首页|期刊导航|路基工程|西南山区软岩大跨度隧道变形规律预测方法

西南山区软岩大跨度隧道变形规律预测方法OA

Prediction Method for Deformation Law of Large-span Soft Rock Tunnels in Southwest Mountainous Areas

中文摘要英文摘要

针对西南山区软岩深埋大跨度隧道变形监测数据非等距采样、强非线性及传统预测模型易局部最优等问题,提出基于时间序列分析与改进灰狼算法优化支持向量回归机的变形预测方法.利用 3次样条函数插值法处理非等距时序数据,将变形分解为趋势项与平稳项分量,采用灰狼算法(GWO)、正余弦优化的灰狼模型(HGWOSCA)优化支持向量回归机(SVR)参数.以某隧道拱顶沉降实测数据为例,采用 3种模型对趋势项变量与平稳项变量、总变形量进行预测对比.结果表明:HGWOSCA-SVR模型在各项预测中均表现最优,能较好反映大跨软岩隧道变形规律;第三断面总变形量预测的相关系数 R达 0.999 2,均方根误差和平均绝对百分比误差显著低于对比模型.HGWOSCA-SVR预测精度更高、误差更小,在实际大跨软岩隧道变形量预测中有效、可行.

To address the problems of non-equidistant sampling,strong nonlinearity of deformation monitoring data,and local optimum defects of traditional prediction models for deep-buried and large-span soft rock tunnels in the mountainous areas of Southwest China,a deformation prediction method based on time series analysis and support vector regression(SVR)optimized by an improved grey wolf optimizer is proposed.The cubic spline interpolation method is adopted to process non-equidistant time series data,and the tunnel deformation is decomposed into trend components and stationary components.The parameters of the SVR model are optimized by the standard grey wolf optimizer(GWO)and the sine-cosine optimized grey wolf model(HGWOSCA).Taking the measured crown settlement data of the Tunnel as the research case,three models are used to predict tand compare the trend variables,stationary variables,and total deformation.The results indicate that the HGWOSCA-SVR model exhibits the optimal performance in all prediction tasks and can well reflect the deformation law of large-span soft rock tunnels.The correlation coefficient R of the total deformation prediction for the third section reaches 0.999 2,and the root mean square error and mean absolute percentage error are significantly lower than those of the comparison models.The HGWOSCA-SVR model features higher prediction accuracy and lower errors,which is effective and feasible for the deformation prediction of actual large-span soft rock tunnels.

张亚辉;郭唯伟;王云成

石家庄铁道大学土木工程学院,石家庄 050043||河北城乡建设学校,石家庄 050031中国电建集团河北省电力勘测设计研究院有限公司,石家庄,050000中国铁路设计集团有限公司测绘地理信息研究院,天津 300308

交通工程

高铁隧道时间序列非等距现场监测数据对比变形预测实用性

high-speed railway tunneltime seriesnon-equidistantfield monitoringdata comparisondeformation predictionpracticality

《路基工程》 2026 (5)

34-40,7

中国国家铁路集团有限公司科技研究开发计划课题(N2020G009)

10.13379/j.issn.1003-8825.202412070

评论