Computer-Aided Civil and Infrastructure Engineering最新文献

{"title":"High-Fidelity Guided-wave Full Waveform Inversion Regularized by Physics-Informed Deep Generative Model","authors":"Shicheng Xue,Hui Li,Yong Huang,Wensong Zhou,Hongbing Chen,Yun Kong,Zhangxuan Wang","doi":"10.1016/j.cacaie.2026.100218","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100218","url":null,"abstract":"","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"13 1","pages":"100218"},"PeriodicalIF":11.775,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148932505","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quantum deep convolutional neural networks for damage classification","authors":"Abu Bakar Siddique,Youngjin Cha","doi":"10.1016/j.cacaie.2026.100217","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100217","url":null,"abstract":"","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"50 1","pages":"100217"},"PeriodicalIF":11.775,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148932506","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimization of ideal speed profile templates considering the trade-off between headway and running time on high-speed railways","authors":"Zili Shen, Hongxiang Zhang, Hexin Hu, Gongyuan Lu","doi":"10.1016/j.cacaie.2026.100211","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100211","url":null,"abstract":"Headway is a key constraint on the capacity utilization of high-speed railway lines and is closely linked to train speed profiles. Regulating speed profiles offers a potential operational approach to compress train headways without modifying existing infrastructure, but it usually introduces an additional running time penalty. To balance this trade-off, this study proposes an Ideal Speed Profile Template (ISPT) approach. The ISPT represents a normative target speed profile shared by successive trains, aiming to reshape the blocking-time envelope (BTE) at critical locations by adjusting local speed states and braking distances. Microscopic train trajectories and section resource occupation are formulated as a speed profile optimization problem (SPOP) on a high-resolution space-velocity (SV) network. To solve the resulting large-scale problem, an improved adaptive large neighborhood search (ALNS) algorithm is developed. Based on the BTE evaluation, customized destroy and repair operators are designed to identify headway bottlenecks and reconstruct the critical speed-profile segments. Numerical experiments verify the solution quality, computational efficiency, and stability of the proposed algorithm. The results reveal substantial potential for headway compression under a moderate running-time penalty. In the Beijing-Shanghai High-Speed Railway case, the optimized ISPT reduces the headway from 265.73 s to 169.80 s, achieving a 36.10% compression with only a 4.71% increase in running time while slightly reducing total energy consumption. These results indicate that the proposed method is an effective operational strategy for improving railway line capacity while maintaining acceptable running time and energy performance.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"35 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148884906","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"ProbCrack-Diff: A structure-aware probabilistic framework for concrete crack segmentation, quantification, and uncertainty estimation","authors":"Yingjie Wu, Yancheng Li, Mingyang Ren, Jianchun Li","doi":"10.1016/j.cacaie.2026.100203","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100203","url":null,"abstract":"Accurate concrete crack segmentation is critical for reliable structural assessment, yet it remains challenging due to irregular morphology of cracks and the ambiguity of their boundaries. This leads to a fundamental limitation in existing methods: while engineering assessment requires reliable predictive uncertainty estimation, most existing segmentation models produce a single deterministic mask, resulting in unstable geometric measurements and limited interpretability. Recent diffusion-based approaches introduce probabilistic segmentation but often lack explicit structural guidance during the denoising process. To address these limitations, this study proposes ProbCrack-Diff, a structure-aware probabilistic framework for concrete crack segmentation, quantification, and uncertainty estimation. The framework integrates a Lightweight Local–Global Encoder (LLGE) to capture multi-scale structural representations and a Structure-aware Discriminator (SA-Discriminator) to provide geometry-aware supervision through structural distillation. By incorporating crack-specific structural priors into the diffusion process, ProbCrack-Diff promotes the preservation of crack continuity, boundary integrity, and geometric fidelity. Extensive experiments demonstrate that ProbCrack-Diff achieves superior segmentation accuracy and geometric fidelity compared with existing discriminative and diffusion-based approaches. Furthermore, uncertainty–reliability analyses indicate that higher predictive uncertainty estimates are generally associated with larger segmentation and crack-geometry estimation errors, suggesting that the estimated predictive uncertainty may provide useful supplementary information regarding prediction reliability in automated crack assessment.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"71 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148853374","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Estimation of 3D Global Shape Parameters of Calcareous Sand from 2D Projections: Insights from Noise-Generated Synthetic Particles","authors":"Chen-Xi Tong, Hui-Lin Jiang, Jia-Jun Li, Dian-Hong Huang, Sheng Zhang","doi":"10.1016/j.cacaie.2026.100207","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100207","url":null,"abstract":"The complex particle morphology of calcareous sand significantly influences its macroscopic mechanical behavior, making accurate three-dimensional (3D) shape characterization essential for numerical analyses such as Discrete Element Method (DEM) simulations. While two-dimensional (2D) Dynamic Image Analysis (DIA) offers a high-throughput and cost-effective alternative to 3D scanning, systematic quantitative relationships between 2D projections and 3D descriptors remain poorly established for highly irregular geomaterials. This study developed a Worley-noise-based framework to generate large databases of bulky and flaky particles with controllable external 3D morphologies, thereby alleviating the sample-size limitations of natural-particle scanning. Principal-axis and random projections were evaluated, together with the effects of sample size, projection number, regression form, generation parameters, and pixel resolution. A baseline sample size of 500 synthetic particles was selected from convergence analysis. Under principal-axis projection, descriptors governed primarily by global axial proportions, particularly elongation, diameter sphericity, and intercept sphericity, showed the strongest 2D-3D correspondence. By contrast, true sphericity, convexity, and roundness were more sensitive to particle morphology and projection-induced loss of local surface information. Under isotropic random projection, mean descriptor values generally performed better for bulky particles, whereas minimum values were more informative for several descriptors of flaky particles. Practical numerical thresholds of 15 and 20 projection directions were identified for bulky and flaky particles, respectively. Assessment using measured natural particles further confirmed the applicability of the proposed relationships to selected global descriptors within the investigated morphological domain. The framework provides a systematic basis for estimating selected 3D global shape characteristics from 2D particle contours and supports the future calibration of efficient image-based characterization methods for calcareous sand.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"32 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148853375","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Learning to Design: A Collaborative Multi-Agent Reinforcement Learning Framework for Arch Dam Shape Design","authors":"Xiaogang Xie, Gang Ma, Zongliang Zhang, Rui Liu, Yaolai Liu, Zhiping Chen, Wei Zhou","doi":"10.1016/j.cacaie.2026.100206","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100206","url":null,"abstract":"Shape design remains challenging in high arch dam engineering because it involves high-dimensional geometry, strict constraints, and conflicting objectives. Conventional experience-driven iteration is labor-intensive, while surrogate-assisted evolutionary optimization improves efficiency but does not accumulate reusable design knowledge. To address these limitations, this study develops a surrogate-assisted collaborative multi-agent reinforcement learning framework for arch dam shape design. The problem is reformulated as a preference-conditioned sequential decision process, where two specialist agents focus on economic efficiency and structural safety. A hybrid experience-sharing mechanism combining preference relabeling and curriculum-guided scheduling is introduced to improve optimization efficiency and Pareto-front coverage. Active learning is coupled with a probabilistic Multi-gate Mixture-of-Experts surrogate to construct an accurate surrogate environment under limited finite-element budgets. A real-world case study shows slightly faster and more stable surrogate refinement than Latin hypercube sampling, while finite-element validation confirms simultaneous improvements of 1.4% in the economic objective and 10.1% in the safety objective.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"55 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148852988","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"RTK-informed geometry-to-analysis framework for automated finite element modeling of masonry pagodas","authors":"Xulin Zhou, Qiankun Zhu, Qiong Zhang, Yanhui Liu, Yongfeng Du","doi":"10.1016/j.cacaie.2026.100199","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100199","url":null,"abstract":"Masonry pagodas exhibit complex geometries, incomplete structural documentation, and uncertain material properties, making finite element model development highly dependent on manual intervention. This study proposes a Real-Time Kinematic (RTK)-informed framework integrating computer vision-based 3D reconstruction, computer-aided finite element model generation, and frequency-based model updating. Unmanned aerial vehicle(UAV) imagery and RTK observations are integrated through a robust similarity transformation to recover the real-world scale and orientation of the reconstructed geometry. The RTK-informed triangulated surface mesh is subsequently converted automatically into an analysis-ready Abaqus solid model through vertical slicing, contour extraction, arc-length resampling, and cross-layer point correspondence. Validation against total-station measurements demonstrates that the relative geometric discrepancies are below 2%. Non-contact ambient-vibration measurements are then conducted using a laser Doppler vibrometer, and the first three measured natural frequencies are used to calibrate selected material parameters through a simulated annealing–particle swarm optimization (SA-PSO) algorithm. Following model updating, the discrepancies between the measured and calculated frequencies are reduced to below 4%. The proposed framework provides a repeatable mesh-to-analysis workflow for the digital documentation and subsequent structural assessment of masonry heritage structures.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"10 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148853152","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An Adaptive Elasto-Plastic Decomposition Framework for Efficient Nonlinear Dynamic Analysis of Soil-Structure Interaction Systems","authors":"Zhe Lin, Quan Gu, Surong Huang","doi":"10.1016/j.cacaie.2026.100202","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100202","url":null,"abstract":"Nonlinear dynamic analysis of complex soil-structure interaction (SSI) systems is computationally demanding because conventional fully coupled solution procedures require repeated global stiffness updates, matrix factorizations, and nonlinear iterations. However, nonlinearities in many SSI systems are not uniformly distributed over the entire domain. During seismic loading, plasticity and stiffness degradation often develop locally and intermittently in critical regions, such as weak soil layers, soil-foundation interaction zones, and selected structural components, while the remaining part of the system behaves elastically or only weakly nonlinearly. Motivated by this feature, this study proposes an Elasto-Plastic Decomposition (EPD) framework for efficient nonlinear dynamic analysis of SSI systems. The key idea is to retain an invariant elastic main system and to identify evolving nonlinear substructures adaptively during the analysis. By applying the Sherman-Morrison-Woodbury formula, nonlinear stiffness modifications are represented as low-rank corrections to the initial elastic system, whose stiffness matrix is factorized only once. The effects of local plasticity and stiffness degradation are further transferred to the global system through localized corrective forces, enabling an iterative implicit solution procedure without repeated global stiffness refactorization. In this way, nonlinear elements are activated on demand when prescribed state thresholds are exceeded, while their tangent-stiffness changes are included in the localized stiffness correction only when necessary. Furthermore, a limited-iteration strategy is introduced to further improve computational efficiency by reducing unnecessary nonlinear iterations while maintaining an acceptable level of response accuracy. The proposed framework is implemented in OpenSees and validated using two representative examples: a two-dimensional layered soil site with a localized weak zone and a three-dimensional nonlinear pile-soil-structure interaction system with an eight-story reinforced-concrete superstructure. The results show that the EPD framework reduces the measured computational time while maintaining close agreement with the full Newton-Raphson solutions. The proposed method provides an adaptive and platform-compatible computational strategy for nonlinear dynamic analysis of SSI systems with localized and evolving nonlinearities.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"71 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148853150","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Specification-guided generative augmentation and mixture-of-experts learning for stainless steel beam parametric design","authors":"Wenkang Zuo, Yuner Huang, Ben Young","doi":"10.1016/j.cacaie.2026.100200","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100200","url":null,"abstract":"Structural machine learning (ML) is increasingly used for rapid performance prediction and parametric design, but its reliability is often limited by scarce structural data and by the weak formalisation of engineering knowledge embedded in design specifications. This study proposes a specification-guided ML framework that converts specification knowledge into learnable, generative and deployable components for parametric structural design. The framework is demonstrated on cold-formed stainless steel RHS/SHS beams, where AISC 370 provisions define the code baseline, feasibility constraints and expert partitioning rules. A two-stage generative augmentation strategy is developed to mitigate the adverse effects of limited data. Feasible design variables are first conditionally generated, followed by synthesising the structural capacity through residual learning anchored to specification-based resistance predictions. To reflect regime-dependent behaviour, a mixture-of-experts (MoE) predictor is constructed, in which a slenderness-based gate blends non-slender and slender experts. The augmented MoE models substantially improve test accuracy, reduce prediction scatter and mitigate overfitting compared with code calculations and single global learners trained on limited real data. Finally, the trained model is deployed in Rhino–Grasshopper, enabling real-time capacity assessment, geometry visualisation and interactive exploration. The proposed framework demonstrates how design specifications can be formalised as computable priors for data generation, model specialisation and design automation.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"20 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148853151","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DuEx-Crack: A reliability-aware dual-expert framework for cross-domain crack segmentation","authors":"Yuan Dai, Wei Zhou, Tingwei Zhang, Zuen Zheng, Yong Zhu, Kejun Long","doi":"10.1016/j.cacaie.2026.100201","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100201","url":null,"abstract":"Automated crack segmentation for civil infrastructure must deliver both label efficiency and cross-domain robustness. Recent pipelines couple a domain-specific detector with a frozen vision foundation model such as the Segment Anything Model (SAM) in a cascade, and therefore inherit the failure modes of the foundation model: both the regime in which it returns no foreground at all, and the regime in which it returns noisy false positives. We propose DuEx-Crack, a rule-based framework with three core contributions: (i) the <ce:italic>silent rate</ce:italic>, a training-free, label-free, dataset-level probe that quantifies how often a frozen foundation model returns an empty mask on the target domain before fusion; (ii) a deterministic adaptive rule driven by the silent rate that selects between connectivity-driven and conservative-union fusion without learned gating or per-dataset tuning; and (iii) a parallel dual-expert architecture that replaces the cascade design, providing structural graceful degradation when the foundation model is silent and bounding rather than propagating the contamination from noisy predictions, as quantified against a true cascade under identical weights. Instantiated with YOLO26-seg-s (11.4 M parameters, bounding-box supervision) and frozen SAM3, DuEx-Crack is evaluated on the Roboflow and CrackVision12 K benchmarks, surpassing the supervised state of the art on Roboflow by <ce:math altimg=\"si108.svg\"></ce:math> IoU with only <ce:math altimg=\"si109.svg\"></ce:math> of the trainable parameters, and outperforming it by <ce:math altimg=\"si110.svg\"></ce:math> and <ce:math altimg=\"si111.svg\"></ce:math> IoU under cross-domain transfer. As an end-to-end engineering integration, a mask-grounded vision-language module converts the fused masks into structured JSON diagnostic reports, demonstrating a practical inspection use case built on the framework.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"6 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148852991","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}