　金：国家自然科学基金项目(41472271)

摘　　要：为研究随机森林(Random Forest,简称RF)智能算法对于砂土地震液化预测判别模型的适用性,在综合考虑地震因素、土层埋藏情况、砂土特性的基础上,选取地震等级、埋深、上覆压力、探锥阻力、最大地面加速度、有效上覆压力等6个评价指标,构建了基于随机森林算法的砂土地震液化判别模型。利用规范法基于工程实例对随机森林砂土液化判别模型进行对比验证,验证结果表明随机森林砂土液化判别模型具有较高的预测精度,是一种可行的砂土液化判别方法。To study the applicability of the random forest intelligent algorithm to the prediction model of sand liquefaction prediction,based on a comprehensive consideration of seismic factors,six factors including earthquake magnitude,depth of penetration,overburden pressure,cone resistance,normalized peak horizontal acceleration at ground surface and effective overburden pressure are selected as the evaluating indices to build the seismic liquefaction discriminant model.Based on engineering examples,the results are compared from random forest sand liquefaction discrimination model and normative method.The comparison results show that the random forest sand liquefaction discrimination model has high prediction accuracy and is a feasible new sand liquefaction discrimination method.

关 键 词：砂土液化 随机森林 原位测试 液化势 

分 类 号：TV16[水利工程]