Computer-Aided Civil and Infrastructure Engineering最新文献

{"title":"CoMT: A coupled multi-task learning algorithm for UAV-based road damage detection","authors":"Xuanyu Ge, Yilong Ren, Han Jiang, Liang Xu, Haiyang Yu","doi":"10.1016/j.cacaie.2026.100187","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100187","url":null,"abstract":"UAV-based road damage detection is essential for intelligent infrastructure maintenance, as it achieves continuous monitoring and timely assessment of pavement conditions. Existing approaches commonly formulate this task as a object detection problem, while overlooking a critical structural prior: road damage inherently occurs within road regions. Consequently, object-detection-based predictions often suffer from limited spatial credibility, particularly in complex scenes containing shadows, lane markings, and other damage-like visual patterns. To address these limitations, this paper proposes CoMT, a coupled multi-task learning algorithm that simultaneously performs road damage detection and road region segmentation in an end-to-end manner. Specifically, a coupled multi-task branch is designed to promote cross-task collaboration through bidirectional feature sharing and interaction. In this design, road-region structural cues serve to constrain damage localization, while damage-related semantic features provide complementary contextual information for region segmentation. Moreover, an orthogonal attention module is introduced to strengthen direction-sensitive damage representations by modeling feature consistency and discrepancy along horizontal and vertical orientations. Furthermore, a spatial-distance fusion loss is developed by integrating Complete IoU and Normalized Wasserstein Distance, enabling the joint optimization of spatial overlap and distributional similarity for more robust bounding-box regression. Experimental results demonstrate that CoMT achieves excellent performance, with 72.3% and 82.0% mAP@50 on UAV-PDD2023 and BDD100K datasets, respectively. The proposed framework also achieves real-time inference and competitive segmentation performance, highlighting its effectiveness and practical potential for UAV-based intelligent road infrastructure inspection and maintenance.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"159 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148767552","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":"QuakeF2G: A region-specific transformer-based ground motion prediction model considering fault segments modeling","authors":"Yitian Feng, Weiqiang Zhu, Xinzheng Lu","doi":"10.1016/j.cacaie.2026.100183","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100183","url":null,"abstract":"Rupture directivity remains a major challenge for near-fault ground motion prediction, particularly for urban regions located close to active faults. Empirical ground motion prediction equations (GMPEs), even with directivity correction factors, often oversimplify rupture geometries, while physics-based wave propagation simulations are computationally prohibitive for rapid and dense analysis. To bridge this gap, we present QuakeF2G, a Transformer-based architecture for probabilistic ground motion prediction that fuses heterogeneous geometric and observational inputs through a unified cross-attention framework. To avoid explicit rupture parameterization, the model tokenizes the rupture as a set of discrete fault segments, encodes each with location and orientation attributes, and fuses the resulting fault sequence with station tokens through cross-attention augmented by a geometry-aware relative-position bias. A flexible masking strategy enables the same model to perform pure prediction from fault geometry and site metadata alone, or conditional interpolation when PGV observations are available at some stations. After training on approximately 626,000 synthetic scenarios for peak ground velocity (PGV) from the CyberShake dataset in the Southern California, QuakeF2G achieves a speedup of more than three orders of magnitude in prediction time compared to physics-based simulations while maintaining high predictive fidelity. Ablation experiments demonstrate that explicitly incorporating finite-fault geometry reduces root-mean-square error (RMSE) by 36% compared to a point-source baseline. In spatial interpolation, the model outperforms Ordinary Kriging across all sampling densities, achieving a lower RMSE using only 50% of training stations than Ordinary Kriging achieves with all training stations. QuakeF2G demonstrates a general methodology for fusing irregular geometric inputs with sparse observations in engineering prediction tasks.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"6 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148767664","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":"Decision-oriented computational framework for drought hazard-state classification with a California case study","authors":"Fan Hu, Khalid M. Mosalam","doi":"10.1016/j.cacaie.2026.100166","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100166","url":null,"abstract":"This study develops a decision-oriented computational framework for multi-level drought hazard-state classification under class imbalance and class-boundary ambiguity. Using station-level climate observations from the Global Historical Climatology Network-Daily (GHCND) and operational drought labels from the U.S. Drought Monitor (USDM), the framework is evaluated through a case study of 48 monitoring stations across California from 2010 to 2023. An initial six-class formulation, comprising No Drought, D0 abnormally dry conditions, and four drought severity levels (D1 to D4), is assessed using several machine learning models, including <ce:italic>logistic regression</ce:italic>, <ce:italic>support vector machine</ce:italic>, <ce:italic>k-nearest neighbors</ce:italic>, <ce:italic>multilayer perceptron</ce:italic>, and <ce:italic>random forest</ce:italic>. Results reveal systematic misclassification near the transitional No Drought–D0 boundary, reflecting class imbalance, categorical proximity, and the inherent ambiguity of early drought conditions. To address this challenge, a two-stage classification framework is proposed. In the first stage, a <ce:italic>decision tree</ce:italic> model separates No Drought from drought-relevant D0–D4 conditions for operational screening. In the second stage, a <ce:italic>random forest</ce:italic> model further classifies routed drought-relevant records into D0–D4 hazard states. Additional validation experiments, including a persistence baseline, end-to-end two-stage evaluation, temporal holdout, spatial holdout, feature ablation, feature importance, transition-week diagnostics, and missing-data auditing, are conducted to assess the robustness and limitations of the framework beyond random-split accuracy. Under the expanded stratified random split, the single-stage RF achieved slightly higher aggregate accuracy (0.9739 versus 0.9684) and weighted F1-score (0.9740 versus 0.9686) than the end-to-end two-stage framework. The value of the two-stage formulation lies instead in its explicit separation of screening and D0–D4 hazard-state classification, stage-specific diagnosis, and decision-structured interpretation. Although California serves as the case study, the proposed framework is formulated to support adaptation to other regions and can support nowcasting-oriented drought hazard assessment, infrastructure resilience screening, regional monitoring, and risk-informed resource allocation where comparable monitoring data are available.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"9 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148717251","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":"Multi-agent deep reinforcement learning with centralized training and decentralized execution for transportation infrastructure management","authors":"M. Saifullah, K.G. Papakonstantinou, A. Bhattacharya, S.M. Stoffels, C.P. Andriotis","doi":"10.1016/j.cacaie.2026.100175","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100175","url":null,"abstract":"Life-cycle management of large-scale transportation systems is a computationally intensive task. It requires determining a sequence of inspection and maintenance decisions to minimize long-term risks and costs while dealing with multiple uncertainties and constraints that lie in high-dimensional spaces. Traditional approaches, such as static age- or condition-based maintenance and risk-based or periodic inspection plans, have been widely applied but often suffer from limitations related to optimality, scalability, and the ability to properly handle uncertainty in the loop of optimization. Moreover, many existing methods rely on unconstrained formulations that overlook critical hard and soft operational constraints. We address these issues in this work by casting the optimization problem within the framework of constrained Partially Observable Markov Decision Processes (POMDPs), which provide a robust mathematical foundation for stochastic sequential decision-making under observation uncertainties, in the presence of risk and resource limitations. To tackle the high dimensionality of state and action spaces, we propose DDMAC-CTDE, a Deep Decentralized Multi-Agent Actor-Critic (DDMAC) reinforcement learning architecture with Centralized Training and Decentralized Execution (CTDE). To demonstrate the utility of the proposed framework, we also develop a new comprehensive benchmark environment representing an existing transportation network in Virginia, U.S., with heterogeneous pavement and bridge assets undergoing nonstationary degradation. This environment incorporates multiple practical constraints related to budget limits, performance guidelines from State and Federal agencies, traffic delays, and risk considerations. On this benchmark, DDMAC-CTDE consistently outperforms standard transportation management baselines, producing more reliable, cost- and risk-efficient policies that respect operational constraints. Together, the proposed framework and benchmark provide (i) a scalable, constraint-aware methodology for sequential decision-making under uncertainty, and (ii) a realistic, rigorous testbed for comprehensive evaluation of Deep Reinforcement Learning (DRL) solutions for transportation infrastructure management.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"11 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148717249","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":"Hybrid Image-based Crack Characteristic Extraction in Reinforced Concrete Experiments","authors":"Morteza M. Hagh, Stephen J. Foster, Hamid Valipour, Evan Bentz","doi":"10.1016/j.cacaie.2026.100184","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100184","url":null,"abstract":"Controlling crack widths is essential to the serviceable design of reinforced concrete structures. Yet cracking behaviour in concrete is inherently random, making it challenging to model its key characteristics mathematically without extensive empirical data. Recent advancements in digital image correlation (DIC) techniques enable high-resolution measurements that capture this variability; however, existing crack-width prediction models, particularly for diagonal shear cracks, are based on historical panel tests that relied on manual measurements. These crack-characteristic measurements are limited in scope, often omitting key parameters such as crack spacing and inclination, and are insufficiently robust for serviceability studies. This study addresses these limitations by proposing a hybrid, image-based method that incorporates compatibility to extract crack characteristics, such as spacing, inclination, and width, in a quasi-continuous manner, thereby capturing the probabilistic component of cracking. The performance of the proposed algorithm is evaluated against manual measurements using a dataset consisting of shear panel experiments reported in the literature. Validation results indicate coefficients of determination (<ce:italic>R<ce:sup loc=\"post\">2</ce:sup></ce:italic>) of 0.91 and 0.96 for reported average crack widths and spacings, respectively, with coefficients of variation (CoV) of 0.16 and 0.20, respectively. The performance and scope of the proposed algorithm are further compared with those of existing image-based techniques, and its primary advantage is its ability to extract high-resolution crack data from both past and current experiments. The algorithm’s viability for probabilistic serviceability studies is further demonstrated by its ability to evaluate the characteristic behaviour of diagonal shear cracks in reinforced concrete elements.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"17 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148717248","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":"Terrain-perception-based determination of tunnel locations for mountain railway alignment optimization","authors":"Qizhe Lu, Taoran Song, Paul Schonfeld, Cuntian Ren, Hao Pu, Wei Guo","doi":"10.1016/j.cacaie.2026.100182","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100182","url":null,"abstract":"The determination of tunnel locations is a critical aspect of railway planning and alignment design, especially in mountainous regions whose complex terrain challenges traditional methods. To address this issue, a tunnel location determination method based on detailed terrain analysis is proposed for mountain railway alignment design that integrates spectral feature analysis with multi-constraint optimization. First, terrain data are parameterized as spatial waves using a spatial Fourier transform, which abstracts the topographic information into frequency-domain representations. Low-pass filtering is then applied to retain macro-scale terrain features while suppressing high-frequency insensitive topographic data, thereby highlighting key topographic structures relevant to long tunnel locations. Subsequent steps involve further valley identification via local ternary patterns and density-based clustering to refine candidate tunnel portal regions. Afterward, feasible tunnel portal combinations are screened and connected through a customized shortest path search algorithm considering multiple constraints, with generated paths further optimized into smooth railway alignments. The proposed method is validated through a complex mountain railway case. The sensitivity analysis finds that the best low-pass filter cutoff frequency for identifying major tunnel locations is 1/2,000 m⁻¹ in this example. Then, compared with the best manually-designed solution, the proposed method can find an alternative that reduces the overall engineering cost by 5.71%, demonstrating its effectiveness in solving real-world problems.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"14 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148717246","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":"Size Effect on Cross-Scale Uncertainty Propagation in Concrete Elements: Insights from the Lattice Discrete Particle Model and Explainable Data-Driven Modeling","authors":"Baixi Chen, Alessandro Fascetti","doi":"10.1016/j.cacaie.2026.100179","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100179","url":null,"abstract":"Heterogeneous cementitious composites exhibit pronounced multiscale characteristics, which lead to uncertainty and size effects in the macroscopic mechanical behavior. This study aims to reveal the size effect in the propagation of uncertainty from mesoscale features to macroscale responses of concrete through a combination of mesoscale simulations and data-driven modeling. At the mesoscale, two sources of uncertainty are considered: mesoscale mechanical uncertainty, characterized by seven mesoscale mechanical parameters, and mesoscale spatial uncertainty, represented by the spatial distribution of the constituent phases. At the macroscale, both the stress–strain response and fracture patterns are investigated. The Lattice Discrete Particle Model (LDPM) is first employed to perform mesoscale simulations and establish the relationship between the mesoscale features and macroscale responses. Subsequently, Monte Carlo simulation (MCS) is used to propagate uncertainty across scales. Based on the data generated from MCS, an explainable stochastic data-driven model is developed using heterogeneous sparse Gaussian process regression (HSGPR) to link mesoscale uncertainty to macroscale responses and to interpret how specimen size influences the uncertainty propagation mechanisms and governs the underlying mesoscale uncertainty drivers. The results provide mechanistic insights that can guide the selection of representative element sizes for stochastic multiscale analysis of concrete structures and inform mesoscale material design for controlling macroscale structural performance.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"10 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148691045","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":"Shear behavior of reinforced concrete beams: A stochastic lattice approach for material spatial variability","authors":"Dawei Gu, Mladena Luković","doi":"10.1016/j.cacaie.2026.100178","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100178","url":null,"abstract":"Conventional deterministic approaches fail to consistently predict the shear capacity of reinforced concrete (RC), particularly for beams without transverse reinforcement. This study investigates the influence of material spatial randomness on the shear behavior of RC using a stochastic discrete lattice modeling framework. To understand the effect of heterogeneity, both concrete and mortar beams are investigated. Material strength is represented as a spatially correlated random field following a mixed Gauss-Weibull distribution, and multiple realizations with different autocorrelation lengths and marginal distribution parameters are analyzed. The results reveal that, for concrete beams without stirrups, spatial randomness induces substantial scatter in predicted shear capacities and crack patterns. Even for identical beam geometries and autocorrelation lengths, the maximum predicted shear capacity can be up to 1.9 times the corresponding minimum across different stochastic realizations. This variability is governed by the competition between weak-region and local averaging effects within the critical damage localization zone, rather than by a classical weakest-link mechanism. In contrast, beams with stirrups exhibit smaller realization-to-realization scatter because transverse reinforcement stabilizes diagonal cracking and provides a stable shear-resisting contribution. Extended validation using independent experimental datasets further confirms that the proposed stochastic lattice framework can capture the underlying fracture mechanisms, the size-effect on nominal shear strength, and the response scatter in repeated tests. These findings demonstrate that shear behavior of RC beams, particularly without stirrups, should be interpreted probabilistically rather than through a single deterministic prediction. The study provides new insight into the intrinsic variability of shear resistance and offers a fundamental explanation for the long-standing difficulty in achieving universally accurate shear predictions.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"345 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148691050","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":"Progressive lower bound evaluation for urban road renovation scheduling","authors":"Robbert Bosch, Patricia Rogetzer, Wouter van Heeswijk, Martijn Mes","doi":"10.1016/j.cacaie.2026.100174","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100174","url":null,"abstract":"Urban infrastructure degrades over time, necessitating periodic renovation to maintain functionality and safety. When renovation is delayed beyond the infrastructure’s remaining lifespan, costly emergency interventions become necessary to prevent failure. Decision makers must therefore balance expected emergency intervention costs against traffic congestion impacts. We formalize this trade-off as a road network maintenance scheduling problem with uncertain deadlines, which presents optimization challenges including computationally expensive evaluation and an exponentially growing solution space. To address these challenges, this paper contributes a hybrid optimization approach combining machine learning with genetic algorithms for large-scale infrastructure renovation scheduling under uncertainty. We formulate the problem as a bi-level multi-objective optimization problem that explicitly accounts for uncertain infrastructure lifespans through probabilistic failure models. We develop a progressive lower bound evaluation method: candidate schedules receive quick, optimistic estimates of their traffic impact from machine learning based surrogate models, and these estimates are refined by exact traffic simulation only for promising candidates, so that full evaluation is reserved almost exclusively for schedules that can advance the Pareto front. Embedded in a multi-objective genetic algorithm, this evaluation strategy improves solution quality by enabling many more iterations within a fixed computational budget. We demonstrate the method’s effectiveness and scalability on substantially larger problem instances than previously addressed in the literature: 76 and 80 projects on the Sioux Falls and Anaheim networks, respectively. Under an identical 24-hour computational budget across 30 random seeds, the method completes up to 87 times as many optimization iterations as standard NSGA-II and achieves statistically significant improvements across multiple performance metrics over both standard NSGA-II and a surrogate-assisted baseline. For road agencies, these gains make city-scale renovation programming tractable within a single day of computation: planners obtain a Pareto front of implementable schedules that quantifies the trade-off between the risk of late renovations and network congestion, from which a preferred schedule can be selected and defended, or adjustments explored with their consequences quantified.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"32 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148691109","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":"MV-HOD-UAM: A Multi-View Heterogeneous Origin–Destination Proxy Demand Forecasting Model for Scenario-Based Urban Air Mobility Planning","authors":"Zhirong Zhang, Yong Fang, Zhanyi Li, Lili Pei, Jiazhen Li, Qixiu Cheng","doi":"10.1016/j.cacaie.2026.100176","DOIUrl":"https://doi.org/10.1016/j.cacaie.2026.100176","url":null,"abstract":"Urban air mobility (UAM) has emerged as a promising solution for future urban transportation. However, the lack of real-world UAM operational data limits the characterization of actual UAM demand patterns. Therefore, ground taxi OD records are widely used to construct proxy demand representations for early-stage UAM planning. Existing studies often rely on proxy demand scenarios but remain limited in capturing OD structural dependencies and contextual influences under different UAM deployment scenarios. To address this issue, this paper proposes MV-HOD-UAM, a Multi-View Heterogeneous Origin-Destination Proxy Demand Forecasting Model for scenario-based UAM planning. The proposed context-aware multi-view learning framework integrates heterogeneous information into a unified OD representation space and captures proxy demand evolution, functional differences between origins and destinations, and OD-pair interactions through structured modeling. Furthermore, a dynamic directed dependency learning mechanism is developed to characterize asymmetric and time-varying relationships among OD pairs by incorporating historical OD patterns and contextual information. A context-aware temporal attention decoder is also introduced to adaptively identify informative historical time steps under varying scenarios. Experiments on a simulated air taxi dataset demonstrate that MV-HOD-UAM consistently outperforms competitive baselines. Compared with the strongest baseline, MV-HOD-UAM reduces MSE, RMSE, MAE, and MAPE by 27.31%, 15.17%, 15.49%, and 12.57%, respectively. Ablation results further verify the contributions of structured multi-view OD modeling and context-aware temporal attention, demonstrating the effectiveness of structured OD representations for proxy demand forecasting in early-stage UAM planning.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"452 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2026-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148691961","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}