隧洞围岩损失位移估计的智能优化反分析

隧洞开挖过程中围岩监测断面的布置一般滞后于掌子面开挖,监测断面布置前围岩已发生的位移称为损失位移。采用优化反分析思路求取损失位移,该思路将损失位移的求解转化为以实测位移与计算位移的误差作为目标函数、岩体力学参数作为决策变量的全局优化反分析问题。针对该全局优化反分析问题是一类高度非线性多峰值且计算代价较高的优化问题,将性能优异的粒子群优化算法与高斯过程机器学习方法相融合,结合FLAC3D数值计算程序,提出隧洞围岩损失位移优化反分析的粒子群-高斯过程-FLAC3D智能协同优化方法。算例研究表明,该方法是可行的,不仅能获得可靠的损失位移预测结果,而且可获取合理的围岩计算模型力学参数,具有全局性好、计算效率高的特点,克服了传统优化反分析方法容易陷入局部最优或过于依赖初始学习样本的局限性。将该方法应用到锦屏二级水电站辅助洞BK14+599断面的损失位移反分析,获得了该断面围岩的损失位移和力学参数,其中,损失位移较大,原因在于岩体开挖后在短时间内弹性变形大。因此,对于地下工程,特别是深部地下岩体工程,在围岩稳定性评价与围岩参数反分析中,损失位移不可忽视,应给予足够重视。     

围岩力学参数反演的GSA-BP神经网络模型及应用

针对遗传算法（GA）存在早熟现象和局部寻优能力较差等缺陷,引入具有很强局部搜索能力的模拟退火算法（SA）,组成改进的遗传模拟退火算法（GSA）提高优化问题的能力和求解质量。针对BP神经网络容易陷入局部最小和收敛速度慢等方面的不足,应用改进的遗传模拟退火算法搜索BP神经网络的最优权值和阀值,提高BP神经网络的预测精度,建立了围岩力学参数反分析的GSA-BP神经网络模型。将该模型应用于乌东德水电站右岸地下厂房围岩力学参数的反演分析中,根据监测围岩变形数据反演围岩力学参数,反演所得参数应用到正计算分析中,得出的计算位移与实测值吻合较好,说明该方法的有效性和应用于该工程的可行性。     

地铁深基坑支护的遗传神经网络位移反分析

针对目前已有的各种位移反分析方法存在的缺陷,利用神经网络具有的非线性映射能力和遗传算法具有的全局随机搜索能力,提出了一种基于遗传神经网络进行深基坑支护的位移反分析方法。该方法改变了BP算法依赖梯度信息的指导来调整网络权值的方法,而是利用遗传算法全局性搜索的特点,寻找最合适的网络连接权和网络结构等来达到优化的目的。结合地铁深基坑支护位移计算,应用该方法对某一地铁深基坑土体的力学参数进行了反演。结果表明：将位移观测值作为网络输入数据,土体力学参数作为输出数据,在较大的解空间内,该位移反分析方法收敛速度快、解的稳定性好、反演结果精度高,是一种理想的位移反分析方法。最后,采用该软件结合一个工程实例实现了应用遗传神经网络进行的基坑支护位移反分析。     

A hybrid Bayesian back‐propagation neural network approach to multivariate modelling

There is growing interest in the use of back‐propagation neural networks to model non‐linear multivariate problems in geotehnical engineering. To overcome the shortcomings of the conventional back‐propagation neural network, such as overfitting, where the neural network learns the spurious details and noise in the training examples, a hybrid back‐propagation algorithm has been developed. The method utilizes the genetic algorithms search technique and the Bayesian neural network methodology. The genetic algorithms enhance the stochastic search to locate the global minima for the neural network model. The Bayesian inference procedures essentially provide better generalization and a statistical approach to deal with data uncertainty in comparison with the conventional back‐propagation. The uncertainty of data can be indicated using error bars. Two examples are presented to demonstrate the convergence and generalization capabilities of this hybrid algorithm. Copyright © 2003 John Wiley & Sons, Ltd.     

基于神经网络和演化算法的土石坝位移反演分析

构造了基于综合应用人工神经网络和演化算法的位移反演分析方法。该法使用具有较强非线性映射能力的神经网络模型代替有限元计算,提高了计算效率。采用演化算法和Vogl快速算法,同时优化神经网络的结构和权值,增加其适应性并加快训练速度;使用多种群演化等策略,改善演化算法的全局收敛性和收敛速度。以三峡茅坪溪防护土石坝的变形反演分析为例,研究了神经网络演化代数以及训练样本数量对神经网络模拟能力的影响,证明了所建立的反演分析方法的有效性。     

天然岩体渗透系数反演研究

天然岩体存在很多裂隙,而水在这些裂隙中的渗流严重影响岩石工程的稳定性,因此确定天然岩体渗透系数具有重大的实际意义。反演方法是确定岩体渗透系数的一种较理想的方法,渗透系数反演可归结为一个复杂的非线性函数优化问题。由于采用传统优化技术存在不少问题,而目前采用的全局优化算法—遗传算法也存在本质的问题,因此,提出仿生算法-免疫进化规划进行岩体渗透系数反演,并用一个大坝坝基工程的算例证明了算法的有效性。结果表明,其方法可以在仅知道水头的条件下,得到接近实际的渗透系数值。     

基于遗传－广义回归神经元算法的坞石隧道三维弹塑性位移反分析研究

广义回归神经元网络在逼近能力、学习速度和网络稳定性方面均优于BP神经元网络,且具有网络人为调节参数少的优点。本文将广义回归神经元网络引入坞石隧道工程的三维弹塑性位移反分析。为了在网络训练过程中快速搜索到最优的网络阈值,采用十进制遗传算法对网络阈值进行优化。在确定最优的网络结构后,采用遗传算法在每个待反演参数的搜索范围内搜索出与实测位移最接近的围岩力学与初始应力场参数组合。用反分析得来的参数进行下步开挖位移预测,预测值与实测值吻合较好,表明所提出的这种反分析方法在工程上是可行的,可以推广使用。     

基于改进遗传算法的粘弹性岩体力学参数反演

把模式搜索嵌入目前广为应用的遗传算法中,使之和神经网络有机结合,提出了搜索—遗传—神经网络算法。该方法用经过最佳预测学习算法训练的神经网络来表达粘弹性岩体力学参数和位移之间的映射关系,除具有一般遗传算法的优点外,还提高了参数反演的精度,节省了参数反演的计算时间。结合某工程实例,验证了该方法在粘弹性岩体力学参数反演中的优越性。     

PSO-LSSVM模型在位移反分析中的应用

提出了一种基于均匀设计原理、最小二乘支持向量机(LSSVM)和粒子群优化算法(PSO)的快速位移反分析方法。该方法利用均匀设计和有限差分法获得学习样本,再用粒子群算法搜索最优的最小二乘支持向量机模型参数。并用最小二乘支持向量机回归模型建立反演参数与监测点位移值之间的非线性映射关系,最后用粒子群算法从全局空间上搜索与实测位移最吻合的反演参数。该反演模型利用了粒子群算法高效简单、均匀设计构造高质量小样本以及最小二乘支持向量机的小样本、泛化性能好的特点。将该模型应用于龙滩水电站左岸地下厂房区岩体地应力场的反演分析中,计算结果与实测的位移值和地应力值均吻合较好,说明了该模型在岩土工程快速反演分析中具有良好的应用价值。     

位移反分析的粒子群优化-高斯过程协同优化方法

针对采用随机全局优化技术进行岩土工程位移反分析存在数值计算量大、效率低的问题,将粒子群优化算法与高斯过程机器学习技术相结合,提出了位移反分析的粒子群优化-高斯过程协同优化方法。该方法利用全局寻优性能优异的粒子群优化算法进行寻优的基础上,采用高斯过程机器学习模型不断地总结历史经验,预测包含全局最优解的最有前景区域,通过提高粒子群搜索效率并降低适应度评价次数,进而有效地降低位移反分析过程中的数值计算工作量。多种测试函数的数学验证和工程算例的研究结果表明该方法是可行的,与传统方法相比较,可显著地降低位移反分析的计算耗时。     

A Hybrid Multi‐Objective Evolutionary Algorithm for Optimal Groundwater Management under Variable Density Conditions

In this paper, a new hybrid multi-objective evolutionary algorithm (MOEA), the niched Pareto tabu search combined with a genetic algorithm (NPTSGA), is proposed for the management of groundwater resources under variable density conditions. Relatively few MOEAs can possess global search ability contenting with intensified search in a local area. Moreover, the overall searching ability of tabu search (TS) based MOEAs is very sensitive to the neighborhood step size. The NPTSGA is developed on the thought of integrating the genetic algorithm (GA) with a TS based MOEA, the niched Pareto tabu search (NPTS), which helps to alleviate both of the above difficulties. Here, the global search ability of the NPTS is improved by the diversification of candidate solutions arising from the evolving genetic algorithm population. Furthermore, the proposed methodology coupled with a density-dependent groundwater flow and solute transport simulator, SEAWAT, is developed and its performance is evaluated through a synthetic seawater intrusion management problem. Optimization results indicate that the NPTSGA offers a tradeoff between the two conflicting objectives. A key conclusion of this study is that the NPTSGA keeps the balance between the intensification of nondomination and the diversification of near Pareto-optimal solutions along the tradeoff curves and is a stable and robust method for implementing the multi-objective design of variable-density groundwater resources.     

高速公路路基性态反分析及沉降预报

本文用三元件粘弹性流变模型模拟路基土体的流变特性, 根据施工期实测资料, 采用粘弹性有限元和非线性二乘法阻尼优化方法, 反演路基土体的地质力学参数, 然后计算路面铺筑后的工后沉降, 编制了土体粘弹性有限元正反分析程序, 并对沪宁高速公路某路段进行了反分析计算, 得到了满意的结果。     

A numerical method for the study of shear band propagation in soft rocks

This paper investigates the possibility of interpreting progressive shear failure in hard soils and soft rocks as the result of shear propagation of a pre‐existing natural defect. This is done through the application of the principles of fracture mechanics, a slip‐weakening model (SWM) being used to simulate the non‐linear zone at the tips of the discontinuity. A numerical implementation of the SWM in a computation method based on the boundary element technique of the displacement discontinuity method (DDM) is presented. The crack and the non‐linear zone at the advancing tip are represented through a set of elements, where the displacement discontinuity (DD) in the tangential direction is determined on the basis of a friction law. A residual friction angle is assumed on the crack elements. Shear resistance decreases on elements in the non‐linear zone from a peak value at the tip, which is characteristic of intact material, to the residual value. The simulation of a uniaxial compressive test in plane strain conditions is carried out to exemplify the numerical methodology. The results emphasize the role played by the critical DD on the mechanical behaviour of the specimen. A validation of the model is shown through the back analysis of some experimental observations. The results of this back analysis show that a non‐linear fracture mechanics approach seems very promising to simulate experimental results, in particular with regards to the shear band evolution pattern. Copyright © 2009 John Wiley & Sons, Ltd.     

基于Cauchy积分解法与遗传算法的随机位移反分析研究

针对位移反分析的正算计算量大和全局最优解遗失问题,提出正算过程采用平面弹性复变方法中的Cauchy积分解法,优化方法选用遗传算法的随机位移反分析方法,并研究了位移观测噪声对识别结果的影响规律。实例应用表明,所提出的方法从根本解决了采用优化技术进行位移反分析的求解效率问题,具有较好的容错性能和全局搜索性能,且能够反映地应力场和实际岩体的不确定性。     

隧道围岩稳定性正算反演分析研究——以厦门海底隧道穿越风化深槽施工安全监控为例介绍

为了降低海底隧道施工风险,确保隧道施工顺利穿越海底几处风化深槽和风化囊区域,解决难以获得隧道围岩力学参数的技术难题,采用位移反分析方法建立了动态反演预测模型;作为比较,还简单介绍了弹塑性反演的一种全局优化方法。根据隧道典型断面实际监控量测的围岩拱顶沉降量和周边收敛位移量,结合先行服务隧道揭露的水文地质情况,进行优化反演分析,得到该类围岩初期支护后的等效弹性模量和等效侧压力系数。在相应的同类地质条件下,对后续将开挖的左、右主洞围岩采用边界元法进行正演数值计算,使之能为主洞施工方案比选以及支护设计参数调整与修正提供定量依据,做到信息化动态设计与施工。工程实例分析表明,利用正算反演分析法得出的围岩等效力学参数是可靠的,可据此对类似地质条件下主隧道围岩进行正演计算分析,预测主洞围岩的变形破坏模式,判断其围岩稳定性。位移反分析法是隧道施工变形理论预测分析与工程实际相联系的有效平台,为工程设计施工技术决策提供了一种切实有效的途径。     

泥岩巷道围岩弹塑性参数反演分析与长期稳定性预测

以实测的围岩蠕变变形资料为基础,基于现场监测所获得的蠕变本构模型,对某一矿区泥岩体的力学参数进行了弹塑性反演分析,得到的巷道围岩体基本力学参数分别为弹性模量E=2.0 GPa,凝聚力c=1.31 MPa,内摩擦角? =24º。待反演参数水平均值的极差结果显示,通过极差的大小可以判断岩土力学参数的敏感性,对蠕变的影响而言,凝聚力最敏感,其次是内摩擦角,弹性模量再次之。在此基础上,利用反演所得的围岩基本力学参数进行了正演,计算得到了各测点处对应的蠕变位移增量,与实测值相比吻合较好,表明反分析所获得的岩体材料参数综合反映了巷道围岩的力学特性。最后,对巷道围岩变形大小及塑性区范围进行了预测。     

大型地下洞室考虑开挖卸荷效应的位移反分析

基于现场监测资料的位移反分析是地下工程动态监控、信息化施工的重要组成部分。以乌江彭水水电站大型地下厂房（开挖跨度为30 m,高度为78.5 m）为例,从围岩实测位移出发,建立了基于均匀设计－神经网络－遗传算法的围岩力学参数的系统反分析方法,反演考虑开挖卸荷效应的围岩力学参数。根据数值分析结果形成训练样本,利用BP人工神经网络映射围岩的变形与力学参数的关系,同时针对传统人工神经网络存在初始权值难以确定的问题,应用遗传算法优化神经网络的初始权值;利用现场监测的增量变形反演了围岩的力学参数;最后利用反演出的参数,进行地下厂房开挖预测分析。结果表明,预测位移与现场监测位移较为接近,进行统计检验结果为优,说明该参数反演方法是正确合理的。     

A Generalized Backward Euler algorithm for the numerical integration of an isotropic hardening elastoplastic model for mechanical and chemical degradation of bonded geomaterials

An extended version of the classical Generalized Backward Euler (GBE) algorithm is proposed for the numerical integration of a three‐invariant isotropic‐hardening elastoplastic model for cemented soils or weak rocks undergoing mechanical and non‐mechanical degradation processes. The restriction to isotropy allows to formulate the return mapping algorithm in the space of principal elastic strains. In this way, an efficient and robust integration scheme is developed which can be applied to relatively complex yield surface and plastic potential functions. Moreover, the proposed algorithm can be linearized in closed form, thus allowing for quadratic convergence in the global Newton iteration. A series of numerical experiments are performed to illustrate the accuracy and convergence properties of the algorithm. Selected results from a finite element analysis of a circular footing on a soft rock layer undergoing chemical weathering are then presented to illustrate the algorithm performance at the boundary value problem level. Copyright © 2002 John Wiley & Sons, Ltd.     

Reservoir geomechanical parameters identification based on ground surface movements

Determination of geomechanical parameters of petroleum reservoir and surrounding rock is important for coupled reservoir–geomechanical modeling, borehole stability analysis and hydraulic fracturing design. A displacement back analysis technique based on artificial neural network (ANN) and genetic algorithm (GA) combination is investigated in this paper to identify reservoir geomechanical parameters based on ground surface displacements. An ANN is used to map the nonlinear relationship between Young’s modulus, E, Poisson’s ratio, v, internal friction angle, Φ, cohesion, c and ground surface displacements. The necessary training and testing samples for ANN are created by using numerical analysis. GA is used to search the set of unknown reservoir geomechanical parameters. Results of the numerical experiment show that the displacement back analysis technique based on ANN–GA combination can effectively identify reservoir geomechanical parameters based on ground surface movements as a result of oil and gas production.     

Developing constitutive models from EPR‐based self‐learning finite element analysis

A constitutive model that captures the material behavior under a wide range of loading conditions is essential for simulating complex boundary value problems. In recent years, some attempts have been made to develop constitutive models for finite element analysis using self‐learning simulation (SelfSim). Self‐learning simulation is an inverse analysis technique that extracts material behavior from some boundary measurements (eg, load and displacement). In the heart of the self‐learning framework is a neural network which is used to train and develop a constitutive model that represents the material behavior. It is generally known that neural networks suffer from a number of drawbacks. This paper utilizes evolutionary polynomial regression (EPR) in the framework of SelfSim within an automation process which is coded in Matlab environment. EPR is a hybrid data mining technique that uses a combination of a genetic algorithm and the least square method to search for mathematical equations to represent the behavior of a system. Two strategies of material modeling have been considered in the SelfSim‐based finite element analysis. These include a total stress‐strain strategy applied to analysis of a truss structure using synthetic measurement data and an incremental stress‐strain strategy applied to simulation of triaxial tests using experimental data. The results show that effective and accurate constitutive models can be developed from the proposed EPR‐based self‐learning finite element method. The EPR‐based self‐learning FEM can provide accurate predictions to engineering problems. The main advantages of using EPR over neural network are highlighted.