基于PSO-BP算法的隧道近接施工围岩参数反演与数值模拟
发布时间：2019-05-24 13:41
【摘要】：21世纪,我国城市路面资源的利用趋近于饱和,各城市大量修建地铁以减缓路面交通的运输量。而由于地铁的埋深较浅,对土体的开挖扰动较敏感,在开挖过程中造成地表沉降,更严重可能会形成地面塌陷,周边原有建筑物也会受到地铁近接施工的影响从而发生沉降,在新建地铁隧道的同时又要保证原有建筑物的稳定性是实际施工时不可忽视的。而实际工程中,通常根据隧道围岩的力学参数确定地铁施工的开挖方式以及支护结构的材料参数,但是岩体是一种非均质、非线性、不连续的材料,如何能够利用简单、高效的方式得到准确的力学参数尤为重要,通过获得围岩的力学参数,能准确预测并采取相应措施控制施工中引起的沉降值,对实际施工具有指导意义。本文结合大连地铁凌水一号桥区间的实际工程,提出基于粒子群的BP神经网络优化算法,利用现场监测的数据反演得到围岩的力学参数,对实际的工程应用有一定意义。本文的主要内容如下:(1)利用有限元分析软件ABAQUS分别对采用全断面开挖法、正台阶法、CD法和CRD法的地铁隧道施工进行数值模拟,分析得到四种不同施工方式下地表沉降值、围岩应力场和位移场、锚杆以及衬砌的内力变化情况。通过对比分析,结果表明:采用CRD法施工,隧道的控制效果较好,而全断面开挖法对隧道稳定性的控制相对较差。(2)根据正交试验设计原理,利用极差法和方差法分析各试验因素对围岩位移沉降值的影响情况,结果表明弹性模量相对于其他因素为主要的影响因素。选取岩体物理力学参数的取值范围,构造位移反分析学习样本,分别采用BP神经网络算法和PSO—BP算法进行计算,对比分析试验值和学习值,采用自验证的方式,得到PSO—BP算法的误差较小,有较好的预测能力。(3)结合凌水一号桥区间实际的监测数据,采用正交试验设计构造了PSO—BP算法的学习样本,将反演围岩的力学参数带入ABAQUS中,考虑近接施工,采用正台阶法模拟地铁动态开挖,分析近接施工对围岩及桥桩稳定性的影响,并将地表沉降、拱顶沉降、边墙净空收敛值、各桩基的沉降值与模拟值对比,得到最大的误差为18.4%,说明数值模拟分析的可靠性。
[Abstract]:In the 21st century, the utilization of urban pavement resources in China is approaching saturation, and a large number of subways are built in various cities to slow down the traffic volume of pavement traffic. However, due to the shallow buried depth of the subway and the sensitivity to the excavation disturbance of the soil, the surface subsidence may be caused by the excavation process, and the ground collapse may be formed, and the original buildings around the subway will also be affected by the adjacent construction of the subway, resulting in settlement. At the same time, the stability of the original building should be guaranteed when the subway tunnel is built, which can not be ignored in the actual construction. In practical engineering, the excavation mode of subway construction and the material parameters of supporting structure are usually determined according to the mechanical parameters of tunnel surrounding rock, but the rock mass is a kind of heterogeneous, nonlinear and discontinuous material, so how to make use of it is simple. It is particularly important to obtain accurate mechanical parameters in an efficient way. By obtaining the mechanical parameters of surrounding rock, we can accurately predict and take corresponding measures to control the settlement value caused by construction, which is of guiding significance to the actual construction. In this paper, based on the actual project of Lingshui No. 1 Bridge in Dalian Metro, a BP neural network optimization algorithm based on particle swarm optimization is proposed. The mechanical parameters of surrounding rock are obtained by inversion of field monitoring data, which is of certain significance for practical engineering application. The main contents of this paper are as follows: (1) the finite element analysis software ABAQUS is used to simulate the subway tunnel construction using full section excavation method, positive step method, CD method and CRD method, respectively. The changes of surface settlement, surrounding rock stress field and displacement field, anchor rod and lining internal force under four different construction modes are analyzed. Through comparative analysis, the results show that the control effect of CRD method is better, while the control effect of full section excavation method on tunnel stability is relatively poor. (2) according to the principle of orthogonal test design, The influence of each test factor on the displacement and settlement value of surrounding rock is analyzed by means of extreme difference method and variance method. The results show that the elastic modulus is the main influencing factor compared with other factors. The range of physical and mechanical parameters of rock mass is selected, and the learning samples of displacement back analysis are constructed. BP neural network algorithm and PSO-BP algorithm are used to calculate the values, and the experimental values and learning values are compared and analyzed, and the self-verification method is adopted. The error of PSO-BP algorithm is small and the prediction ability is good. (3) combined with the actual monitoring data of Lingshui No. 1 bridge interval, the learning sample of PSO-BP algorithm is constructed by orthogonal experiment design. The mechanical parameters of surrounding rock are inversed into ABAQUS, and the dynamic excavation of subway is simulated by forward step method. The influence of adjacent construction on the stability of surrounding rock and bridge pile is analyzed, and the surface settlement, arch roof settlement and side wall clearance convergence value are analyzed. By comparing the settlement value of each pile foundation with the simulated value, the maximum error is 18.4%, which shows the reliability of numerical simulation analysis.
【学位授予单位】：沈阳工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U455;U451.2
[Abstract]:In the 21st century, the utilization of urban pavement resources in China is approaching saturation, and a large number of subways are built in various cities to slow down the traffic volume of pavement traffic. However, due to the shallow buried depth of the subway and the sensitivity to the excavation disturbance of the soil, the surface subsidence may be caused by the excavation process, and the ground collapse may be formed, and the original buildings around the subway will also be affected by the adjacent construction of the subway, resulting in settlement. At the same time, the stability of the original building should be guaranteed when the subway tunnel is built, which can not be ignored in the actual construction. In practical engineering, the excavation mode of subway construction and the material parameters of supporting structure are usually determined according to the mechanical parameters of tunnel surrounding rock, but the rock mass is a kind of heterogeneous, nonlinear and discontinuous material, so how to make use of it is simple. It is particularly important to obtain accurate mechanical parameters in an efficient way. By obtaining the mechanical parameters of surrounding rock, we can accurately predict and take corresponding measures to control the settlement value caused by construction, which is of guiding significance to the actual construction. In this paper, based on the actual project of Lingshui No. 1 Bridge in Dalian Metro, a BP neural network optimization algorithm based on particle swarm optimization is proposed. The mechanical parameters of surrounding rock are obtained by inversion of field monitoring data, which is of certain significance for practical engineering application. The main contents of this paper are as follows: (1) the finite element analysis software ABAQUS is used to simulate the subway tunnel construction using full section excavation method, positive step method, CD method and CRD method, respectively. The changes of surface settlement, surrounding rock stress field and displacement field, anchor rod and lining internal force under four different construction modes are analyzed. Through comparative analysis, the results show that the control effect of CRD method is better, while the control effect of full section excavation method on tunnel stability is relatively poor. (2) according to the principle of orthogonal test design, The influence of each test factor on the displacement and settlement value of surrounding rock is analyzed by means of extreme difference method and variance method. The results show that the elastic modulus is the main influencing factor compared with other factors. The range of physical and mechanical parameters of rock mass is selected, and the learning samples of displacement back analysis are constructed. BP neural network algorithm and PSO-BP algorithm are used to calculate the values, and the experimental values and learning values are compared and analyzed, and the self-verification method is adopted. The error of PSO-BP algorithm is small and the prediction ability is good. (3) combined with the actual monitoring data of Lingshui No. 1 bridge interval, the learning sample of PSO-BP algorithm is constructed by orthogonal experiment design. The mechanical parameters of surrounding rock are inversed into ABAQUS, and the dynamic excavation of subway is simulated by forward step method. The influence of adjacent construction on the stability of surrounding rock and bridge pile is analyzed, and the surface settlement, arch roof settlement and side wall clearance convergence value are analyzed. By comparing the settlement value of each pile foundation with the simulated value, the maximum error is 18.4%, which shows the reliability of numerical simulation analysis.
【学位授予单位】：沈阳工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U455;U451.2
【相似文献】
相关期刊论文 前10条
1 谢新民;既有桥高锥体近接工程支护型式研究[J];铁道学报;2003年04期
2 廖惠生,処文，
本文编号：2484907
本文链接：https://www.wllwen.com/kejilunwen/daoluqiaoliang/2484907.html
