基于免疫多输出支持向量回归算法的隧道工程位移反分析新方法  

New Method for Back Analysis of Tunnel Displacement Based on Immune Multiple Output Support Vector Regression Algorithm

作　　者：宋威 刘开云[1] 梁军平[2] 徐冲[3] SONG Wei;LIU Kaiyun;LIANG Junping;XU Chong(School of Civil Engineering,Beijing Jiaotong University,Beijing 100044,China;School of Environment,Tsinghua University,Beijing 100084,China;China Railway First Survey and Design Institute Group Co.,Ltd.,Xi’an710043,China)

机构地区：[1]北京交通大学土木建筑工程学院,北京100044 [2]清华大学环境学院,北京100084 [3]中铁第一勘察设计院集团有限公司,陕西西安710043

出　　处：《铁道学报》2022年第2期126-134,共9页Journal of the China Railway Society

基　　金：国家自然科学基金(51378052)。

摘　　要：智能位移反分析作为计算模型参数辨识的一种方法在隧道工程中得到了广泛应用,鉴于现阶段智能位移反分析方法存在的主要问题,在一维支持向量回归算法的基础上,引入多输出的损失函数,建立了隧道工程围岩位移反分析的多输出支持向量回归(MSVR)模型,该模型将多个输出变量统一建模,解决了位移变量间的相关性问题,并引入免疫克隆选择算法(ISCA)与其耦合,形成免疫多输出支持向量回归(ICSA-MSVR)算法。在样本训练过程中采用免疫克隆选择算法自动搜索多输出支持向量回归模型的参数以建立岩体物理力学参数与位移之间的免疫多输出支持向量回归模型,然后采用免疫克隆选择算法自动搜索,求解出使MSVR计算位移与实测位移最接近的岩体物理力学参数组合,获得最优的计算模型参数。三口隧道施工期的应用结果表明,所提出的基于免疫多输出支持向量回归算法的位移反分析方法具有精度高和计算快捷的优点,可为类似工程提供借鉴。Intelligent back-analysis of displacement is widely used in tunnel engineering as a method of parameter identification of calculation model.In view of the main problems of intelligent back-analysis of displacement at the current stage,a multi-output loss function based on one-dimensional support vector regression algorithm was introduced and a multi-output support vector regression(MSVR)model for back-analysis of displacement of surrounding rock of tunnel engineering was established.In this model,multiple output variables were modeled in a unified way to solve the problem of correlation between displacement variables.An immune multi-output support vector regression(ICSA-MSVR)algorithm was formed by introducing immune clonal selection algorithm(ISCA)and its coupling.In the process of sample training,the immune clonal selection algorithm was used to automatically search the parameters of the multi-output support vector regression model to establish the immune multi-output support vector regression model between the physical and mechanical parameters of rock mass and displacement.Then the immune clonal selection algorithm was used to automatically search for the combination of the physical and mechanical parameters of rock mass that makes the calculated displacement of MSVR closest to the measured displacement,in order to obtain the optimal calculation model parameters.The application results during the construction period of the Sankou tunnel show that the back analysis method based on the immune multi-output support vector regression algorithm proposed in this paper has the advantages of high accuracy and fast calculation,which provides a new reference for similar projects.

关 键 词：隧道工程 位移反分析 免疫克隆选择算法 多输出支持向量回归 参数辨识 

分 类 号：U452[建筑科学—桥梁与隧道工程]