首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于GWO-SVM岩爆分级预测模型
引用本文:吴菡,郭永刚,何军杰,苏立彬.基于GWO-SVM岩爆分级预测模型[J].路基工程,2023,0(1):49-54.
作者姓名:吴菡  郭永刚  何军杰  苏立彬
作者单位:西藏农牧学院水利土木工程学院,西藏林芝 860000
基金项目:西藏自治区科技重点研发计划项目资助(XZ202201ZY0034G)
摘    要:引入灰狼算法(GWO)优化支持向量机(SVM)模型,构建GWO-SVM岩爆分级预测模型。选取围岩最大切向应力、岩石单轴抗压强度、岩石单轴抗拉强度等,组成3种不同的输入指标组合,构建岩爆分级预测指标体系。将153组岩爆案例作为数据集输入4种模型进行训练、测试,比较不同输入组合下模型的预测效果。结果表明:GWO-SVM比标准SVM模型预测准确率提升7.41%~18.52%,在输入指标组合2下,GWO-SVM模型预测准确率最高达92.59%,基于GWOSVM的岩爆分级预测方法优于其他方法。

关 键 词:深部岩土工程  岩石力学  岩爆预测  灰狼算法  预测指标
收稿时间:2022-10-28

Rock Burst Classification Prediction Model Based on GWO-SVM
Abstract:Grey Wolf Optimizer (GWO) is introduced to optimize Support Vector Machine (SVM) model, and GWO-SVM rock burst classification prediction model is constructed. The maximum tangential stress of surrounding rock, rock uniaxial compressive strength, rock uniaxial tensile strength, etc. are selected to form three different input index combinations, and a reasonable rock burst classification prediction index system is constructed. 153 groups of rock burst actual cases are input into four models (GWO-SVM model, PNN model, Elman model, SVM model) as data sets for training and testing, and the prediction effects under different input combinations are compared. The results show that the prediction accuracy of GWO-SVM model is 7.41 %~18.52 % higher than that of the standard SVM model, and under the input index combination 2, the prediction accuracy of GWO-SVM model is up to 92.59 %, The prediction method of rock burst classification based on GWO-SVM is superior to other methods.
Keywords:
点击此处可从《路基工程》浏览原始摘要信息
点击此处可从《路基工程》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号