Study On Displacement Back Analysis Method Of Geotechnical Engineering Using Support Vector Regression Optimized By Beetle Swarm Optimization Algorithm

The complexity of rock mass structure determines that geotechnical engineering is a highly non-linear uncertain system.In the process of design,construction and maintenance of geotechnical engineering,how to reasonably and accurately determine the physical and mechanical parameters of rock mass,so as to analyze the stress state and deformation of geotechnical engineering is a difficult problem.A displacement back analysis method of geotechnical engineering based on support vector regression optimized by beetle swarm optimization Algorithm（BSO-SVR）is presented in this paper.It uses the optimized support vector regression model instead of the numerical model to establish the complex non-linear mapping relationship between rock mass physical and mechanical parameters and its displacement,Inversion of physical and mechanical parameters of rock mass based on measured displacement information,Then the forward analysis of geotechnical engineering is carried out by finite difference model（FDM）.This method combines the advantages of advanced methods such as intelligent optimization method and numerical calculation method,and opens up a new way for back analysis of displacement in geotechnical engineering.The main work of this paper is as follows:（1）The principle of Support Vector Regression（SVR）is described in detail,and its modeling and implementation process are deduced in detail.The advantages,disadvantages and shortcomings of SVR are analyzed.The influence of parameters of SVR,penalty value-C and kernel width-?²,on the mapping ability of non-linear problems is pointed out.（2）The bionic principle of beetle antennae search（BAS）algorithm is described in detail.The process of modeling and programming is deduced in detail.The advantages,disadvantages and shortcomings of BAS are analyzed.Drawing on the idea of swarm intelligence algorithm,the beetle swarm with interactive information is generated,and the concept of beetle swarm speed is introduced to optimize BAS to form beetle swarm（BSO）algorithm.The modeling and programming process of BSO are described in detail.Based on the simulation test,the results of BAS and BSO are compared.The results reflect the accuracy and efficiency of BSO for the optimization of high-dimensional nonlinear complex functions,and prove the correctness of the optimization idea.（3）A displacement back analysis method of geotechnical engineering based on BSO-SVR is proposed.The modeling and programming process of BSO-SVR displacement back analysis（DBA）model are described in detail.The displacement back analysis method of geotechnical engineering based on BSO-SVR can be divided into three stages:the first stage is SVR model optimization stage,in which BSO searches SVR’s best parameter;the second stage is rock physical and mechanical parameters inversion stage,in which BSO searches best rock physical and mechanical parameters;and the third stage is FDM forward calculation and analysis,In this stage,the stress state and deformation of geotechnical engineering are analyzed and predicted.（4）Based on the above work,combined with in-situ scale model test of tunnel anchor of Jiangjin Youxi Yangtze River Bridge,BSO-SVR displacement back analysis method of geotechnical engineering is used to carry out engineering application research,and compared with traditional PSO-SVR method concurrently.Firstly,the values of elastic modulus-E,cohesion-C,internal friction angle-φand tensile strength-Rt of sandstone are inverted;then,the inverted parameters are substituted for the FDM software FLAC^3D to calculate the displacement of each measuring point,and compared with the measured displacement values;finally,the stress state and failure characteristics of the anchorage and its surrounding rock under different load ratios are analyzed.