Mesh generation and optimization from digital rock fractures based on neural style transfer

The complex geometric features of subsurface fractures at different scales makes mesh generation challenging and/or expensive.In this paper,we make use of neural style transfer(NST),a machine learning technique,to generate mesh from rock fracture images.In this new approach,we use digital rock fractures at multiple scales that represent’content’and define uniformly shaped and sized triangles to represent’style’.The 19-layer convolutional neural network(CNN)learns the content from the rock image,including lower-level features(such as edges and corners)and higher-level features(such as rock,fractures,or other mineral fillings),and learns the style from the triangular grids.By optimizing the cost function to achieve approximation to represent both the content and the style,numerical meshes can be generated and optimized.We utilize the NST to generate meshes for rough fractures with asperities formed in rock,a network of fractures embedded in rock,and a sand aggregate with multiple grains.Based on the examples,we show that this new NST technique can make mesh generation and optimization much more efficient by achieving a good balance between the density of the mesh and the presentation of the geometric features.Finally,we discuss future applications of this approach and perspectives of applying machine learning to bridge the gaps between numerical modeling and experiments.

Robust construction of minimal surface from general initial mesh

We analyze three commonly used energy functions in solving Plateau-Mesh Prob- lem, that is, Dirichlet, area, and the discrete mean curvature（DMC）. They all possess unique advantages compared to others, but their drawbacks restrict their usages individually. Our algo- rithm combines the three steps together to make full use of their features. At first the Dirichlet energy is optimized for faster approximation with better topology. Then the area energy is used to come close to the constrained domain. Finally the DMC energy is engaged to achieve a better converging step. Results show that our method can work under a rather noisy initial mesh, which is even topologically different from the final result.

Combined Size and Shape Optimization of Structures with DOE,RSM and GA

In this paper,size and shape optimization problem of a machine gun system is addressed with an efficient hybrid method,in which a novel and flexible mesh morphing technique is employed to achieve fast parameterization and modification of complexity structure without going back to CAD for reconstruction of geometric models or to finite element analysis（ FEA） for remodeling. Design of experiments（ DOE） and response surface method（ RSM） are applied to approximate the constitutive parameters of a machine gun system based on experimental tests. Further FEA,secondary development technique and genetic algorithm（ GA） are introduced to find all the optimal solutions in one go and the optimal design of the demonstrated machine gun system is obtained. Results of the rigid-flexible coupling dynamic analysis and exterior ballistics calculation validate the proposed methodology,which is relatively time-saving,reliable and has the potential to solve similar problems.

Modeling Methods and Test Verification of Root Insert Contact Interface for Wind Turbine Blade

Two modeling methods of the root insert for wind turbine blade are presented,i.e.,the local mesh optimization method(LMOM)and the global modeling method(GMM).Based on the optimized mesh of the local model for the metal contact interface,LMOM is proposed to analyze the load path and stress distribution characteristics,while GMM is used to calculate and analyze the stress distribution characteristics of the resin layer established between the bushing and composite layers of root insert.To validate the GMM,a tension test is carried out.The result successfully shows that the shear strain expresses a similar strain distribution tendency with the GMM′s results.

A survey on image and video stitching

Image/video stitching is a technology for solving the field of view(FOV)limitation of images/videos.It stitches multiple overlapping images/videos to generate a wide-FOV image/video,and has been used in various fields such as sports broadcasting,video surveillance,street view,and entertainment.This survey reviews image/video stitching algorithms,with a particular focus on those developed in recent years.Image stitching first calculates the corresponding relationships between multiple overlapping images,deforms and aligns the matched images,and then blends the aligned images to generate a wide-FOV image.A seamless method is always adopted to eliminate such potential flaws as ghosting and blurring caused by parallax or objects moving across the overlapping regions.Video stitching is the further extension of image stitching.It usually stitches selected frames of original videos to generate a stitching template by performing image stitching algorithms,and the subsequent frames can then be stitched according to the template.Video stitching is more complicated with moving objects or violent camera movement,because these factors introduce jitter,shakiness,ghosting,and blurring.Foreground detection technique is usually combined into stitching to eliminate ghosting and blurring,while video stabilization algorithms are adopted to solve the jitter and shakiness.This paper further discusses panoramic stitching as a special-extension of image/video stitching.Panoramic stitching is currently the most widely used application in stitching.This survey reviews the latest image/video stitching methods,and introduces the fundamental principles/advantages/weaknesses of image/video stitching algorithms.Image/video stitching faces long-term challenges such as wide baseline,large parallax,and low-texture problem in the overlapping region.New technologies may present new opportunities to address these issues,such as deep learning-based semantic correspondence,and 3D image stitching.Finally,this survey discusses the challenges of image/video stitching and proposes potential solutions.

Enhanced optimal delaunay triangulation methods with connectivity regularization

In this paper,we study the underlying properties of optimal Delaunay triangulations(ODT)and propose enhanced ODT methods combined with connectivity regularization.Based on optimizing node positions and Delaunay triangulation iteratively,ODT methods are very effective in mesh improvement.This paper demonstrates that the energy function minimized by ODT is nonconvex and unsmooth,thus,ODT methods suffer the problem of falling into a local minimum inevitably.Unlike general ways that minimize the ODT energy function in terms of mathematics directly,we take an outflanking strategy combining ODT methods with connectivity regularization for this issue.Connectivity regularization reduces the number of irregular nodes by basic topological operations,which can be regarded as a perturbation to help ODT methods jump out of a poor local minimum.Although the enhanced ODT methods cannot guarantee to obtain a global minimum,it starts a new viewpoint of minimizing ODT energy which uses topological operations but mathematical methods.And in terms of practical effect,several experimental results illustrate the enhanced ODT methods are capable of improving the mesh furtherly compared to general ODT methods.

The ICM method with objective function transformed by variable discrete condition for continuum structure

ICM （Independent Continuous Mapping） method can solve topological optimization problems with the minimized weight as the objective and subjected to displacement constraints. To get a clearer topological configuration, by introducing the discrete condition of topological variables and integrating with the original objective, an optimal model with multi-objectives is formulated to make the topological variables approach 0 or 1 as near as possible, and the model reduces the effect of deleting rate on the result. The image-filtering method is employed to eliminate the checkerboard patterns and mesh dependence that occurred in the topology optimization of a continuum structure. The computational efficiency is enhanced through selecting quasi-active displacement constraints and a design region. Numerical examples indicate that this algorithm is robust and practicable, though the number of iterations is slightly increased with respect to the original algorithm.

Removing local irregularities of triangular meshes with highlight line models

The highlight line model is a powerful tool in assessing the quality of a surface. It increases the flexibility of an interactive design environment. In this paper, a method to generate a highlight line model on an arbitrary triangular mesh is presented. Based on the highlight line model, a technique to remove local shape irregularities of a triangular mesh is then presented. The shape modification is done by solving a minimization problem and performing an iterative procedure. The new technique improves not only the shape quality of the mesh surface, but also the shape of the highlight line model. It provides an intuitive and yet suitable method for locally optimizing the shape of a triangular mesh.

3D finite element modelling optimization for deep tunnels with material nonlinearity

3D modeling of tunnels using a nonlinear ground model is still a time-consuming task because it usually requires a large number of incremental phases with iterative processes,to ensure accuracy while minimizing computational effort.Optimization of thefinite element mesh is of utmost importance.Despite the current tendency towards 3D modeling of tunnels,few publications are concerned with mesh optimization considering model size,grid refinement and order of elements.This paper improves the understanding of key issues that affect 3D modeling of tunnels.Our results shown that:(1)2nd order elements are more effcient when material nonlinearity is present and should be preferred;(2)the plastic zone size has a strong influence on the model dimensions and may require discretizations much larger than those currently accepted.The paper provides recommendations for mesh refinement and model dimensions(width and length)as a function of the plastic zone size,for accurate 3D models with reduced computational cost.

Robust and Quality Boundary Constrained Tetrahedral Mesh Generation

A novel method for boundary constrained tetrahedral mesh generation is proposed based on Advancing Front Technique(AFT)and conforming Delaunay triangulation.Given a triangulated surface mesh,AFT is firstly applied to mesh several layers of elements adjacent to the boundary.The rest of the domain is then meshed by the conforming Delaunay triangulation.The non-conformal interface between two parts of meshes are adjusted.Mesh refinement and mesh optimization are then preformed to obtain a more reasonable-sized mesh with better quality.Robustness and quality of the proposed method is shown.Convergence proof of each stage as well as the whole algorithm is provided.Various numerical examples are included as well as the quality of the meshes.

Optimal contract wall for desired orientation of fibers and its effect on flow behavior

The orientation of suspended fibers in the turbulent contraction is strongly related to the contraction ratio,which in some cases may be detrimental to the actual production.Here for a certain contraction ratio,the contraction geometry shape is optimized to obtain the desired fiber orientation.In view of the nonlinearity and the complexity of the turbulent flow equations,the parameterized shape curve,the dynamic mesh and a quasi-static assumption are used to model the contraction with the variable boundary and to search the optimal solution.Furthermore the Reynolds stress model and the fiber orientation distribution function are solved for various wall shapes.The fiber orientation alignment at the outlet is taken as the optimization objective.Finally the effect of the wall shape on the flow mechanism is discussed in detail.