机器学习算法反求水文地质参数

传统方法求解优化问题时,一般都是依据最小二乘原理来确定目标函数。鉴于这种方法没有考虑到原始测量数据的误差对计算结果及精度带来的影响。为此,提出机器学习算法改进传统的目标函数,同时结合双评价粒子群算法来求解水文地质参数。结果表明,该算法具有良好的收敛性和稳定性,求解效率高,简单易实现。      

基于改进SPH模型的溃坝洪水演进模拟方法

溃坝洪水演进模拟的准确性是制约水库洪水预演有效性的关键。基于光滑粒子流体动力学(Smoothed Particle Hydrodynamics, SPH)方法提出了适用于溃坝洪水演进分析的数值模拟方法。通过设置溃口粒子与粒子库,基于黎曼不变量对SPH粒子状态进行修正,构建施加边界条件的改进SPH溃坝洪水演进模型,将SPH瞬时全溃整体模型转换为考虑溃口水流变化的入流边界模型,实现SPH方法与溃口计算模型的耦合。以Malpasset溃坝事件为例,检验了该模型计算溃坝洪水的精度,结果表明该模型精度相对较高,与实测值吻合较好;应用该模型模拟了某水库溃坝洪水演进预演过程,评估其对下游输水干渠及交叉建筑物排水倒虹吸的洪水冲击风险,结果表明在上游水库遭遇超标准洪水漫顶溃坝工况下,洪水演进至排水倒虹吸处的最大洪水位未超过校核洪水位。改进SPH模型精度高,可靠性强,与溃口计算模型耦合性好,可作为溃坝洪水演进模拟的通用手段之一。     

基于粒子群优化算法的未知波速声发射定位数值模拟

针对岩石力学试验中基于时差定位算法中声发射定位精度受岩石波速等诸多因素的影响问题,提出基于粒子群优化的未知波速声发射定位算法。该算法将岩石波速作为未知值,根据拾取到的时差建立基于最小二乘法原理的目标函数,利用粒子群优化算法,求解目标函数,寻得目标位置和波速。权重系数是影响该算法精度及其稳定性的重要因素,首先通过数值试验模型确定最优权重系数为0.729 8,可以满足算法精度及其稳定性。数值仿真结果表明,由选取的权重系数最后计算精度高于传统已知波速算法。为验证该算法的实际应用效果,进行断铅试验,结果表明该算法优于传统已知波速算法。     

改进型粒子群算法及其在GPR全波形反演中的应用

探地雷达作为高精度的物探工作方法,其主要目的是反演解释地下结构的物性参数。笔者提出社会学习型粒子群优化反演方法,它以信号均方误差为目标函数,用时域有限差分方法作正演,并且针对反射波信号较弱、反演效果不佳的情况设计了对正演结果进行振幅补偿的方法,对反射波的振幅进行增益,以提高反演精度。通过与经典粒子群优化反演方法的结果对比,说明了该算法在准确度以及效率方面都有相当大的提高。经过分析多层介质仿真数据的一维反演结果,说明了该算法对多参数反演的有效性和良好的抗噪性。     

基于免疫粒子群算法的马斯京根模型参数识别

目前在河道汇流计算中广泛应用的是马斯京根模型,该模型参数的优化求解是影响能否精确模拟实测水文过程的关键问题.以往的模型参数求解大多采用试错法、矩法等方法,计算过程繁琐,计算精度不高,影响到模型的模拟精度.针对此类问题,笔者将免疫原理与粒子群优化算法有机结合.提出了免疫粒子群优化新方法,有效解决了传统方法的计算结果精度不高的问题,并在马斯京根模型参数优化求解中得以应用,结果表明免疫粒子群算法的计算结果精度令人满意,为河道洪水演算方面研究提供了一种新的研究方法和研究模式.     

基于紧支径向基函数的点插值无网格方法

紧支径向基函数能使支配方程中的刚度矩阵具有稀疏性,很适合应用于无网格方法中,其缺点是在插值计算时精度不高.点插值方法的插值函数具有Delta函数性质,可以很方便的施加本质边界条件,但在计算插值函数时矩阵易出现奇异.为了提高计算精度并避免点插值法的局限性,首先对紧支径向基函数进行完备性修正,然后用完备性修正的紧支径向基函数代替多项式来形成插值函数,建立了紧支径向基函数点插值方法.由于该方法中的形函数满足Delta函数性质,因此本质边界条件可以像传统的有限元方法一样很容易施加.然后将该方法用于二维弹性静力问题的求解,导出了其相应的离散方程.最后将该方法应用于一个悬臂梁的分析中,初步验证了该方法的有效性与合理性.     

基于三维并行SPH模型的土体流滑特性研究

土体的大变形流滑导致了许多地质灾害的发生,对人们的生命财产安全构成了极大的威胁,因此越来越多的研究开始关注土体的大变形流滑特性。其中,光滑粒子流体动力学（SPH）方法是常用的模拟方法之一,但SPH方法的粒子特性导致其计算时间过长,影响了在工程地质领域的进一步应用。对此,本研究基于SPH方法的基本原理、非牛顿流体理论和等效黏度概念,提出了适用于土体大变形流滑分析的三维SPH仿真模型。结合OpenMP并行计算原理,实现了SPH算法的并行优化。在此基础上,对土体流滑模型试验进行了二维和三维分析,得到了滑动距离、滑动冲击力和冲击力峰值等动力学参数,分析了计算维数和边界条件对流滑特性的影响机制。通过不同线程数下计算时间的对比,获得了计算效率随线程数的变化规律。结果证明了本文的OpenMP并行优化具有较高的计算效率,显著降低了三维SPH模拟的计算耗时,对工程地质数值方法的效率提升具有重要的借鉴意义。     

大定圆回线频率域电磁场特征研究

大定圆回线非中心回线装置的电磁场计算公式中含有双重贝塞尔函数,因其高震荡性,使用常规方法计算误差较大。为提高计算精度,本文使用一种通过改进核函数的快速汉克尔变换算法来进行求解,计算结果表明:该算法能够很好的压制贝塞尔函数的强震荡性对计算结果的干扰,即使在非准静态近似的条件下,也能精确地计算出大圆回线源位于地表及接收点位于地面任意一点的电磁场,此外,大定源回线电磁场的分布规律不仅与接收装置位于回线内、外有关,还与圆回线所处的位置成一定的线性关系。     

煤矿井下无线传感器网络三维定位算法研究

针对煤矿无线传感器网络定位存在节点模型不合理、定位精度差、算法不稳定等问题,提出一种新的煤矿井下巷道节点模型,并在此基础上提出基于粒子群算法的煤矿井下无线传感器网络节点三维定位算法。该算法将三维空间中未知节点与相邻的信标节点之间的估算距离和测量距离的均方误差作为优化的目标函数,通过粒子群优化来提高定位精度。理论分析和仿真结果表明,该算法具有定位精度高以及稳定性较好等特点,适用于煤矿井下无线传感器网络节点定位。      

土石坝动力分析的改进二次加速度积分法

土石坝动力分析常用的是Wilson-? 积分方法,但该方法用于分析高频问题和大步长问题时精度较低。针对该缺点提出一种改进的二次加速度积分法,该算法与一步二次加速度积分法相比,计算精度不变、稳定性要求容易满足,计算用时大大减少。对算法的基本原理、稳定性及精度进行了分析,并给出一个土石坝动力分析的算例,验证了该算法的可行性。与常用的Wilson-? 法、Newmark法和中心差分法进行了数值比较。计算结果表明：该算法在分析高频问题和大步长问题时,计算精度较高。     

基于光滑质点流体动力学的海冰热力-动力数值模式

考虑海冰热力因素对其厚度、密集度的影响,在光滑质点流体动力学(SPH)基础上发展了一个海冰热力-动力数值模式。该模式既解决了传统欧拉有限差分法和质点网格法存在的数值扩散问题,同时弥补了光滑质点动力学海冰动力模式未考虑热力因素的不足,具有精确模拟冰缘线运动、计算精度高等优点。首先介绍了光滑质点流体动力学的基本原理,并对海冰生消的热力因素进行了分析,将影响冰厚和密集度的热力因素引入到光滑质点流体动力学的海冰动力模式中,得到该热力-动力模式的控制方程。应用该数值模式对渤海海冰进行了48 h数值模拟,得到了海冰厚度和速度矢量的分布规律;对JZ20-2海域的海冰厚度、冰内温度场分布以及热力因素的变化特性进行了讨论。数值模拟结果表明,该数值模式能够很好地适用于渤海海冰数值模拟,是一种有效的海冰数值模拟方法。     

Identification of failure slip surfaces for landslide risk assessment using smoothed particle hydrodynamics

ABSTRACT

Location of failure slip surfaces plays a critical role in landslide risk assessment and mitigation, particularly for unstable slopes, because it is a key input to design of stabilisation measures for unstable slopes and it determines the volume of the sliding soil mass (i.e. landslide consequence). The failure slip surfaces in the numerical analysis (e.g. finite element/different method, FEM/FDM) are often identified using shear strength reduction (SSR) method. A careful examination of FEM results showed that, although the SSR method performs well for stable slopes, it might provide misleading results for unstable slopes. To properly locate failure slip surfaces for unstable slopes, this paper presents a particle-based numerical method called smoothed particle hydrodynamics (SPH), which is mesh-free, immune to the mesh distortion problem in FEM/FDM, and able to directly simulate large deformation of soils that occurs during landslides. A series of slope stability analyses is performed using an in-house SPH programme. Failure slip surfaces are properly identified by SPH for both stable and unstable slopes. Furthermore, because SPH provides a spatial distribution of the post-landslide large displacement of soils, the failure slip surfaces can be identified conveniently using soil displacement. A displacement-based criterion is proposed to locate the failure slip surfaces.     

A comprehensive study on the parameters setting in smoothed particle hydrodynamics (SPH) method applied to hydrodynamics problems

Smoothed particle hydrodynamics (SPH) is a meshfree, Lagrangian particle method which has advantages in handling solids with extremely large deformation. Like any other numerical methods, cares must be taken to ensure its desirable accuracy and stability through considering several correction techniques in calculation. The selection of values for parameters in those correction approaches is a key step in SPH simulation, which is always difficult for new beginners to deal well with effectively. This paper examines the common inconsistency and instability problems in SPH method and studies its computational efficiency when applied to hydrodynamics problems with material strength like soil column collapse. We analyzed in detail how the correction techniques mitigate these inconsistency and instability problems. Also, the numerical testing results associate with different values for the parameters used in the correction techniques are provided for better understanding the influence of these parameters and for finding out the desirable values. It is found that (1) the SPH method is easily subjected to an inconsistency problem in the boundary area due to the boundary deficiency, and it can be treated well by adopting “virtual particles” contributing to the particle summations. (2) The numerical oscillation in SPH simulation can be mitigated effectively by artificial viscosity with the suggested parameter values. (3) The tension cracking treatment, artificial viscosity and artificial stress work well in removing the tensile instability problem in SPH method. In addition, the nearest neighboring particle searching (NNPS) algorithm, spacing ratio, smoothing length and time step influence the efficiency and accuracy of SPH method significantly. It is shown that SPH method with suggested parameters values can produce a very good result compared with the experimental result.     

Lagrangian meshfree particles method (SPH) for large deformation and failure flows of geomaterial using elastic–plastic soil constitutive model

Simulation of large deformation and post‐failure of geomaterial in the framework of smoothed particle hydrodynamics (SPH) are presented in this study. The Drucker–Prager model with associated and non‐associated plastic flow rules is implemented into the SPH code to describe elastic–plastic soil behavior. In contrast to previous work on SPH for solids, where the hydrostatic pressure is often estimated from density by an equation of state, this study proposes to calculate the hydrostatic pressure of soil directly from constitutive models. Results obtained in this paper show that the original SPH method, which has been successfully applied to a vast range of problems, is unable to directly solve elastic–plastic flows of soil because of the so‐called SPH tensile instability. This numerical instability may result in unrealistic fracture and particles clustering in SPH simulation. For non‐cohesive soil, the instability is not serious and can be completely removed by using a tension cracking treatment from soil constitutive model and thereby give realistic soil behavior. However, the serious tensile instability that is found in SPH application for cohesive soil requires a special treatment to overcome this problem. In this paper, an artificial stress method is applied to remove the SPH numerical instability in cohesive soil. A number of numerical tests are carried out to check the capability of SPH in the current application. Numerical results are then compared with experimental and finite element method solutions. The good agreement obtained from these comparisons suggests that SPH can be extended to general geotechnical problems. Copyright © 2008 John Wiley & Sons, Ltd.     

Flow analysis of liquefied soils based on smoothed particle hydrodynamics

To overcome the disadvantages of traditional flow analysis methods for liquefied soils that exhibit fluidization and large deformation characteristics, Smoothed particle hydrodynamics (SPH) is adopted in this study to analyze the flow processes of liquefied soils. Bingham model with the use of the Mohr–Coulomb yield criterion, the concepts of equivalent Newtonian viscosity, and the Verlet neighbor list method are introduced into the framework of SPH to build an algorithm for the analysis of flowing liquefied soils. This modeling involves a simulation of physical model test of flowing liquefied soils that can be compared with numerical results. In addition, a shaking table test is selected from the literature for SPH analysis to verify the validation of the SPH method and extend its applications. The SPH simulation can reproduce the flow processes of liquefied soils and constrain estimates of the horizontal displacement, vertical displacement, and velocity of soils after liquefaction. According to the dynamic behaviors of the materials involved, designs can be implemented to improve the seismic safety of structures.     

A new SPH-based approach to simulation of granular flows using viscous damping and stress regularisation

The smoothed particle hydrodynamics (SPH) method was recently extended to simulate granular materials by the authors and demonstrated to be a powerful continuum numerical method to deal with the post-flow behaviour of granular materials. However, most existing SPH simulations of granular flows suffer from significant stress oscillation during the post-failure process, despite the use of an artificial viscosity to damp out stress fluctuation. In this paper, a new SPH approach combining viscous damping with stress/strain regularisation is proposed for simulations of granular flows. It is shown that the proposed SPH algorithm can improve the overall accuracy of the SPH performance by accurately predicting the smooth stress distribution during the post-failure process. It can also effectively remove the stress oscillation issue in the standard SPH model without having to use the standard SPH artificial viscosity that requires unphysical parameters. The predictions by the proposed SPH approach show very good agreement with experimental and numerical results reported in the literature. This suggests that the proposed method could be considered as a promising continuum alternative for simulations of granular flows.     

An improved SPH method for saturated soils and its application to investigate the mechanisms of embankment failure: Case of hydrostatic pore‐water pressure

The method of smoothed particle hydrodynamics (SPH) has recently been applied to computational geomechanics and has been shown to be a powerful alternative to the standard numerical method, that is, the finite element method, for handling large deformation and post‐failure of geomaterials. However, very few studies apply the SPH method to model saturated or submerged soil problems. Our recent studies of this matter revealed that significant errors may be made if the gradient of the pore‐water pressure is handled using the standard SPH formulation. To overcome this problem and to enhance the SPH applications to computational geomechanics, this article proposes a general SPH formulation, which can be applied straightforwardly to dry and saturated soils. For simplicity, the current work assumes hydrostatic pore‐water pressure. It is shown that the proposed formulation can remove the numerical error mentioned earlier. Moreover, this formulation automatically satisfies the dynamic boundary conditions at a submerged ground surface, thereby saving computational cost. Discussions on the applications of the standard and new SPH formulations are also given through some numerical tests. Furthermore, techniques to obtain the correct SPH solution are also proposed and discussed throughout. As an application of the proposed method, the effect of the dilatancy angle on the failure mechanism of a two‐sided embankment subjected to a high groundwater table is presented and compared with that of other solutions. Finally, the proposed formulation can be considered a basic formulation for further developments of SPH for saturated soils. Copyright © 2011 John Wiley & Sons, Ltd.     

Run-out analysis of flow-like landslides triggered by the Ms 8.0 2008 Wenchuan earthquake using smoothed particle hydrodynamics

Flow-like landslides have caused significant damage and casualties worldwide. However, studying such phenomena with traditional simulation methods is made difficult by their complex fluidization characteristics. In this paper, we use smoothed-particle hydrodynamics (SPH) for the run-out analysis of flow-like landslides. Compared with conventional methods, the proposed SPH modeling technique is the combination of a Bingham flow model and Navier?CStokes equations in the framework of computational fluid dynamics. At first, two benchmark problems of dam break and granular flow are simulated and verified to evaluate the accuracy of the SPH model. Then, run-out analyses are performed for flow-like landslides triggered by the Ms 8.0 Wenchuan earthquake that occurred on 12 May 2008 in Sichuan Province, China. Run-out analyses of the Tangjiashan, Wangjiayan, and Donghekou landslides are conducted by the application of SPH models to real flow-like landslides. All simulations show good agreement with characteristics of flow-like landslides observed in the field. We have found that numerical modeling can capture the fundamental dynamic behavior of these flow-like landslides and produce preliminary results for hazard assessment and site selection for reconstruction in earthquake-prone areas.     

基于HBP本构模型的泥石流动力过程SPH数值模拟

本构模型是描述泥石流流变特性的关键,也是决定其动力过程数值模拟准确性的核心问题之一。泥石流流体属多相混合物,现有的研究已证实其存在剪切增稠或剪切变稀的现象,传统基于Bingham及Cross线性本构关系的数值模型难以准确描述泥石流流变特性。文中探讨了Bingham模型在低剪应变率下的数值发散问题,在光滑粒子流体动力学（SPH）方法框架上建立了整合Herschel-Bulkley-Papanastasiou(HBP）本构关系的稀性泥石流动力过程三维数值模型。相比传统基于浅水波假设的二维数值模型,所述方法从三维尺度建立SPH形式下的泥石流浆体纳维?斯托克斯方程并进行数值求解,可获取泥石流速度场时空分布及堆积形态,同时采用HBP本构关系描述泥石流流变特性,能在确保数值收敛的前提下反映泥石流流体在塑性屈服过渡段及大变形状态下应力?应变的非线性变化。为验证提出方法的合理性,结合小型模型槽实验观测进行了对比,结果表明数值模拟与实测结果基本吻合。     

Parallel simulations of pore-scale flow through porous media

Smoothed particle hydrodynamics (SPH) is a versatile technique which can be applied to single and multiphase flow through porous media. The versatility of SPH is offset by its computational expense which limits the practicability of SPH for large problems involving low Reynolds number flow. A parallel pore-scale numerical model based on SPH is described for modeling flow phenomena in porous media. Aspects of SPH which complicate parallelization are emphasized. The speed of the method is demonstrated to be proportional to the number of processors for test cases where load balance was achieved. The parallel algorithm permits the application of SPH to more complicated porous media problems than previously considered. For such problems, best performance is achieved when several soil grains are simulated by each processor. Finally, future applications of the method and possible extensions are discussed.