<rss version="2.0" xmlns:atom="https://www.w3.org/2005/Atom">
  <channel>
    <title>Transport Research International Documentation (TRID)</title>
    <link>https://trid.trb.org/</link>
    <atom:link href="https://trid.trb.org/Record/RSS?s=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" rel="self" type="application/rss+xml" />
    <description></description>
    <language>en-us</language>
    <copyright>Copyright © 2026. National Academy of Sciences. All rights reserved.</copyright>
    <docs>http://blogs.law.harvard.edu/tech/rss</docs>
    <managingEditor>tris-trb@nas.edu (Bill McLeod)</managingEditor>
    <webMaster>tris-trb@nas.edu (Bill McLeod)</webMaster>
    <image>
      <title>Transport Research International Documentation (TRID)</title>
      <url>https://trid.trb.org/Images/PageHeader-wTitle.jpg</url>
      <link>https://trid.trb.org/</link>
    </image>
    <item>
      <title>Marine-YOLO: A high-precision object detection algorithm for complex maritime environments</title>
      <link>https://trid.trb.org/View/2676165</link>
      <description><![CDATA[Sea Surface Object Detection is of great significance for intelligent shipping, marine monitoring, and search and rescue. However, in complex sea-surface scenarios, challenges remain, such as severe weather, illumination changes, large variations in object scale, and missed detections of small targets. To address these issues, this paper proposes Marine-YOLO based on YOLOv11 to improve detection accuracy and robustness in complex sea-surface environments. First, Marine-YOLO introduces a C3k2-SP module into the backbone network to enhance feature representation and environmental robustness. Second, a SAGA attention module is added to the neck to strengthen multi-scale modeling capability. Finally, an AFAE-Head module is designed in the detection head, and a P2 layer is incorporated to optimize small-object detection performance. Experimental results on the WSODD dataset show that Marine-YOLO achieves 76.7% and 44.1% on mAP50 and mAP50–95, respectively, representing improvements of 3.5% and 1.3% over YOLOv11n. While achieving higher accuracy, Marine-YOLO (4.2 M parameters and 13.3 GFLOPs) has more parameters and computational cost than YOLOv11n (2.6 M parameters and 6.5 GFLOPs), but still far less than other high-performance models. Compared with RT-DETR-X (67.3 M parameters and 232.4 GFLOPs), Marine-YOLO maintains higher accuracy while reducing parameters and computation by about 94%, and improving mAP50 and mAP50–95 by 12.2% and 13.9%, respectively. The results indicate that Marine-YOLO achieves a good balance between accuracy and efficiency, making it an ideal choice for real-time marine object detection.]]></description>
      <pubDate>Tue, 01 Sep 2026 14:02:30 GMT</pubDate>
      <guid>https://trid.trb.org/View/2676165</guid>
    </item>
    <item>
      <title>Hydrodynamic mechanisms and surrogate-based prediction of resistance reduction in two-catamaran formations</title>
      <link>https://trid.trb.org/View/2737353</link>
      <description><![CDATA[The significance of energy conservation and emission reduction in maritime operations is becoming increasingly prominent, particularly with the rapid development of unmanned and multihull vessels. This study proposes a novel formation model that exploits favorable hydrodynamic interactions between catamarans to improve the energy efficiency of the overall formation or individual vessels. The numerical framework is first assessed through grid-convergence analysis and calm-water resistance validation. The resistance characteristics of two catamarans in tandem, lateral, and parallel formations are then numerically investigated over Fr = 0.2-0.7. A GPR surrogate model was developed from the 336-case CFD database. Validation on an independent test set confirmed its suitability for the rapid screening of energy-efficient formation layouts. The results show that the leading catamaran is only weakly affected, whereas the following catamaran dominates the hydrodynamic benefit in tandem and lateral formations. The resistance response is particularly sensitive near Fr = 0.5, where both substantial resistance reductions and penalties occur. The maximum reduction reaches 41.61% in tandem formation and 51.38% in lateral formation. Resistance reduction can be achieved when a catamaran is positioned near the wave trough of the leading catamaran's wake, where favorable wave superposition occurs.]]></description>
      <pubDate>Wed, 12 Aug 2026 14:59:37 GMT</pubDate>
      <guid>https://trid.trb.org/View/2737353</guid>
    </item>
    <item>
      <title>An enhanced autonomous surface manipulator system for additive manufacturing of polyurethane foam on water surfaces in oceanic manufacturing scenarios</title>
      <link>https://trid.trb.org/View/2733083</link>
      <description><![CDATA[Deploying robots for additive manufacturing (AM) tasks on water surfaces enhances task safety and automation. The autonomous surface manipulator system (ASMS), comprising a manipulator and an unmanned surface vehicle (USV), has the potential for these tasks. However, traditional ASMS has limitations such as poor configuration stability. Hence, this study developed an enhanced ASMS for AM tasks with polyurethane foam on water surfaces. Firstly, the dynamic coupling between the manipulator and the USV is reduced by redesigning the manipulator's structure and fabricating the manipulator's arms using carbon fiber. Secondly, dynamic models for both the traditional and enhanced ASMS are developed using Newton-Euler equations, and dynamic analysis results confirm that the enhanced ASMS exhibits superior configuration stability because the rolling and pitching torques applied by the manipulator to the USV are reduced during printing processes. Thirdly, the relationship between printing parameters and the printing quality of polyurethane foam is investigated on a digital twin-driven hardware-in-the-loop simulation platform, and the optimal printing parameters are determined. Finally, two AM tasks are conducted using ASMS's physical prototypes, involving printing floating platforms for unmanned aerial vehicles landing and printing roads on water surfaces. Experimental results validated that the enhanced ASMS is more suitable for AM tasks.]]></description>
      <pubDate>Tue, 11 Aug 2026 16:35:12 GMT</pubDate>
      <guid>https://trid.trb.org/View/2733083</guid>
    </item>
    <item>
      <title>A hierarchical deep learning framework for few-shot corrosion type identification in natural gas pipelines</title>
      <link>https://trid.trb.org/View/2733037</link>
      <description><![CDATA[Accurate identification of corrosion types is essential for ensuring pipeline integrity. However, due to limited sample availability and complex morphological variations, this task remains challenging. Traditional manual methods are inefficient, and deep learning models often struggle under small-sample conditions. To address this issue, this paper proposes a few-shot corrosion identification method that integrates a hierarchical attention mechanism into ResNet50 (HA-ResNet50). A defect dataset containing various corrosion types is first constructed, and a generative adversarial network is introduced to generate high-quality synthetic samples, alleviating data scarcity. The core idea lies in designing a hierarchical attention mechanism that simulates the human visual cognitive process. This mechanism comprises multiple functional modules that progressively refine corrosion features across different levels and granularities: it first stabilizes low-level features, then focuses on discriminative local regions, and ultimately achieves a global understanding of the entire corrosion scene. Experiments on industrial corrosion datasets demonstrate that the proposed method achieves over 98% classification accuracy under small-sample conditions, significantly outperforming existing baseline models. These results validate its ability to effectively simulate a coherent recognition process from local feature perception to global structural comprehension, offering a reliable and practical solution for intelligent corrosion detection in pipeline integrity management.]]></description>
      <pubDate>Tue, 11 Aug 2026 16:35:12 GMT</pubDate>
      <guid>https://trid.trb.org/View/2733037</guid>
    </item>
    <item>
      <title>Extreme value probability estimation in ship motions based on collaborative MLG-VAE and IDCGAN neural networks</title>
      <link>https://trid.trb.org/View/2737382</link>
      <description><![CDATA[Aiming at the extremity and randomness of ship motion responses in harsh sea conditions, a novel method based on the collaboration of Multi-layer Gumbel-distribution-constrained Variational Autoencoder (MLG-VAE) and Improved Deep Convolutional Generative Adversarial Network (IDCGAN) is proposed to achieve accurate and reliable prediction of ship motion extreme values. First, to address the challenge of scarce extreme motion data, an MLG-VAE model is constructed. The training process is optimized via a triple-constraint loss function, and virtual samples of motion extreme values that match the statistical characteristics of the original data are generated. Meanwhile, an adaptive sliding window technique is adopted to realize precise extraction of extreme value features. On this basis, IDCGAN is utilized to perform high-fidelity fitting of the extreme value distribution for the augmented dataset, and an extreme value probability model is established. Comparative experiments demonstrate that the proposed method outperforms the traditional Generalized Pareto Distribution (GPD) method significantly in the fitting accuracy of extreme value tails. Furthermore, based on the constructed extreme value probability model, short-term and long-term probability extrapolation estimations of ship motions are carried out. Quantitative estimates of motion extreme values under different return periods are provided, which offers effective technical support for navigation safety assessment of ships in irregular waves.]]></description>
      <pubDate>Mon, 10 Aug 2026 15:03:59 GMT</pubDate>
      <guid>https://trid.trb.org/View/2737382</guid>
    </item>
    <item>
      <title>Identification of gas-liquid flow patterns in horizontal pipes using phase fraction-integrated physics-guided neural network</title>
      <link>https://trid.trb.org/View/2737844</link>
      <description><![CDATA[Accurate identification of gas-liquid two-phase flow patterns in pipelines is a key challenge for optimizing offshore oil and gas transport, marine energy systems, and sustainable maritime operations. This paper integrates the gas-liquid flow pattern transition mechanism with a physics-guided neural network (PGNN) and proposes an identification model for gas-liquid flow patterns in horizontal pipes that combines data fitting with physical constraints. A database of horizontal gas-liquid flow patterns is developed with different pipe diameters (12.7 - 300 mm), gas superficial velocities (0.014 - 200.6 m/s), liquid superficial velocities (0.002 - 8.930 m/s), and liquid viscosities (1.002 - 166 mPa⸱s), comprising 5625 data points. Based on the visualization images of gas-liquid flow in a 65 mm ID horizontal pipe, the flow patterns are classified into annular flow, intermittent flow, dispersed bubble flow, and stratified flow, and the flow pattern data in the database are subsequently reclassified. The predictive abilities of 14 phase fraction correlations are evaluated using an existing flow pattern-phase fraction database, and the correlation proposed by Rouhani is found to have the best coefficient of determination and error control (RRMSE = 10.25%, R² = 0.9). As this correlation serves as the phase fraction physics-guided module, PGNN achieves 97.9% flow pattern identification accuracy on the independent test set, which is superior to other models. The effectiveness of the physics-guided module is quantitatively verified, showing high fitting accuracy (R² > 0.93) for physics-guided quantities. In cases where the flow pattern boundary is blurred and difficult to identify, PGNN maintains an accuracy of 96.88%, surpassing models such as random forest (95.3%) and multilayer perceptron (85.2%).]]></description>
      <pubDate>Wed, 05 Aug 2026 09:12:29 GMT</pubDate>
      <guid>https://trid.trb.org/View/2737844</guid>
    </item>
    <item>
      <title>Generalization capabilities and failure modes of parametric surrogate models for harbor wave fields</title>
      <link>https://trid.trb.org/View/2721522</link>
      <description><![CDATA[Rapid tranquility assessment is essential for iterative harbor design; however, high-fidelity Boussinesq-type simulations remain computationally prohibitive. This short communication develops and validates a parametric DNN surrogate model to predict spatial wave fields based on continuous scalar structural inputs. Using both simplified and real bathymetry, the model was evaluated against three critical practical needs: (1) predictive accuracy for wave fields resulting from breakwater length alterations, (2) the generalization capability for unlearned parameters within (interpolation) and outside (extrapolation) the training ranges, and (3) comparative predictability of Hₛ versus ηₘₐₓ. The results revealed a clear contrast in performance. Regarding structural alterations, the model demonstrated exceptional robustness (R² > 0.98) for unlearned breakwater lengths, regardless of interpolation or extrapolation. This confirms that scalar-based parameterization enables high-precision, seamless layout optimization without iterative grid regeneration. Conversely, wave condition extrapolation exhibited significant degradation (R² < 0.5) outside the training range, underscoring the difficulty of extrapolating nonlinear wave physics compared to interpolation. Furthermore, regarding physical quantities, ηₘₐₓ proved more challenging than Hₛ due to phase sensitivity, though accuracy remained practical within the structural domain. These findings establish a vital engineering guideline: while data-driven models are powerful for structural optimization, avoiding extrapolation of wave climates is crucial.]]></description>
      <pubDate>Mon, 27 Jul 2026 09:46:44 GMT</pubDate>
      <guid>https://trid.trb.org/View/2721522</guid>
    </item>
    <item>
      <title>Prediction-driven distributed cooperative interception for multi-unmanned surface vehicles via spatio-temporal fusion</title>
      <link>https://trid.trb.org/View/2721521</link>
      <description><![CDATA[This paper investigates a prediction-driven cooperative interception method for multiple unmanned surface vessels in complex marine environments. A distributed spatio-temporal information fusion framework is developed to improve state consistency and prediction reliability under heterogeneous observations. Each vessel first performs local target-state estimation using an Extended Kalman Filter (EKF), and the resulting local posteriors are fused through covariance-weighted Bayesian information fusion to obtain a globally consistent target state. An exponentially weighted sliding-window mechanism is then introduced at the prediction initialization stage to fuse historical information, enhance temporal continuity, and suppress maneuver-induced prediction jitter. Based on the fused state and stabilized predicted trajectory, a cooperative interception strategy is further designed for proactive pursuit of maneuvering targets. Simulation results in hybrid maneuvering scenarios show that the proposed method achieves higher estimation accuracy, smoother predicted trajectories, and better interception performance than single-USV and non-predictive baselines. The results further indicate that jointly strengthening spatial consistency and history-informed temporal continuity can improve perception reliability, operational robustness, and cooperative interception performance in complex marine environments.]]></description>
      <pubDate>Mon, 27 Jul 2026 09:46:44 GMT</pubDate>
      <guid>https://trid.trb.org/View/2721521</guid>
    </item>
    <item>
      <title>Research on ice particle impact erosion in seawater pipelines of polar vessels</title>
      <link>https://trid.trb.org/View/2718450</link>
      <description><![CDATA[Safety risks associated with ice particle impact on seawater pipelines in polar vessels are investigated in this study. The limitations of existing research regarding ice particle fragmentation and the accurate prediction of erosion behavior are addressed. The mechanical properties of ice particles were evaluated using a coupled CFD-DPM approach, along with paint removal experiments and acoustic emission (AE) technology. The mechanisms underlying ice particle breakage and pipeline erosion were analyzed, and the service life of pipelines constructed from different materials was assessed. A quantitative relationship between impact parameters and material damage was also established. The results reveal that pipeline erosion by ice-laden seawater is a complex process involving kinetic energy-induced fragmentation, flow field-driven aggregation, and turbulence-enhanced collisions. Ice particle fragmentation increased the erosion rate by 25%. Among the materials tested, glass-reinforced epoxy (GRE) showed the poorest erosion resistance, with a service life only one-fifth that of metallic materials. The distribution of erosion hotspots was governed by the orientation of the pipeline elbow, while increases in Stokes number, ice particle diameter, and volume fraction intensified material damage. Future work could integrate shipboard monitoring data with machine learning techniques to enhance the prediction of erosion hotspots.]]></description>
      <pubDate>Mon, 27 Jul 2026 09:46:44 GMT</pubDate>
      <guid>https://trid.trb.org/View/2718450</guid>
    </item>
    <item>
      <title>AI-empowered computational fluid dynamics in ship and ocean engineering: A review of data, algorithms and validation</title>
      <link>https://trid.trb.org/View/2721603</link>
      <description><![CDATA[The integration of artificial intelligence (AI) with computational fluid dynamics (CFD) offers new opportunities to overcome high computational costs and high-dimensional nonlinearities in ship and ocean engineering. This review focuses specifically on AI-assisted CFD in the marine domain, covering four interconnected themes: dataset construction, image-recognition algorithms, surrogate models and flow-field reconstruction, and scientific validation of AI outputs. Following a transparent literature search (2019–2026) with a PRISMA workflow, we critically analyze 124 papers. Key findings include: (1) hybrid “simulation + experiment” datasets balance scale and authenticity; (2) convolutional neural networks and vision transformers, transferred from natural image recognition, show promise for ship detection and flow feature extraction but face physical consistency challenges; (3) neural operators (Fourier neural operator, graph neural operator) and physics-informed neural networks enable fast surrogate modelling, yet require large training datasets and do not inherently guarantee conservation laws; (4) a three-level validation framework (numerical accuracy, physical consistency, generalization) is necessary but lacks standardized benchmarks. We identify data authenticity bias, insufficient interpretability, and missing verification protocols as core gaps. Future research should prioritize physics-constrained learning, multi-fidelity datasets, and cross-industry validation standards.]]></description>
      <pubDate>Mon, 27 Jul 2026 09:46:44 GMT</pubDate>
      <guid>https://trid.trb.org/View/2721603</guid>
    </item>
    <item>
      <title>A Synergistic VMD-LSTM-LightGBM model for significant wave height prediction in the Northern Taiwan Strait</title>
      <link>https://trid.trb.org/View/2725520</link>
      <description><![CDATA[Significant wave height is a critical parameter in marine engineering and maritime safety; therefore, accurate wave height prediction is essential for ensuring the safety of offshore operations. To address the high computational cost of traditional numerical simulation methods and the limitations of data-driven approaches in modeling nonstationary time series, this study proposes a synergistic VMD–LSTM–LightGBM prediction framework based on variational mode decomposition (VMD) and a frequency-differentiated modeling strategy. First, VMD is applied to decompose the wave height time series into intrinsic mode functions (IMFs). Based on their frequency characteristics, long short-term memory (LSTM) is used to model low-frequency components, while the light gradient boosting machine (LightGBM) captures high-frequency components. The final predictions are obtained through reconstruction. Using measured data from multiple buoy stations in the Northern Taiwan Strait, the results demonstrate that the proposed model outperforms several benchmark models, including LSTM, LightGBM, convolutional neural network–LSTM (CNN–LSTM), and random forest (RF) and Support Vector Regression (SVR), in terms of root mean square error (RMSE) and coefficient of determination. Furthermore, the proposed framework exhibits strong generalization capability and robustness in multistation and multistep prediction tasks. Ablation experiments are conducted to verify the contribution of each model component.]]></description>
      <pubDate>Mon, 27 Jul 2026 09:46:44 GMT</pubDate>
      <guid>https://trid.trb.org/View/2725520</guid>
    </item>
    <item>
      <title>Elastic preloading-enhanced parallel-column brace for vibration suppression of deep-sea mining towers</title>
      <link>https://trid.trb.org/View/2725468</link>
      <description><![CDATA[Deep-sea mining towers are installed on the deck of mining vessels and are used for the deployment and recovery of mining equipment. Tower vibration induced by ship rocking can reduce operational efficiency, accelerate fatigue damage, and threaten system safety. In this study, an energy dissipation enhancement strategy based on elastic preloading is proposed. By using a pre-compressed linear spring to regulate the buckling, contact, and configuration transition processes of a parallel-column structure, a novel pre-compressed parallel-column structure (PPCS) is designed. Finite element simulations and theoretical analyses show that, without increasing the peak strain, the energy dissipation capacity of the PPCS can be increased to more than twice that of the original structure. The stiffness–damping performance index is improved from 13 to 27, indicating a significant enhancement in stiffness–damping synergy. Subsequently, an equivalent hysteretic modeling method based on key turning-point extraction and piecewise linear fitting is proposed to introduce the nonlinear hysteretic response of the PPCS into the dynamic model of a deep-sea mining tower. Under ship-induced rocking excitation, the PPCS-braced tower exhibits rapid vibration attenuation, with a dimensionless damping coefficient of ξ = 2.7. These results demonstrate that the proposed PPCS brace provides a high-stiffness, recoverable, and high-energy-dissipation solution for vibration suppression of deep-sea mining towers and other offshore structures.]]></description>
      <pubDate>Mon, 27 Jul 2026 09:46:44 GMT</pubDate>
      <guid>https://trid.trb.org/View/2725468</guid>
    </item>
    <item>
      <title>Toward Novel Smart Ocean Data Collection: A Secure Batch Data Concept in IOTA</title>
      <link>https://trid.trb.org/View/2717494</link>
      <description><![CDATA[Ocean data collection, instrumental for addressing global issues such as climate change and biodiversity conservation, relies on IoT devices deployed on marine vessels. Despite their significance, traditional blockchain-based data collection systems have scalability and energy efficiency limitations. Moreover, ensuring the secure journey of data from its origin to a remote storage location is a critical concern often overlooked. This paper presents a system design fortified with a novel secure batch data-assisted IOTA scheme, enhancing the efficiency and security of data collection while reducing the computational load on small IoT devices. The proposed system is based on three core components: 1) lightweight cryptography, employing Hash-based Message Authentication Codes (HMAC) and Elliptic Curve Integrated Encryption Scheme (ECIES), chosen for their scalability and suitability for low-power IoT devices; 2) batch data aggregation, introduced in IOTA to augment throughput and reduce latency by processing large data volumes concurrently; and 3) secure and scalable data storage using the recent version of IOTA, namely Chrysalis. Rigorous testing on small IoT devices substantiates the acceptable performance of the system in terms of computation overhead compared to existing systems. By integrating IOTA Chrysalis, hybrid cryptography, and the batch data concept, this method offers an efficient, secure solution for large-scale ocean data collection.]]></description>
      <pubDate>Wed, 15 Jul 2026 16:27:17 GMT</pubDate>
      <guid>https://trid.trb.org/View/2717494</guid>
    </item>
    <item>
      <title>Euclidean distance-based cooperative target encircling of uncrewed surface vehicles in GPS-denied environments</title>
      <link>https://trid.trb.org/View/2699499</link>
      <description><![CDATA[This paper investigates the cooperative target encircling problem for uncrewed surface vehicles (USVs) under a global positioning system (GPS)-denied environment. That is, the positions of the USV and the target are unknown. A distributed cooperative control method with aperiodic communications is proposed using only the Euclidean distance measurements, which incorporates distance keeping and uniformly spaced encircling. First, a fixed-time state observer is designed to reconstruct the distance rate between the USV and the target, allowing the distance keeping loop to be closed without global coordinates. By virtue of distance measurements, a phase angle prediction-based uniformly spaced encircling controller is constructed for velocity coordination, such that all vehicles can encircle the unknown target with the preassigned radius and distribution pattern. Additionally, intermittent communication is realized using dynamic event-triggered mechanisms, where each USV broadcasts the estimated phase angles to the neighboring vehicles based on predefined event conditions. Rigorous theoretical analysis validates the stability of the closed-loop system. Finally, both numerical and hardware-in-loop simulations are provided to illustrate the effectiveness and resilience of the proposed distributed cooperative control method.]]></description>
      <pubDate>Fri, 12 Jun 2026 09:40:51 GMT</pubDate>
      <guid>https://trid.trb.org/View/2699499</guid>
    </item>
    <item>
      <title>Adaptive wind feedforward disturbance rejection controller for an active disturbance rejection based unmanned surface vehicle</title>
      <link>https://trid.trb.org/View/2710031</link>
      <description><![CDATA[In complex wind fields, unmanned surface vehicles (USVs) often experience significant yaw deviations due to insufficient wind-disturbance rejection capability. Active Disturbance Rejection Control (ADRC), as a heading controller, relies mainly on an extended state observer (ESO) to estimate the total disturbance and provide feedback compensation. However, under rapidly varying wind disturbances, the ESO faces a heavier estimation burden, which leads to delayed disturbance compensation in ADRC. To achieve fast and accurate wind-disturbance compensation, this paper proposes an Adaptive Wind Feedforward-Active Disturbance Rejection Controller (AWF-ADRC). First, a wind feedforward model is established and incorporated into the ADRC framework, so that the feedforward compensation directly acts on the yaw control moment for rapid wind-disturbance compensation. Meanwhile, a wind-feedforward ESO is designed to effectively reduce the ESO estimation burden under severe wind variations. Second, a gradient-descent-based adaptive algorithm for the wind feedforward gain is proposed, which uses the gradient descent principle to realize online automatic adjustment of the wind feedforward compensation gain and improve the accuracy of wind-disturbance feedforward compensation. Finally, comparisons with ADRC and PID show that the proposed controller achieves better wind-disturbance rejection and smaller trajectory deviations in complex wind fields.]]></description>
      <pubDate>Wed, 10 Jun 2026 16:38:52 GMT</pubDate>
      <guid>https://trid.trb.org/View/2710031</guid>
    </item>
  </channel>
</rss>