Computer-Aided Design最新文献

ADPINet: Attention-based discrete physics-informed network for solving geometric PDEs ADPINet：求解几何偏微分方程的基于注意力的离散物理信息网络

IF 3.1 3区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Computer-Aided Design

Pub Date : 2026-05-01 Epub Date: 2026-02-11 DOI: 10.1016/j.cad.2026.104051

Ting Zhang, Hanchao Liu, Shengjun Liu, Xinru Liu

Geometric partial differential equations (GPDEs), defined on Riemannian manifolds, play a fundamental role in modeling surface evolution processes in science and engineering. Traditional numerical methods, however, often require costly remeshing and recomputation when handling dynamically evolving surfaces or mesh refinements. Physics-informed neural networks (PINNs), which leverage automatic differentiation to compute derivatives and embed physical constraints directly into neural network training, offer a promising alternative. Yet, unlike classical PDEs defined on fixed domains, GPDEs involve time-varying computational surfaces, where differential operators defined on irregular discrete meshes cannot be directly handled through automatic differentiation. This poses unique challenges for applying conventional PINNs to irregular geometric domains. To address these issues, we propose an attention-based discrete physics-informed neural network (ADPINet) architecture for explicitly solving GPDEs defined on triangular mesh surfaces. The proposed method adopts a dual-discrete neural framework in both time and space, enabling direct learning on irregular spatiotemporal domains. An attention-based backbone is designed to extract global correlations among mesh vertices, while a physics-informed loss formulated via discrete differential geometry ensures physical consistency with governing equations without requiring labeled data. Once trained, ADPINet can efficiently predict solutions on refined meshes or newly introduced vertices by directly inputting their coordinates, without retraining or interpolation. Furthermore, ADPINet demonstrates shape-level generalization, being able to predict GPDE solutions for different initial surfaces with similar topological and geometric structures. Extensive numerical experiments, including comparisons with traditional numerical solvers and alternative neural architectures, demonstrate the superior accuracy, mesh-quality preservation, computational efficiency, and generalization capability of the proposed method.

在黎曼流形上定义的几何偏微分方程（GPDEs）在科学和工程表面演化过程的建模中起着重要作用。然而，当处理动态变化的表面或网格细化时，传统的数值方法通常需要昂贵的重新网格划分和重新计算。物理信息神经网络（pinn）利用自动微分来计算导数，并将物理约束直接嵌入神经网络训练中，提供了一个很有前途的替代方案。然而，与定义在固定域上的经典偏微分方程不同，GPDEs涉及时变计算曲面，其中定义在不规则离散网格上的微分算子不能通过自动微分直接处理。这对将传统pin应用于不规则几何区域提出了独特的挑战。为了解决这些问题，我们提出了一种基于注意力的离散物理信息神经网络（ADPINet）架构，用于显式求解三角网格表面上定义的gpde。该方法在时间和空间上采用双离散神经网络框架，实现了不规则时空域的直接学习。基于注意力的主干设计用于提取网格顶点之间的全局相关性，而通过离散微分几何制定的物理通知损失确保与控制方程的物理一致性，而不需要标记数据。经过训练后，ADPINet可以通过直接输入其坐标来有效地预测精细网格或新引入的顶点的解，而无需重新训练或插值。此外，ADPINet展示了形状级泛化，能够预测具有相似拓扑和几何结构的不同初始表面的GPDE解。大量的数值实验，包括与传统数值求解器和替代神经结构的比较，证明了该方法具有优越的精度、网格质量保存、计算效率和泛化能力。

{"title":"ADPINet: Attention-based discrete physics-informed network for solving geometric PDEs","authors":"Ting Zhang, Hanchao Liu, Shengjun Liu, Xinru Liu","doi":"10.1016/j.cad.2026.104051","DOIUrl":"10.1016/j.cad.2026.104051","url":null,"abstract":"<div><div>Geometric partial differential equations (GPDEs), defined on Riemannian manifolds, play a fundamental role in modeling surface evolution processes in science and engineering. Traditional numerical methods, however, often require costly remeshing and recomputation when handling dynamically evolving surfaces or mesh refinements. Physics-informed neural networks (PINNs), which leverage automatic differentiation to compute derivatives and embed physical constraints directly into neural network training, offer a promising alternative. Yet, unlike classical PDEs defined on fixed domains, GPDEs involve time-varying computational surfaces, where differential operators defined on irregular discrete meshes cannot be directly handled through automatic differentiation. This poses unique challenges for applying conventional PINNs to irregular geometric domains. To address these issues, we propose an attention-based discrete physics-informed neural network (ADPINet) architecture for explicitly solving GPDEs defined on triangular mesh surfaces. The proposed method adopts a dual-discrete neural framework in both time and space, enabling direct learning on irregular spatiotemporal domains. An attention-based backbone is designed to extract global correlations among mesh vertices, while a physics-informed loss formulated via discrete differential geometry ensures physical consistency with governing equations without requiring labeled data. Once trained, ADPINet can efficiently predict solutions on refined meshes or newly introduced vertices by directly inputting their coordinates, without retraining or interpolation. Furthermore, ADPINet demonstrates shape-level generalization, being able to predict GPDE solutions for different initial surfaces with similar topological and geometric structures. Extensive numerical experiments, including comparisons with traditional numerical solvers and alternative neural architectures, demonstrate the superior accuracy, mesh-quality preservation, computational efficiency, and generalization capability of the proposed method.</div></div>","PeriodicalId":50632,"journal":{"name":"Computer-Aided Design","volume":"194 ","pages":"Article 104051"},"PeriodicalIF":3.1,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146196901","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Anisotropic mesh spacing prediction using neural networks 基于神经网络的各向异性网格间距预测

IF 3.1 3区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Computer-Aided Design

Pub Date : 2026-04-01 Epub Date: 2026-02-02 DOI: 10.1016/j.cad.2026.104040

Callum Lock, Oubay Hassan, Ruben Sevilla, Jason Jones

This work presents a framework to predict near-optimal anisotropic spacing functions suitable to perform simulations with unseen operating conditions or geometric configurations. The strategy consists of utilising the vast amount of high-fidelity data available in industry to compute a target anisotropic spacing and train an artificial neural network to predict the spacing for unseen scenarios. The trained neural network outputs the metric tensor at the nodes of a coarse background mesh that is then used to generate meshes for unseen cases. Examples are used to demonstrate the effect of the network hyperparameters and the training dataset on the accuracy of the predictions. The potential of the method is demonstrated for examples involving up to 11 geometric parameters on CFD simulations involving a full aircraft configuration.

这项工作提出了一个框架来预测近乎最优的各向异性间距函数，适合在不可见的操作条件或几何配置下进行模拟。该策略包括利用工业中可用的大量高保真数据来计算目标各向异性间距，并训练人工神经网络来预测未知场景的间距。经过训练的神经网络在粗背景网格的节点处输出度量张量，然后用于生成未见情况的网格。用实例证明了网络超参数和训练数据集对预测精度的影响。该方法的潜力在涉及整个飞机结构的CFD模拟中涉及多达11个几何参数的例子中得到了证明。

引用次数: 0

Advances and challenges in surface–surface intersection computation — An overview 曲面相交计算的进展与挑战综述

IF 3.1 3区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Computer-Aided Design

Pub Date : 2026-04-01 Epub Date: 2026-02-06 DOI: 10.1016/j.cad.2026.104039

Kai Li , Jieyin Yang , Xiaohong Jia

Surface–Surface Intersection (SSI) represents a foundational challenge in Computer-Aided Design (CAD) and numerous engineering disciplines. The lack of topologically consistent surface intersection algorithms stands as a primary factor to the poor reliability of CAD systems. SSI algorithms form the backbone of Boolean operations in CAD modeling, and their numerical inaccuracies as well as topological incorrectness are among the foremost causes of the notorious watertightness issues in CAD models. Such issues directly compromise the integrity of subsequent simulation and manufacturing workflows, underscoring the critical need for robust SSI solutions.

In this review, we systematically analyze the challenges and recent breakthroughs in SSI computation. We dissect representative methodologies and evaluate their performance across geometric accuracy, topological consistency, and computational efficiency. We further delve into applications of SSI in Boolean operation, collision detection and physical simulation of Computer-Aided Manufacturing (CAM).

曲面-曲面交叉（SSI）是计算机辅助设计（CAD）和众多工程学科的基础挑战。缺乏拓扑一致的曲面相交算法是导致CAD系统可靠性差的主要原因。SSI算法构成了CAD建模中布尔运算的主干，它们的数值不准确性以及拓扑不正确性是CAD模型中臭名昭著的水密性问题的主要原因之一。这些问题直接损害了后续仿真和制造工作流程的完整性，强调了对强大的SSI解决方案的迫切需求。在这篇综述中，我们系统地分析了SSI计算面临的挑战和最近的突破。我们剖析了代表性的方法，并评估了它们在几何精度、拓扑一致性和计算效率方面的性能。我们进一步深入研究了SSI在布尔运算、碰撞检测和计算机辅助制造（CAM）物理模拟中的应用。

引用次数: 0

Conditioned Numerical Shape Functions on Unfitted Reduced Coarse Elements for Robust Analysis of Complex Solid Structures 复杂实体结构鲁棒分析中非拟合简化粗单元的条件数值形状函数

IF 3.1 3区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Computer-Aided Design

Pub Date : 2026-04-01 Epub Date: 2026-01-16 DOI: 10.1016/j.cad.2026.104038

Wei Chen, Ming Li

The unfitted finite element (FE) methods offer significant advantages in engineering analysis by embedding the structure within a simple background mesh and eliminating the need for complex and labor-intensive mesh generation. A high solution accuracy can still be achieved via constructing, on each coarse element, standard polynomial shape functions, or numerical (piecewise) shape functions. However, the strategy incurs significant technical challenges due to the unavoidable occurrence of cut elements of arbitrarily small size that may greatly deteriorate the condition number of the stiffness matrix. To address the issue, we propose the concept of reduced coarse elements. By formulating high-order numerical shape functions as the product of a boundary interpolator and a boundary–interior mapping, a detailed condition number analysis reveals the dependence of the numerical shape function conditioning on the boundary interpolator. Based on these findings, we develop a new type of reduced coarse elements and their associated numerical shape functions to address the conditioning challenges. The stability, accuracy, convergence rate, and efficiency of the approach are tested through various numerical examples in comparison with other cutting-edge approaches. Its performance on a multi-material printed circuit board (PCB) example of 183 million fine mesh nodes is also tested.

非拟合有限元（FE）方法通过将结构嵌入简单的背景网格中，消除了复杂和劳动密集型网格生成的需要，在工程分析中具有显著的优势。通过在每个粗元上构造标准多项式形状函数或数值（分段）形状函数，仍然可以获得较高的解精度。然而，由于不可避免地会出现任意小尺寸的切削元件，这可能会大大降低刚度矩阵的条件数，因此该策略面临着重大的技术挑战。为了解决这个问题，我们提出了简化粗元素的概念。通过将高阶数值形状函数表述为边界插值器与边界-内映射的乘积，详细分析了数值形状函数条件对边界插值器的依赖性。基于这些发现，我们开发了一种新型的简化粗元及其相关的数值形状函数来解决条件的挑战。通过各种数值算例与其他前沿方法进行比较，验证了该方法的稳定性、准确性、收敛速度和效率。在多材料印刷电路板（PCB）的1.83亿个细网格节点上进行了性能测试。

{"title":"Conditioned Numerical Shape Functions on Unfitted Reduced Coarse Elements for Robust Analysis of Complex Solid Structures","authors":"Wei Chen, Ming Li","doi":"10.1016/j.cad.2026.104038","DOIUrl":"10.1016/j.cad.2026.104038","url":null,"abstract":"<div><div>The unfitted finite element (FE) methods offer significant advantages in engineering analysis by embedding the structure within a simple background mesh and eliminating the need for complex and labor-intensive mesh generation. A high solution accuracy can still be achieved via constructing, on each coarse element, standard polynomial shape functions, or numerical (piecewise) shape functions. However, the strategy incurs significant technical challenges due to the unavoidable occurrence of cut elements of arbitrarily small size that may greatly deteriorate the condition number of the stiffness matrix. To address the issue, we propose the concept of <em>reduced coarse elements</em>. By formulating high-order numerical shape functions as the product of a boundary interpolator and a boundary–interior mapping, a detailed condition number analysis reveals the dependence of the numerical shape function conditioning on the boundary interpolator. Based on these findings, we develop a new type of reduced coarse elements and their associated numerical shape functions to address the conditioning challenges. The stability, accuracy, convergence rate, and efficiency of the approach are tested through various numerical examples in comparison with other cutting-edge approaches. Its performance on a multi-material printed circuit board (PCB) example of 183 million fine mesh nodes is also tested.</div></div>","PeriodicalId":50632,"journal":{"name":"Computer-Aided Design","volume":"193 ","pages":"Article 104038"},"PeriodicalIF":3.1,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146006530","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Paver: Element-based pattern creation on 3D free-form surfaces 铺路器：在3D自由曲面上创建基于元素的图案

IF 3.1 3区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Computer-Aided Design

Pub Date : 2026-04-01 Epub Date: 2026-02-03 DOI: 10.1016/j.cad.2026.104031

Weidan Xiong , Ziyu Hu , Yongli Wu , Peng Song , Jianmin Zheng

This paper presents Paver, an interactive tool for designing structured patterns composed of 3D elements on free-form 3D surfaces. To address the inherent complexity of direct pattern creation in 3D space, we introduce a parametric design model that allows users to define patterns on a 2D parameter plane. Each pattern is defined by an underlying tessellation and isometric elements. A bijective mapping then lifts these 2D designs onto the target input surface. To ensure structural coherence and accommodate surface curvature, we model the tessellation as a 2D mass–spring system and propose an optimization method to adjust the particle positions on the 2D parameter plane under well-designed system forces. Once optimized, the tessellation is transferred to the 3D surface, generating the final element-based pattern. Paver supports diverse pattern types and offers real-time feedback, enabling designers to iteratively explore and refine complex surface decorations with ease. Experimental results are presented to demonstrate the effectiveness of our Paver.

本文介绍了一种交互式工具Paver，用于在自由曲面上设计由三维元素组成的结构化图案。为了解决在3D空间中直接创建模式的固有复杂性，我们引入了一个参数化设计模型，允许用户在2D参数平面上定义模式。每个图案都由底层镶嵌和等距元素定义。然后，双物镜映射将这些2D设计提升到目标输入表面。为了保证结构的一致性和适应表面曲率，我们将曲面镶嵌建模为一个二维质量-弹簧系统，并提出了一种优化方法来调整粒子在二维参数平面上的位置。一旦优化，镶嵌被转移到3D表面，生成最终的基于元素的图案。Paver支持多种图案类型，并提供实时反馈，使设计师能够轻松地迭代探索和完善复杂的表面装饰。实验结果证明了所设计的摊铺机的有效性。

引用次数: 0

AAGATNet: An attentive graph network for machining feature recognition on free-form surfaces AAGATNet：一个用于自由曲面加工特征识别的细心图网络

IF 3.1 3区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Computer-Aided Design

Pub Date : 2026-04-01 Epub Date: 2026-01-31 DOI: 10.1016/j.cad.2026.104041

Feiqi Wang , Yehong Cao , Haidong Wang , Siqi Yin , Xin Hu , Xiangyang Cui

Machining feature recognition is a critical enabling technology for intelligent manufacturing, yet existing methods, particularly deep learning-based approaches, exhibit significant performance limitations on complex CAD models with free-form surfaces. This challenge is compounded by the scarcity of benchmarks that capture such geometric complexity. To address these gaps, we make two primary contributions. First, we construct MFInstSeg++, a large-scale benchmark dataset comprising over 67,000 complex CAD models, which systematically incorporates machining features on and around B-Spline surfaces. To support research reproducibility, a representative subset (MFInstSeg++ Mini) is made publicly available. Second, we propose AAGATNet, a novel attentive graph neural network that builds upon and advances the capabilities of AAGNet, a highly effective prior model. AAGATNet advances the Geometric Attribute Adjacency Graph (gAAG) representation by incorporating a rich set of extended geometric and topological attributes. Furthermore, it introduces a GATv2-based Attention Refinement Layer (ARL) that enables the model to dynamically weigh the importance of neighboring faces, facilitating a more context-aware feature aggregation. Extensive experiments show that AAGATNet significantly outperforms the AAGNet baseline on our challenging MFInstSeg++ dataset, improving the key metrics of mIoU and F1-Score by 2.89% and 0.93% respectively, while maintaining excellent performance on the original MFInstSeg benchmark. Our work provides both a more realistic benchmark and a more powerful model, representing a significant advancement in feature recognition, particularly for CAD models involving complex free-form surfaces.

加工特征识别是智能制造的关键技术，但现有方法，特别是基于深度学习的方法，在具有自由曲面的复杂CAD模型上表现出显着的性能限制。由于缺乏能够捕捉这种几何复杂性的基准，这一挑战变得更加复杂。为了弥补这些差距，我们做出了两项主要贡献。首先，我们构建了mfinstseg++，这是一个包含超过67,000个复杂CAD模型的大型基准数据集，它系统地包含了b样条表面及其周围的加工特征。为了支持研究的可重复性，一个具有代表性的子集（mfinstseg++ Mini）是公开可用的。其次，我们提出了AAGATNet，这是一种新型的关注图神经网络，它建立在AAGNet的基础上并提高了AAGNet的能力，AAGNet是一种高效的先验模型。AAGATNet通过结合丰富的扩展几何和拓扑属性集来改进几何属性邻接图（gAAG）表示。此外，它引入了一个基于gatv2的注意力细化层（ARL），使模型能够动态权衡相邻人脸的重要性，从而促进更具上下文感知的特征聚合。大量实验表明，AAGATNet在具有挑战性的MFInstSeg++数据集上的性能明显优于AAGNet基线，mIoU和F1-Score的关键指标分别提高了2.89%和0.93%，同时在原始MFInstSeg基准上保持了优异的性能。我们的工作提供了一个更现实的基准和一个更强大的模型，代表了特征识别的重大进步，特别是对于涉及复杂自由曲面的CAD模型。

{"title":"AAGATNet: An attentive graph network for machining feature recognition on free-form surfaces","authors":"Feiqi Wang , Yehong Cao , Haidong Wang , Siqi Yin , Xin Hu , Xiangyang Cui","doi":"10.1016/j.cad.2026.104041","DOIUrl":"10.1016/j.cad.2026.104041","url":null,"abstract":"<div><div>Machining feature recognition is a critical enabling technology for intelligent manufacturing, yet existing methods, particularly deep learning-based approaches, exhibit significant performance limitations on complex CAD models with free-form surfaces. This challenge is compounded by the scarcity of benchmarks that capture such geometric complexity. To address these gaps, we make two primary contributions. First, we construct MFInstSeg++, a large-scale benchmark dataset comprising over 67,000 complex CAD models, which systematically incorporates machining features on and around B-Spline surfaces. To support research reproducibility, a representative subset (MFInstSeg++ Mini) is made publicly available. Second, we propose AAGATNet, a novel attentive graph neural network that builds upon and advances the capabilities of AAGNet, a highly effective prior model. AAGATNet advances the Geometric Attribute Adjacency Graph (gAAG) representation by incorporating a rich set of extended geometric and topological attributes. Furthermore, it introduces a GATv2-based Attention Refinement Layer (ARL) that enables the model to dynamically weigh the importance of neighboring faces, facilitating a more context-aware feature aggregation. Extensive experiments show that AAGATNet significantly outperforms the AAGNet baseline on our challenging MFInstSeg++ dataset, improving the key metrics of mIoU and F1-Score by 2.89% and 0.93% respectively, while maintaining excellent performance on the original MFInstSeg benchmark. Our work provides both a more realistic benchmark and a more powerful model, representing a significant advancement in feature recognition, particularly for CAD models involving complex free-form surfaces.</div></div>","PeriodicalId":50632,"journal":{"name":"Computer-Aided Design","volume":"193 ","pages":"Article 104041"},"PeriodicalIF":3.1,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146190204","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Self-intersection detection algorithm of sweep surfaces based on geometric features of spine curves 基于脊柱曲线几何特征的扫描曲面自交检测算法

IF 3.1 3区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Computer-Aided Design

Pub Date : 2026-03-01 Epub Date: 2025-12-09 DOI: 10.1016/j.cad.2025.104024

Chuan He , Aizeng Wang , Gang Zhao

In modern CAD/CAM workflows, sweep surfaces play a pivotal role in aerospace and automotive design, yet robust self‐intersection analysis remains an open challenge. This paper presents an approach for detecting self‐intersections in parametric sweep surfaces generated by a planar profile along a planar spine. We first classify self‐intersections into local and global categories and further distinguish those arising from individual offset‐curve interactions versus offset‐strip interactions. For local self‐intersections, we develop an efficient algorithm based on curvature extrema and offset‐distance extrema. For global self‐intersections caused by single offset curve, we address spine‐curve minimal‐distance pairs by decomposing the NURBS spine into curvature‐monotonic and globally convex Bézier segments; and introduce a critical offset distance concept for detecting endpoint‐induced self-intersections. For global self‐intersections caused by offset‐strip self‐intersections, we reduce detection to curve–curve intersection and curve–surface intersection subproblems. Numerical experiments demonstrate the proposed methods’ completeness, numerical stability, and applicability to high‐precision sweep surface modeling and downstream process verification.

在现代CAD/CAM工作流程中，扫描曲面在航空航天和汽车设计中发挥着关键作用，但强大的自交分析仍然是一个开放的挑战。本文提出了一种检测由平面轮廓沿平面脊线生成的参数化扫描曲面的自交的方法。我们首先将自交分为局部和全局两类，并进一步区分由个别偏移曲线相互作用和偏移带相互作用产生的自交。对于局部自交，我们提出了一种基于曲率极值和偏移距离极值的有效算法。对于由单偏移曲线引起的全局自交，我们通过将NURBS脊柱分解为曲率单调段和全局凸段来解决脊柱-曲线最小距离对；并引入了检测端点诱导自交的临界偏移距离概念。对于偏置条自交引起的全局自交，我们将检测简化为曲线-曲线交和曲线-曲面交子问题。数值实验证明了该方法的完备性、数值稳定性以及对高精度扫描曲面建模和下游工艺验证的适用性。

{"title":"Self-intersection detection algorithm of sweep surfaces based on geometric features of spine curves","authors":"Chuan He , Aizeng Wang , Gang Zhao","doi":"10.1016/j.cad.2025.104024","DOIUrl":"10.1016/j.cad.2025.104024","url":null,"abstract":"<div><div>In modern CAD/CAM workflows, sweep surfaces play a pivotal role in aerospace and automotive design, yet robust self‐intersection analysis remains an open challenge. This paper presents an approach for detecting self‐intersections in parametric sweep surfaces generated by a planar profile along a planar spine. We first classify self‐intersections into local and global categories and further distinguish those arising from individual offset‐curve interactions versus offset‐strip interactions. For local self‐intersections, we develop an efficient algorithm based on curvature extrema and offset‐distance extrema. For global self‐intersections caused by single offset curve, we address spine‐curve minimal‐distance pairs by decomposing the NURBS spine into curvature‐monotonic and globally convex Bézier segments; and introduce a critical offset distance concept for detecting endpoint‐induced self-intersections. For global self‐intersections caused by offset‐strip self‐intersections, we reduce detection to curve–curve intersection and curve–surface intersection subproblems. Numerical experiments demonstrate the proposed methods’ completeness, numerical stability, and applicability to high‐precision sweep surface modeling and downstream process verification.</div></div>","PeriodicalId":50632,"journal":{"name":"Computer-Aided Design","volume":"192 ","pages":"Article 104024"},"PeriodicalIF":3.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790214","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Gen-Porous: An INR-based generative framework for multiscale TPMS-like porous structure design and optimization Gen-Porous：一个基于inr的多尺度tpms类多孔结构设计和优化生成框架

IF 3.1 3区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Computer-Aided Design

Pub Date : 2026-03-01 Epub Date: 2025-12-02 DOI: 10.1016/j.cad.2025.104020

Shengfa Wang , Hao Zheng , Jiangbei Hu , Yu Jiang , Liang Du , Hao Du , Na Lei , Zhongxuan Luo

Porous structures originating from triply periodic minimal surfaces (TPMSs), referred to as TPMS-like porous structures, have been widely applied across various domains due to their advantageous structural properties. However, the limited morphological diversity and the simulation challenges posed by the complex features have impeded the further development and practical utilization of TPMS-like porous structures. Recent breakthroughs in artificial intelligence have demonstrated remarkable potential for creativity in industrial design. Capitalizing on these developments, we propose Gen-Porous - a novel framework employing implicit neural representations (INR) to design and optimize TPMS-like porous structures. The proposed methodology delineates a comprehensive paradigm for the acquisition and synthesis of various TPMS-like porous structures through Implicit Neural Representation (INR), facilitating simulation-driven optimization in a meshfree manner directly within the acquired latent space. Initially, we collect a dataset comprising multiscale TPMS-like porous structures, employing implicit functions to achieve this. Subsequently, we utilize an Auto-Decoder network architecture to encode these structures into a compact latent space via implicit neural representation learning, thereby permitting the generation of novel configurations with augmented morphological diversity. Moreover, we develop a differentiable computational framework using neural meshfree simulations. This integration not only makes optimization more efficient, but also simplifies sensitivity analysis through automatic differentiation, enabling diverse constraints and objectives to be more easily incorporated. We demonstrate the efficacy of the framework through a lightweight design case study, where optimized solutions are identified through latent space exploration. The core innovation lies in the neural unification of representation and simulation, achieving significant improvements in computational efficiency by circumventing traditional meshing bottlenecks and enabling full differentiability.

源自三周期极小表面（tpms）的多孔结构被称为类tpms多孔结构，由于其优越的结构特性而被广泛应用于各个领域。然而，有限的形态多样性和复杂特征带来的模拟挑战阻碍了tpms类多孔结构的进一步发展和实际应用。最近人工智能的突破显示了工业设计创造力的巨大潜力。利用这些发展，我们提出了Gen-Porous -一个采用隐式神经表征（INR）来设计和优化类似tpms的多孔结构的新框架。所提出的方法描述了通过隐式神经表征（INR）获取和合成各种类似tpms的多孔结构的综合范例，促进了在获取的潜在空间内直接以无网格方式进行模拟驱动的优化。首先，我们收集了一个包含多尺度tpms类多孔结构的数据集，采用隐式函数来实现这一目标。随后，我们利用Auto-Decoder网络架构通过隐式神经表征学习将这些结构编码到紧凑的潜在空间中，从而允许生成具有增强形态多样性的新配置。此外，我们开发了一个可微计算框架使用神经网格模拟。这种集成不仅提高了优化效率，而且通过自动区分简化了敏感性分析，使不同的约束和目标更容易合并。我们通过轻量级设计案例研究证明了该框架的有效性，其中通过潜在的空间探索确定了优化的解决方案。核心创新在于表征和模拟的神经统一，通过绕过传统网格瓶颈和实现完全可微性，显著提高了计算效率。

{"title":"Gen-Porous: An INR-based generative framework for multiscale TPMS-like porous structure design and optimization","authors":"Shengfa Wang , Hao Zheng , Jiangbei Hu , Yu Jiang , Liang Du , Hao Du , Na Lei , Zhongxuan Luo","doi":"10.1016/j.cad.2025.104020","DOIUrl":"10.1016/j.cad.2025.104020","url":null,"abstract":"<div><div>Porous structures originating from triply periodic minimal surfaces (TPMSs), referred to as TPMS-like porous structures, have been widely applied across various domains due to their advantageous structural properties. However, the limited morphological diversity and the simulation challenges posed by the complex features have impeded the further development and practical utilization of TPMS-like porous structures. Recent breakthroughs in artificial intelligence have demonstrated remarkable potential for creativity in industrial design. Capitalizing on these developments, we propose <em>Gen-Porous</em> - a novel framework employing implicit neural representations (INR) to design and optimize TPMS-like porous structures. The proposed methodology delineates a comprehensive paradigm for the acquisition and synthesis of various TPMS-like porous structures through Implicit Neural Representation (INR), facilitating simulation-driven optimization in a meshfree manner directly within the acquired latent space. Initially, we collect a dataset comprising multiscale TPMS-like porous structures, employing implicit functions to achieve this. Subsequently, we utilize an Auto-Decoder network architecture to encode these structures into a compact latent space via implicit neural representation learning, thereby permitting the generation of novel configurations with augmented morphological diversity. Moreover, we develop a differentiable computational framework using neural meshfree simulations. This integration not only makes optimization more efficient, but also simplifies sensitivity analysis through automatic differentiation, enabling diverse constraints and objectives to be more easily incorporated. We demonstrate the efficacy of the framework through a lightweight design case study, where optimized solutions are identified through latent space exploration. The core innovation lies in the neural unification of representation and simulation, achieving significant improvements in computational efficiency by circumventing traditional meshing bottlenecks and enabling full differentiability.</div></div>","PeriodicalId":50632,"journal":{"name":"Computer-Aided Design","volume":"192 ","pages":"Article 104020"},"PeriodicalIF":3.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145737144","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Feature-aware manifold meshing and remeshing of point clouds and polyhedral surfaces with guaranteed smallest edge length 特征感知的点云和多面体表面的流形网格划分和重划分，保证最小的边缘长度

IF 3.1 3区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Computer-Aided Design

Pub Date : 2026-03-01 Epub Date: 2025-11-17 DOI: 10.1016/j.cad.2025.104010

Henriette Lipschütz , Ulrich Reitebuch , Konrad Polthier , Martin Skrodzki

Point clouds and polygonal meshes are widely used when modeling real-world scenarios. Here, point clouds arise, for instance, from acquisition processes applied in various surroundings, such as reverse engineering, rapid prototyping, or cultural preservation. Based on these raw data, polygonal meshes are created to, for example, run various simulations. For such applications, the utilized meshes must be of high quality. This paper presents an algorithm to derive triangle meshes from unstructured point clouds. The occurring edges have a close to uniform length and their lengths are bounded from below. Theoretical results guarantee the output to be manifold, provided suitable input and parameter choices. Further, the paper presents several experiments establishing that the algorithms can compete with widely used competitors in terms of quality of the output and timing and the output is stable under moderate levels of noise. Additionally, we expand the algorithm to detect and respect features on point clouds as well as to remesh polyhedral surfaces, possibly with features.

Supplementary material, an extended preprint, a link to a previously published version of the article, utilized models, and implementation details are made available online.

点云和多边形网格在建模现实场景时被广泛使用。例如，点云产生于应用于各种环境的获取过程中，例如逆向工程、快速原型或文化保护。基于这些原始数据，创建多边形网格，例如，运行各种模拟。对于这样的应用，使用的网格必须是高质量的。提出了一种从非结构化点云中导出三角形网格的算法。出现的边具有接近均匀的长度，并且它们的长度从下面有界。理论结果保证了输出是多种多样的，提供了合适的输入和参数选择。此外，本文提出了几个实验，证明该算法在输出质量和时序方面可以与广泛使用的竞争对手竞争，并且在中等噪声水平下输出稳定。此外，我们扩展了算法来检测和尊重点云上的特征，以及对多面体表面进行网格划分，可能会有特征。补充材料、扩展的预印本、到文章先前发布版本的链接、使用的模型和实现细节都可以在线获得。

{"title":"Feature-aware manifold meshing and remeshing of point clouds and polyhedral surfaces with guaranteed smallest edge length","authors":"Henriette Lipschütz , Ulrich Reitebuch , Konrad Polthier , Martin Skrodzki","doi":"10.1016/j.cad.2025.104010","DOIUrl":"10.1016/j.cad.2025.104010","url":null,"abstract":"<div><div>Point clouds and polygonal meshes are widely used when modeling real-world scenarios. Here, point clouds arise, for instance, from acquisition processes applied in various surroundings, such as reverse engineering, rapid prototyping, or cultural preservation. Based on these raw data, polygonal meshes are created to, for example, run various simulations. For such applications, the utilized meshes must be of high quality. This paper presents an algorithm to derive triangle meshes from unstructured point clouds. The occurring edges have a close to uniform length and their lengths are bounded from below. Theoretical results guarantee the output to be manifold, provided suitable input and parameter choices. Further, the paper presents several experiments establishing that the algorithms can compete with widely used competitors in terms of quality of the output and timing and the output is stable under moderate levels of noise. Additionally, we expand the algorithm to detect and respect features on point clouds as well as to remesh polyhedral surfaces, possibly with features.</div><div>Supplementary material, an extended preprint, a link to a previously published version of the article, utilized models, and implementation details are made <span><span>available online</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50632,"journal":{"name":"Computer-Aided Design","volume":"192 ","pages":"Article 104010"},"PeriodicalIF":3.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145600401","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Part orientation fused shape optimization for minimisation of print time and material waste in extrusion-based 3D printing 零件方向融合形状优化，以最大限度地减少基于挤压的3D打印的打印时间和材料浪费

IF 3.1 3区计算机科学 Q2 COMPUTER SCIENCE, SOFTWARE ENGINEERING

Computer-Aided Design

Pub Date : 2026-03-01 Epub Date: 2025-12-17 DOI: 10.1016/j.cad.2025.104029

Don Pubudu Vishwana Joseph Jayakody , Bailin Deng , Ravindra S. Goonetilleke , Lauren E.J. Thomas-Seale , Hyunyoung Kim

Support structure generation is a critical requirement in additive manufacturing (AM) to prevent material collapse in overhanging regions. However, it increases print time, material waste, and overall production cost, especially in extrusion-based AM. To mitigate these problems, design engineers often resort to manually finetuning or even redesigning prototype geometry to minimise support structures, which is time-consuming and inefficient. A direct geometric optimisation that preserves locality of shape changes whilst corresponding to the part orientation remains an underdetermined problem.

In this paper, we present a novel alternating optimisation framework that finds the corresponding part geometry and orientation to minimise support structures under minimal geometric deviation. Whilst global-level support structure reduction is realised by the part orientation change, we introduce an efficient energy minimisation-based geometric optimisation framework, which is governed by saliency-aware elementwise projections and a set of manufacturing constraints. The proposed framework is validated through extensive computational and physical printing experiments employing multiple 3D printers and support structure types, on a diverse set of complex models including topologically non-trivial parts such as gyroid structures. Our results show an average reduction of 50 % in support structure print time, 27 % in material usage and 25 % in total print time, demonstrating the effectiveness of the proposed framework and its potential as a paradigm shift in manufacturing-oriented design.

支撑结构生成是增材制造（AM）中防止悬垂区域材料坍塌的关键要求。然而，它增加了打印时间，材料浪费和整体生产成本，特别是在基于挤出的AM中。为了缓解这些问题，设计工程师经常求助于手动微调甚至重新设计原型几何形状，以最大限度地减少支撑结构，这既耗时又低效。一个直接的几何优化，保持局部性的形状变化，同时对应的部分方向仍然是一个未确定的问题。在本文中，我们提出了一种新的交替优化框架，该框架可以在最小的几何偏差下找到相应的零件几何和方向，以最小化支撑结构。虽然全球水平的支撑结构减少是通过改变零件方向来实现的，但我们引入了一个高效的基于能量最小化的几何优化框架，该框架由显著性感知元素投影和一组制造约束控制。所提出的框架通过广泛的计算和物理打印实验进行了验证，该实验采用了多种3D打印机和支撑结构类型，在各种复杂模型上进行了验证，包括拓扑上非琐碎的部件，如陀螺结构。我们的研究结果显示，支撑结构打印时间平均减少了50%，材料使用量减少了27%，总打印时间减少了25%，这表明了所提出框架的有效性及其作为面向制造的设计范式转变的潜力。

{"title":"Part orientation fused shape optimization for minimisation of print time and material waste in extrusion-based 3D printing","authors":"Don Pubudu Vishwana Joseph Jayakody , Bailin Deng , Ravindra S. Goonetilleke , Lauren E.J. Thomas-Seale , Hyunyoung Kim","doi":"10.1016/j.cad.2025.104029","DOIUrl":"10.1016/j.cad.2025.104029","url":null,"abstract":"<div><div>Support structure generation is a critical requirement in additive manufacturing (AM) to prevent material collapse in overhanging regions. However, it increases print time, material waste, and overall production cost, especially in extrusion-based AM. To mitigate these problems, design engineers often resort to manually finetuning or even redesigning prototype geometry to minimise support structures, which is time-consuming and inefficient. A direct geometric optimisation that preserves locality of shape changes whilst corresponding to the part orientation remains an underdetermined problem.</div><div>In this paper, we present a novel alternating optimisation framework that finds the corresponding part geometry and orientation to minimise support structures under minimal geometric deviation. Whilst global-level support structure reduction is realised by the part orientation change, we introduce an efficient energy minimisation-based geometric optimisation framework, which is governed by saliency-aware elementwise projections and a set of manufacturing constraints. The proposed framework is validated through extensive computational and physical printing experiments employing multiple 3D printers and support structure types, on a diverse set of complex models including topologically non-trivial parts such as gyroid structures. Our results show an average reduction of 50 % in support structure print time, 27 % in material usage and 25 % in total print time, demonstrating the effectiveness of the proposed framework and its potential as a paradigm shift in manufacturing-oriented design.</div></div>","PeriodicalId":50632,"journal":{"name":"Computer-Aided Design","volume":"192 ","pages":"Article 104029"},"PeriodicalIF":3.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840006","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0