逼近散乱点云数据的三角形网格精确剖分

基于自组织特征映射神经网络构建的三角形网格模型可以实现测量点云 压缩后的Delaunay 三角逼近剖分,但该模型存在逼近误差和边缘误差。为减小三角形网格 的逼近误差和边缘误差,构建了精确逼近的三角形网格模型。首先采用整个测量点云,对三 角形网格模型中的所有神经元进行整体训练;然后对三角形网格中的网格神经元的位置权 重,沿网格顶点法矢方向进行修正;最后采用测量点云中的边界点集,对三角形网格模型中 的网格边界神经元进行训练。算例表明,应用该模型,可以有效减小三角形网格的边缘误差, 三角形网格逼近散乱点云的逼近精度得到大幅提高并覆盖散乱点云整体分布范围。     

从表面重构的体数据实现三维网格剖分

针对有限元分析中网格最优化问题,本文提出一种改进的生成四面体网格的自组织算法。该算法首先应用几何方法将三角形表面模型重新构造成规定大小的分类体数据,同时由该表面模型建立平衡八叉树,计算用以控制网格尺寸的三维数组;然后将体数据转换成邻域内不同等值面的形态一致的边界指示数组;结合改进的自组织算法和相关三维数据的插值函数,达到生成四面体网格的目的。实验对比表明,该方法能够生成更高比例的优质四面体,同时很好地保证了边界的一致。在对封闭的三维表面网格进行有限元建模时,本文算法为其提供了一种有效、可靠的途径。     

边界简化与多目标优化相结合的高质量四边形网格生成

目的 高质量四边形网格生成是计算机辅助设计、等几何分析与图形学领域中一个富有挑战性的重要问题。针对这一问题,提出一种基于边界简化与多目标优化的高质量四边形网格生成新框架。方法 首先针对亏格非零的平面区域,提出一种将多连通区域转化为单连通区域的方法,可生成高质量的插入边界;其次,提出"可简化角度"和"可简化面积比率"两个阈值概念,从顶点夹角和顶点三角形面积入手,将给定的多边形边界简化为粗糙多边形;然后对边界简化得到的粗糙多边形进行子域分解,并确定每个子域内的网格顶点连接信息;最后提出四边形网格的均匀性和正交性度量目标函数,并通过多目标非线性优化技术确定网格内部顶点的几何位置。结果 在同样的离散边界下,本文方法与现有方法所生成的四边网格相比,所生成的四边网格顶点和单元总数目较少,网格单元质量基本类似,计算时间成本大致相同,但奇异点数目可减少70% 80%,衡量网格单元质量的比例雅克比值等相关指标均有所提高。结论 本文所提出的四边形网格生成方法能够有效减少网格中的奇异点数目,并可生成具有良好光滑性、均匀性和正交性的高质量四边形网格,非常适用于工程分析和动画仿真。     

多体运动大流场计算的动态差分网格

提出无限大流场中多体运动计算的动态差分网格．无限大流场用充分大的人工边界模拟，而运动物体表面是流场的内边界．对于流场的差分网格，外边界（人工边界）点是固定的，内边界点则随时间而移动，因此流场的差分网格是动态的．几个有代表性的算例验证了动态差分网格的有效性．     

闭合三角网格的优化切割与保角映射

提出一种自动地将任意闭合三角网格切开并保角映射到二维平面域的算法.通过对自动提取的模型初始切割线逐步优化得到模型切割线,优化过程由一个与保角映射扭曲度和合法性相关的成本函数控制.为了减小映射扭曲,算法中不预先固定参数域边界,而在参数化过程中自动地确定网格的自然边界.实验结果表明,该算法通过优化切割线和参数域边界有效地降低了三角形形状扭曲,并保证了参数化结果的合法性.     

有限元网格的孔洞修补算法研究

针对板料成形零件的有限元网格模型提出了一种基于曲面的网格孔洞修补算法,该算法首先建立有限元网格模型的孔洞边界信息,其次利用网格孔洞边界和单元信息确定截面线的方向并生成截面线,然后用蒙皮法构造光滑的蒙皮曲面,最后利用基于边界约束的铺砌算法生成混合网格的孔洞网格,根据此算法获得的孔洞网格可以与原有网格光滑地融为一体,可以很好地满足板料成形CAE分析零件的网格孔洞修补要求,应用实践表明该算法是稳定可靠的。     

三维网格模型的边界性度量方法

在三维网格分割中,如何实现网格模型边界的自动准确分割是目前亟待解决的问题。为给自动分割提供理论依据,提出了一种新的三维网格模型表面边界性计算方法,将少量手工标注的边界点视为能量的放射源,根据能量流动原理,自动计算出其他点作为分割边界的可能性。实验表明,该方法是行之有效的,可以依据手工标注的少量边界点找到更多的真实边界点,进而为最终实现网格模型的自动分割提供可行的理论基础。     

Constrained mesh optimization on boundary

This paper presents a new mesh optimization approach aiming to improve the mesh quality on the boundary. The existing mesh untangling and smoothing algorithms (Vachal et al. in J Comput Phys 196: 627–644, 2004; Knupp in J Numer Methods Eng 48: 1165–1185, 2002), which have been proved to work well to interior mesh optimization, are enhanced by adding constrains of surface and curve shape functions that approximate the boundary geometry from the finite element mesh. The enhanced constrained optimization guarantees that the boundary nodes to be optimized always move on the approximated boundary. A dual-grid hexahedral meshing method is used to generate sample meshes for testing the proposed mesh optimization approach. As complementary treatments to the mesh optimization, appropriate mesh topology modifications, including buffering element insertion and local mesh refinement, are performed in order to eliminate concave and distorted elements on the boundary. Finally, the optimization results of some examples are given to demonstrate the effectivity of the proposed approach.     

逼近点云的三角形网格的边缘误差减小研究

基于SOFM神经网络构建的三角形网格模型可以实现测量点云压缩后的Delaunay三角逼近剖分,但该模型存在边缘误差。为减小三角形网格的边缘误差,改进了三角形网格模型的训练模式,提出了3步训练模式。第1步采用整个测量点云,对三角形网格模型中的所有神经元进行整体训练;第2步采用测量点云中的边界点集,对三角形网格模型中的网格边界神经元进行训练;第3步采用边界点集中的角点点集,对与边界角点匹配最佳的网格边界神经元进行训练。算例表明,应用该训练模式,可以有效减小三角形网格的边缘误差,三角形网格逼近散乱点云的逼近精度得到提高并覆盖散乱点云整体分布范围。     

The KW-Boundary Hybrid Digital Waveguide Mesh for Room Acoustics Applications

The digital waveguide mesh is a discrete-time simulation used to model acoustic wave propagation through a bounded medium. It can be applied to the simulation of the acoustics of rooms through the generation of impulse responses suitable for auralization purposes. However, large-scale three-dimensional mesh structures are required for high quality results. These structures must therefore be efficient and also capable of flexible boundary implementation in terms of both geometrical layout and the possibility for improved mesh termination algorithms. The general one-dimensional N-port boundary termination is investigated, where N depends on the geometry of the modeled domain and the mesh topology used. The equivalence between physical variable Kirchoff-model, and scattering-based wave-model boundary formulations is proved. This leads to the KW-hybrid one-dimensional N-port boundary-node termination, which is shown to be equivalent to the Kirchoff- and wave-model cases. The KW-hybrid boundary-node is implemented as part of a new hybrid two-dimensional triangular digital waveguide mesh. This is shown to offer the possibility for large-scale, computationally efficient mesh structures for more complex shapes. It proves more accurate than a similar rectilinear mesh in terms of geometrical fit, and offers significant savings in processing time and memory use over a standard wave-based model. The new hybrid mesh also has the potential for improved real-world room boundary simulations through the inclusion of additional mixed modeling algorithms     

三角网格曲面的球面参数化

提出了一种球面参数化三角网格曲面的方法。结合平面凸参数化和球面参数化,计算出封闭网格的切割线边界,网格边界映射到球面的凸区域边界上。然后分别参数化各子网格,最后将三角网格内部点映射到球面上。并用实例验证了此方法的可行性和有效性。     

解析二维非结构网格生成方法及其应用

提出弱区域指示函数描述流场形状并证明了其存在和不唯一性;利用该函数基于Delaunay方法依据流场尺度生成二维非结构网格。其中,边界引力有效解决了流场边界恢复和内嵌边界问题,弹簧振子和正三角形趋进技术有效改善网格质量。最后,弹簧振子随着弱区域指示函数的演变与监测函数的指导成功应用到自适应网格生成。     

Interactive mesh cloning driven by boundary loop

In order to copy arbitrary irregular mesh between two models continuously, this paper presents an interactive mesh cloning approach based on pyramid spherical coordinates driven by boundary loop. The approach extends an existing algorithm for computing offset membrane on mesh. A parametric paint brush is constructed to define canvas both on the source mesh and the target mesh. They are mapped onto a 2D parametric domain using discrete geodesic polar maps to register correspondingly. During cloning, the boundary loop of the region of interest (ROI) on the target mesh is fitted in real time by B-spline curve to register the boundary loop of the source ROI. Via the reconstructed boundary loop, the ROI is deformed to register the target mesh by pyramid spherical coordinates to ensure that the clone result is seamless and natural. Our approach can clone arbitrary irregular meshes between two 3D models, even if the mesh is non-manifold. The cloning process is operated in real time by GPU acceleration. Experimental results demonstrate the effectiveness of our interactive mesh cloning.     

基于非结构网格的Gas-Kinetic方法

为解决带有复杂几何边界条件的高速流体计算问题,提出基于非结构网格的Gas-Kinetic方法.对于二维非结构网格,以三角形网格作为计算单元,形成在该网格控制单元中物理量导数求解的新方法.通过物理量导数得到在控制体积元边界上的通量,然后用每个计算时间步中求出的边界通量和控制体积元中的物理量,求出下一计算时间步所需的新物理量,依次进行计算直到计算结果收敛为止.采用NACA0012翼型进行数值计算验证,结果表明该方法简单高效,适用于低速和高速流体的计算.     

Simultaneous Refinement and Coarsening for Adaptive Meshing

In the numerical simulation of the combustion process and microstructural evolution, we need to consider the adaptive meshing problem for a domain that has a moving boundary. During the simulation, the region ahead of the moving boundary needs to be refined (to satisfy stronger numerical conditions), and the submesh in the region behind the moving boundary should be coarsened (to reduce the mesh size). We present a unified scheme for simultaneously refining and coarsening a mesh. Our method uses sphere packings and guarantees that the resulting mesh is well-shaped and is within a constant factor of the optimal possible in the number of mesh elements. We also present several practical variations of our provably good algorithm.     

Boundary formulations for high-order finite differences on staggered meshes

Discretizations of the compressible flow equations with flow variables defined at different staggered positions on a regular mesh have accuracy and stability advantages over standard collocated discretization, but implementation of boundary conditions is hampered without all flow variables available at any boundary point. Boundary schemes for implementation of the boundary conditions compatible with staggered mesh discretizations are considered in this study. We focus on a combination of fourth- and fifth-order schemes near the boundary that are stable when sixth-order centered schemes are used for the interior points. Characteristics-based formulations provide a physically meaningful treatment for all the variables, avoiding the use of extrapolation. However its application on staggered meshes has not been systematically studied and its implementation is unclear since this method requires collocation of all the variables at the boundary, which is not natural for standard staggered mesh formulations. We show that including all the flow variables at the boundary can be done in a way that does not affect resolution or accuracy of the formulation. Predictions based upon analysis with model equations are verified with a staggered mesh flow solver.     

Quadtree grid generation: Information handling, boundary fitting and CFD applications

When solving partial differential equations by numerical methods, an automatic mesh generation technique which can accommodate local mesh refinement adaptively is desirable. One efficient technique for producing such meshes in two-dimensional space is to subdivide recursively the domain into quadrants using a quadtree to store and manipulate the mesh information. Here, the quadtree grid generation technique is reviewed and its programming discussed. Three data storage methods are examined. The conversion of the quadtree grid to a triangular finite element mesh is also described, along with methods for fitting the mesh to smooth boundary contours. Results from viscous flow and standing wave simulations are used to illustrate mesh adaptivity about internal and boundary features.     

A highly solid model boundary preserving method for large-scale parallel 3D Delaunay meshing on parallel computers

In this paper, we propose a novel parallel 3D Delaunay triangulation algorithm for large-scale simulations on parallel computers. Our method keeps the 3D boundary representation model information during the whole parallel 3D Delaunay triangulation process running on parallel computers so that the solid model information can be accessed dynamically and the meshing results can be very approaching to the model boundary with the increase of meshing scale. The model is coarsely meshed at first and distributed on CPUs with consistent partitioned shared interfaces and partitioned model boundary meshes across processors. The domain partition aims at minimizing the edge-cuts across different processors for minimum communication cost and distributing roughly equal number of mesh vertices for load balance. Then a parallel multi-scale surface mesh refinement phase is iteratively performed to meet the mesh density criteria followed by a parallel surface mesh optimization phase moving vertices to the model boundary so as to fit model geometry feature dynamically. A dynamic load balancing algorithm is performed to change the partition interfaces if necessary. A 3D local non-Delaunay mesh repair algorithm is finally done on the shared interfaces across processors and model boundaries. The experimental results demonstrate our method can achieve high parallel performance and perfect scalability, at the same time preserve model boundary feature and generate high quality 3D Delaunay mesh as well.     

三维地质界面网格模型构建算法研究

三维地质模型主要包含地质构造模型和地质属性模型。提出一种局部映射-边界控制的曲面三角化网格模型构建算法,与映射法相比,减少三维空间点映射到二维平面的计算过程,避免因多点到一点的映射关系而生成错误的三角化网格模型。基于地质测量数据特点,原始地质数据经处理后采用点集合形式表示,基于点集数据构建三维三角化网格模型,模拟地质界面的展布形态,控制三角网格质量。采用两种网格边界控制方法,在有边界约束数据和无边界约束数据条件下均能自动更新地质界面三角化网格模型边界。基于断层点数据集测试并展示算法构建的三维三角化网格模型可视化效果,通过断层面三角化网格模型能够反映断层面之间空间位置关系。     

基于Hermite插值的网格拼接和融合

网格模型的拼接和融合是3维形状编辑和造型中的一个重要方面。基于Hermite插值技术,提出一种适用于具有一般边界点空间分布的三角网格模型之间无缝光滑拼接和融合方法。首先查找网格模型待拼接区域的边缘点集,并利用二次B样条曲线插值边缘点集分别得到边缘曲线;然后对边缘曲线进行Hermite插值得到拼接区域连续曲面;最后对拼接曲面分别进行三角网格化和Laplacian光顺平滑处理以实现网格模型的光滑拼接和融合。由于利用B样条曲线插值待拼接模型边界,本文方法适用于具有各种不同边界情形的网格模型拼接和融合,它不仅仅可以处理平面边界曲线情形也可以处理空间边界曲线情形。结合Hermite曲面插值拼接过渡区域,使得产生的拼接网格能光滑地衔接待拼接模型。实验结果表明,本文方法能够有效地实现三角网格模型的光滑拼接、模型修复和模型融合。