| Along with the large-scale development and construction of the underground engineering, the particularity, randomness and the complexity of the rock-soil mass geological condition is more and more prominent, which causes the design and construction of underground engineering to be in the experience and qualitative dependence stage since long-period. In the complex geological conditions, to ensure the stability of tunnel, designers tend to artificially increase the support parameters, resulting in the safety factor is too large and great economic waste. Therefore, it is difficult to achieve the optimal design aiming at safety, reliability and economy. At the same time, the initial stress state of the surrounding rock will change and the unloading deformation will occur during the tunnel excavation. The rock mass may collapse if peripheral convergence deformation of tunnel exceeds permission value. Therefore, it is necessary to specially research the optimization for support structure of tunnel and failure probability analysis for tunnel surrounding rock deformation.At first, based on the construction monitoring of Ya Lin Road Jingzhu Mountain Tunnel, this paper analyzed the monitoring results and evaluated the stability of surrounding rock of shallow burying and unsymmetrical loaded section K5+250 by the value of deformation, the time of stability and the rate of deformation. According to the basic theory of optimization back analysis, combined with information feedback of site monitoring, the paper formulated optimization procedure by APDL (ANSYS Parametric Design Language) for optimization back analysis of the grouting reinforcement surrounding rock parameters of shallow burying and unsymmetrical loaded section K5+250, and further researched the internal force and deformation characteristics of primary support structure of shallow buried and bias section by analyzing the original support structure of this section by ANSYS.Secondly, an optimization method of tunnel support structure based on support vector machine (SVM) response surface model was established. The basic idea is to use SVM trained by learning samples, which was constructed by orthogonal experimental design method, to construct response surface model to approximatively express the complex nonlinear relationship between tunnel stability and support structure parameters, and then use MATLAB optimization technology to search the best support structure parameters. In the paper, this method was used to optimize primary support structure parameters of shallow burying and unsymmetrical loaded section K5+250 of Jingzhu Mountain Tunnel. By analyzing the numerical simulation results of the original support structure design and the optimization support structure design, it is found that under the premise of meeting the tunnel stability, the optimization design reduced about 20% of the support structure cost than the original design.Finally, in this paper, the criterion of tunnel surrounding rock displacement and SVM algorithm for function fitting were combined to establish the method of failure probability analysis by SVM for tunnel surrounding rock deformation in complex formation. The operation steps of this method were summarized and the corresponding calculation procedure was formulated. Failure probability analysis for surrounding rock deformation of Long Yang Road Liziping Tunnel in Zhangjiajie was analyzed by this method. It is found that the calculation cost of this method was 0.4% of Monte-Carlo method, the absolute error of calculation result is about 0.233% and relative error 2.422%. For structural reliability analysis of complex nonlinear implicit limit state equation, this method had lower error and higher precision than the traditional quadratic response surface method. |
