{"title":"A training strategy based on iterative self-training enhanced transfer learning for accurate structural time history response prediction","authors":"Qitao Yang , Zuohua Li , Shujuan Ma , Jiafei Ning","doi":"10.1016/j.compstruc.2025.107840","DOIUrl":"10.1016/j.compstruc.2025.107840","url":null,"abstract":"<div><div>Seismic time history responses are vital for structural performance evaluation and deep learning technologies have shown promise in accelerating the response calculation. However, effective training of deep neural networks generally necessitates extensive data, while their acquisition remains constrained by the time-consuming refined simulations and the high-cost experiments. Here a novel Iterative Self-training Enhanced Transfer Learning (ISTL) method is proposed for deep neural network training to improve seismic time history response prediction accuracy, especially in data scarcity scenarios. ISTL method leverages self-training method to augment abundant samples without additional experiments while integrating domain adaptation with novel output conditional distribution regularization to enhance learning through augmented knowledge. Three experiments validate the proposed method by predicting responses from numerical simulations of a nonlinear frame structure, shake-table testing of a linear frame structure, and field-sensing records of an instrumented shear-wall structure. The results show that ISTL method can eliminate the need for additional experiments and improve prediction performance by up to 60 % over conventional direct training process, underscoring its potential for developing robust predictive models for structural time history responses.</div></div>","PeriodicalId":50626,"journal":{"name":"Computers & Structures","volume":"316 ","pages":"Article 107840"},"PeriodicalIF":4.4,"publicationDate":"2025-06-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144253805","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A. Motte , C. Bovet , V. Chiaruttini , O. Jamond , B. Prabel
{"title":"Massively parallel tandem traversal of bounding volume hierarchies for geometric queries on distributed meshes","authors":"A. Motte , C. Bovet , V. Chiaruttini , O. Jamond , B. Prabel","doi":"10.1016/j.compstruc.2025.107843","DOIUrl":"10.1016/j.compstruc.2025.107843","url":null,"abstract":"<div><div>The efficient computation of geometric queries on large sets of elements, such as intersections or nearest neighbor searches, is crucial for many applications such as contact detection in mechanics simulations. While hierarchical data structures based on spatial or geometric partitioning have been extensively studied over the past 40 years, research detailing their adaptation to massively distributed environments using the message passing interface library—specifically how the communications are done and optimized for parallel efficiency—remains under-explored. We quantify the impact of various design choices to achieve rapid contact detection in massively distributed simulations using bounding volume hierarchies. The memory layout is optimized by reducing the hierarchy node size and ordering them with an innovative “L-filling curve” pattern. Next, the impact of a tandem traversal algorithm is analyzed, along with the optimal exploration order of both hierarchies’ nodes. The parallelization of this tandem traversal is presented, focusing on balancing message latency and volume to enhance parallel efficiency. These design choices extend to a wide range of applications and data structures. Finally, the algorithm’s strong performance is demonstrated up to 20,000 MPI tasks to handle tens of billions of intersections on very large meshes exceeding 100 billion elements.</div></div>","PeriodicalId":50626,"journal":{"name":"Computers & Structures","volume":"316 ","pages":"Article 107843"},"PeriodicalIF":4.4,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144229807","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Approximated and neural network assisted quadratic programming and its applications in structural topology optimization","authors":"Yi Xing, Liyong Tong","doi":"10.1016/j.compstruc.2025.107825","DOIUrl":"10.1016/j.compstruc.2025.107825","url":null,"abstract":"<div><div>In this article, we present approximated and Neural Network assisted Quadratic Programming (NNaQP) methods to accelerate the solution process for solving the constrained and unconstrained optimization problems using gradient and Hessian matrix of objective function. Firstly, three schemes are presented for the approximation of diagonalizing inversed Hessian matrix; Secondly, the NNaQP method is presented by using the gradient online learning and prediction (GoLap) to learn and predict gradient and approximated inverted Hessian matrix. Several scaling and restoration schemes for GoLap are also proposed. The combination of NNaQP and the three approximation schemes of inversed Hessian matrix reduces the number of routine iterations involving complex derivative computing and thus decreases the total computational time. Thirdly, the three approximation schemes of the Hessian and the NNaQP are used to solve structural topology optimization problems. The performance and the benefits of approximated quadratic programming and NNaQP, in terms of prediction accuracy and the computational efficiency, are demonstrated by numerical results of solving one unconstrained minimization problem, and one 2D and one 3D minimum compliance topology optimization problems. For the selected structural topology optimization problems, the total computational timesaving can reach up to 98 %.</div></div>","PeriodicalId":50626,"journal":{"name":"Computers & Structures","volume":"316 ","pages":"Article 107825"},"PeriodicalIF":4.4,"publicationDate":"2025-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144213280","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Robustness of a flux-intensifying permanent magnet-assisted synchronous reluctance machine focusing on shifted surface-inset ferrite magnets","authors":"Mihály Katona, Tamás Orosz","doi":"10.1016/j.compstruc.2025.107845","DOIUrl":"10.1016/j.compstruc.2025.107845","url":null,"abstract":"<div><div>Flux-intensifying permanent magnet-assisted synchronous reluctance machines use relatively small amounts of non-rare earth permanent magnets, making them viable alternatives for remanufacturing older machines, aligning with EU directives and circular economy principles. The asymmetric rotor topology is particularly suited for micromobility applications, which benefit from shifting inset magnets, as reverse motoring is rarely required. However, this design could be more sensitive to manufacturing and positioning errors of the magnets. To investigate the effects of the uncertainties of the shifted surface inset magnets, first, an optimal topology is selected based on average torque, torque ripple, and cogging torque using the NSGA-II optimisation method. The effects of the magnet shifting and its robustness are analysed using the Taguchi and ANOVA methods, validated by Full Factorial calculations. Results indicate a 31.25 % reduction in permanent magnet volume without compromising torque output with magnet shifting. The machine’s average and cogging torque remain within a 5 % robustness threshold for a <span><math><mo>±</mo></math></span>0.06 mm discrete manufacturing tolerance. Torque ripple may exceed this limit up to 14.77 %. However, the likelihood of exceeding the threshold is only 12.10 %. The reduced magnet volume and maintained performance make this topology a promising option for remanufactured machines in micromobility applications, supporting circular economy goals.</div></div>","PeriodicalId":50626,"journal":{"name":"Computers & Structures","volume":"316 ","pages":"Article 107845"},"PeriodicalIF":4.4,"publicationDate":"2025-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144203899","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Marcin Łoś , Tomasz Służalec , Paweł Maczuga , Askold Vilkha , Carlos Uriarte , Maciej Paszyński
{"title":"Collocation-based robust variational physics-informed neural networks (CRVPINNs)","authors":"Marcin Łoś , Tomasz Służalec , Paweł Maczuga , Askold Vilkha , Carlos Uriarte , Maciej Paszyński","doi":"10.1016/j.compstruc.2025.107839","DOIUrl":"10.1016/j.compstruc.2025.107839","url":null,"abstract":"<div><div>Physics-informed neural networks (PINNs) have been widely used to solve partial differential equations (PDEs) through strong residual minimization formulations. Their extension to weak scenarios via Variational PINNs (VPINNs) has been shown to lack robustness when the discrete and continuous-level norms are mismatched. Robust Variational PINNs (RVPINNs) address this problem by appropriately incorporating the Gram matrix but suffer from high computational costs due to the weak residual integration and the Gram matrix inversion. In this work, we accelerate RVPINN computations by using a point-collocation approach similar to PINNs, and by employing an LU factorization of the sparse Gram matrix. This leads to the proposed Collocation-Based Robust Variational PINN (CRVPINN). We validate CRVPINN on Laplace, advection–diffusion, Stokes, non-linear stationary Navier–Stokes, and linear elasticity problems in two spatial dimensions, demonstrating improved efficiency without compromising robustness.</div></div>","PeriodicalId":50626,"journal":{"name":"Computers & Structures","volume":"316 ","pages":"Article 107839"},"PeriodicalIF":4.4,"publicationDate":"2025-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144195184","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Explicit topology optimization for piezoelectric energy harvester with ensured connectivity of polarization profile","authors":"Weisheng Zhang , Qi Lai , Jian Zhang","doi":"10.1016/j.compstruc.2025.107836","DOIUrl":"10.1016/j.compstruc.2025.107836","url":null,"abstract":"<div><div>This paper proposes an explicit topology optimization approach for simultaneous design of both the structure and polarization profile of piezoelectric energy harvesters (PEHs), with particular emphasis on the connectivity of polarization profile. The approach is established under the Moving Morphable Component (MMC) framework. Two independent sets of MMC components are adopted to simultaneously describe the structural topology of the PEH and the polarization profile of the piezoelectric material. Resorting to the explicit geometry description of MMC, the connectivity of the polarization profile related to PEH manufacturability can be addressed in a straightforward manner using graph theory. A novel connectivity constraint is formulated by considering both connected and disconnected graphs. Meanwhile, the proposed approach is applicable to both subtractive and additive manufacturing strategies under different manufacturing processes. Some numerical examples are provided for cantilever-type PEH. The results demonstrate the effectiveness of the proposed approach in improving energy harvesting efficiency and the connectivity control of the polarization profile for PEH.</div></div>","PeriodicalId":50626,"journal":{"name":"Computers & Structures","volume":"316 ","pages":"Article 107836"},"PeriodicalIF":4.4,"publicationDate":"2025-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144203901","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Strong form meshless analysis of solids using partially constrained polynomial finite-difference operators and overdetermined equilibrium","authors":"S. Sharma , N. Potnuru , P. Sideris","doi":"10.1016/j.compstruc.2025.107841","DOIUrl":"10.1016/j.compstruc.2025.107841","url":null,"abstract":"<div><div>This paper proposes a new strong-form meshless method for arbitrary (non-rectangular) grids that combines a new class of finite-difference (FD) differential operators with overdetermined equilibrium. The proposed new class of FD operators combine short-range interpolation with long-range approximation over their support. Polynomial approximation is realized using weighted least squares, whereas polynomial interpolation is enforced using Lagrange multipliers. This partially constrained polynomial (PCP) FD operator is implemented through a proposed algorithm that automatically selects the stencil size and short range radius, such that the numerical error of the stencil weights due to potential ill-conditioning is below a selected tolerance. Using the PCP-FD operators, a meshless method for 2D elasto-static problems was formulated. Instabilities that are often inherent in strong form methods are addressed by introducing additional equations of field equilibrium over a set of additional points within the solid/domain. The resulting overdetermined set of equations is solved as a constrained least squares problem using Lagrange multipliers to enforce traction BCs, which bypasses the need for scaling of the traction BC equations, as is often the case in weighted collocation methods. The proposed meshless method is evaluated in several examples and is shown to achieve high accuracy that is usually better than that of the finite element method.</div></div>","PeriodicalId":50626,"journal":{"name":"Computers & Structures","volume":"316 ","pages":"Article 107841"},"PeriodicalIF":4.4,"publicationDate":"2025-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144203900","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A dissipative orthotropic non-local approach for objective modelling of masonry structures","authors":"Héloïse Rostagni , Cédric Giry , Frédéric Ragueneau","doi":"10.1016/j.compstruc.2025.107802","DOIUrl":"10.1016/j.compstruc.2025.107802","url":null,"abstract":"<div><div>This paper presents the theoretical formulation and numerical implementation of an anisotropic non-local model for materials with intrinsic orthotropy. The model incorporates the mechanisms of orthotropic elasticity, orthotropic damage, shear friction and unilateral effect. To restore the objectivity of the results with respect to the mesh when scaling up to the structure, energetic and non-local regularisations are considered by integrating the orientation of the masonry arrangement in the regularisation domain. An orthotropic non-local averaging scheme is specially designed to capture the preferred damage path in masonry structures linked to the architectural features of the material. Numerical examples are presented to demonstrate the effectiveness of the non-local enhancement in preventing mesh dependency.</div></div>","PeriodicalId":50626,"journal":{"name":"Computers & Structures","volume":"315 ","pages":"Article 107802"},"PeriodicalIF":4.4,"publicationDate":"2025-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144189696","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Stanislav Sysala , Michal Béreš , Simona Bérešová , Tomáš Luber , Zdeněk Michalec
{"title":"Advanced continuation and iterative methods for slope stability analysis in 3D","authors":"Stanislav Sysala , Michal Béreš , Simona Bérešová , Tomáš Luber , Zdeněk Michalec","doi":"10.1016/j.compstruc.2025.107842","DOIUrl":"10.1016/j.compstruc.2025.107842","url":null,"abstract":"<div><div>This paper addresses the solution of slope stability problems in 3D using the finite element method and incremental procedures like the shear strength reduction or limit load methods. We build on Mohr–Coulomb plasticity, Davis’ modifications of the non-associated plastic flow rule and recent mathematical results, which relate the factor of safety (FoS) with convex optimization. A complex solution concept is presented in detail and completed with in-house developed, publicly available open-source MATLAB codes. The concept consists of a combination of indirect continuation techniques, inexact Newton-like solvers and deflated Krylov methods with preconditioners. Further, mesh adaptivity is used to reduce overestimation of FoS and determine failure zones more accurately. The solution concept is tested on slope stability benchmarks in 3D and its efficiency is demonstrated. Numerical results are validated against either literature or software COMSOL Multiphysics.</div></div>","PeriodicalId":50626,"journal":{"name":"Computers & Structures","volume":"315 ","pages":"Article 107842"},"PeriodicalIF":4.4,"publicationDate":"2025-05-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144184649","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}