<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>Online multi-modal evacuation during passenger flow outburst in urban transit system: A heterogeneous multi-agent reinforcement learning framework</title>
      <link>https://trid.trb.org/View/2599190</link>
      <description><![CDATA[With growing demand straining urban transit systems’ resilience in managing outburst passenger flows, existing approaches focused on offline and single-modal evacuations remain limited. This study proposes an online multi-modal evacuation framework that coordinates on-duty taxis, buses, and metros while minimizing impact on their regular services. We develop a data-driven agent-based environment to update multi-modal transit data and stranded passenger information in real time. Two coordination strategies are introduced: (1) an independent strategy using a decentralized training and distributed execution algorithm, and (2) a collaborative strategy using a hybrid centralized training and distributed execution algorithm. To dynamically assess evacuation effectiveness, we design a resilience framework with three metrics: robustness, rapidity, and resourcefulness. These metrics are transformed into demand-responsive feedback at each time step, enabling agents to proactively generate resilient evacuation plans. In a real-world case study triggered by a railway disruption, our approach outperforms genetic algorithms and multi-agent deep deterministic policy gradient algorithms in computation time and solution quality under offline conditions. Simulated new environments further validate its online applicability, demonstrating its potential for real-world deployment.]]></description>
      <pubDate>Wed, 17 Jun 2026 16:14:28 GMT</pubDate>
      <guid>https://trid.trb.org/View/2599190</guid>
    </item>
    <item>
      <title>An advanced hybrid approach for emergency healthcare pickup and delivery with unmanned aerial vehicles under a stochastic environment</title>
      <link>https://trid.trb.org/View/2597148</link>
      <description><![CDATA[This paper proposes an advanced hybrid approach for optimizing the pickup and delivery problem using unmanned aerial vehicles (UAVs) in emergency healthcare operations. We specifically account for scenarios involving stochastic demand and flying environments that may arise simultaneously during the distribution of healthcare resources and the collection of biological samples. We address the challenges of emergency logistics, such as inventory shortages, urgent and unpredictable demand, suddenness, and stochastic geographical obstacles. A mixed-integer linear programming model for the healthcare pickup and delivery with UAVs (HPDUP) is first formulated, aiming at maximizing the total weighted coverage from healthcare demands of patient groups. An extended model for HPDUP under stochastic environment (HPDU-SEP) is then developed to manage the uncertainty in demand and traveled distance. An adaptive large neighborhood search (ALNS) integrated with Q-learning for UAV trajectory planning (ALNS-QLTP) is proposed, where Q-learning receives geographical information and feedback distance parameters to the optimization model. Compared with static or semi-dynamic methods, Q-learning achieves higher trajectory optimization efficiency in large-scale uncertain environments by utilizing offline training and scenario updates. ALNS-QLTP exhibits a strong performance on HPDU-SEP instances, guaranteeing 71.70% and 92.47% patient coverage with a limited and sufficient number of UAVs, respectively.]]></description>
      <pubDate>Wed, 17 Jun 2026 16:14:28 GMT</pubDate>
      <guid>https://trid.trb.org/View/2597148</guid>
    </item>
    <item>
      <title>Urban Traffic Flow Simulation in Various Weather Conditions Using Long Short-Term Memory-Based Machine Learning</title>
      <link>https://trid.trb.org/View/2709292</link>
      <description><![CDATA[This study proposes a weather-aware short-term urban traffic prediction framework that combines microscopic traffic simulation with Long Short-Term Memory (LSTM) based Machine Learning (ML) for forecasting vehicle flow at selected urban junctions. The research is motivated by the need to improve traffic-state prediction under varying environmental conditions, since weather disturbances such as rain, fog, and snow significantly affect driverbehaviour, vehicle speed, spacing, and intersection throughput. The simulation environment was developed in the Simulation of Urban Mobility (SUMO) platform using an urban road segment derived from geographic map data and traffic-flow information from the Vilnius city traffic database. 2 monitored junctions were selected as observation points, and traffic behaviour was simulated under 5 weather scenarios: normal conditions, light rain, heavy rain, fog, and snow. Weather-dependent driver parameters, including speed factor, reaction time, acceleration, deceleration, driver imperfection, and minimum gap, were incorporated into the simulation in order to reproduce realistic traffic dynamics. The generated simulation data, including vehicle count, speed, waiting time, temporal features, and encoded weather conditions, were used to train and evaluate multiple LSTM models through an AutoML-based tuning procedure. Among different configurations, the best-performing model consisted of 3 recurrent layers and achieved a validation RMSE of 0.1056 with a validation loss of 0.0652. The results show that the proposed framework is capable of reproducing the general temporal structure of urban traffic flow and preserving the relative ordering of traffic intensity across weather scenarios. Prediction quality was highest under normal conditions, with RMSE of approximately 0.21 veh/min, while the poorest accuracy was observed under snow conditions (about 2.5 veh/min). The model captured dominant traffic trends effectively, although it tended to smooth short-term local oscillations, especially under more unfavourable weather conditions. Overall, the study demonstrates that integrating SUMObased simulation with LSTM forecasting provides an effective and flexible approach for short-term urban traffic prediction under varying meteorological conditions and may support future intelligent traffic management and weather-adaptive mobility applications.]]></description>
      <pubDate>Wed, 17 Jun 2026 16:14:28 GMT</pubDate>
      <guid>https://trid.trb.org/View/2709292</guid>
    </item>
    <item>
      <title>Forecasting short-term subway passenger flow using Wi-Fi data: comparative analysis of advanced time-series methods</title>
      <link>https://trid.trb.org/View/2701205</link>
      <description><![CDATA[Accurately monitoring passenger demand fluctuations is crucial for streamlined operations of subway systems and informed decision-making. This study presents a detailed Time Series Analysis of the Toronto subway system using Wi-Fi data connection from devices as a predictor of passenger volume. Various time series models were tested for short-term forecasting, including Linear Regression, Exponential Smoothing, ARIMA, Random Forest, N-BEATS, and T-GCN. An end-to-end modeling implementation process was carried out, and the performance of each model was evaluated. The primary objective was to assess the effectiveness of short-term prediction models for univariate time series at the system level and discuss deployment challenges. While conventional time series models are fast to implement and interpretable, they require a more in-depth data exploration phase for validation, making scaling at the system level difficult. Additionally, maintenance is more challenging with conventional models, and their exploratory analysis phases need to be repeated when the models degrade over time. Prediction difficulty varied across each subway station, indicating the need for a more thorough calibration or hybrid approach, especially for transfer stations. Despite the different uses and qualities of each model in our scenario, Random Forest and Exponential Smoothing emerged as the best performers and could be a satisfactory option for robust demand forecasting at the system level.]]></description>
      <pubDate>Wed, 17 Jun 2026 16:14:27 GMT</pubDate>
      <guid>https://trid.trb.org/View/2701205</guid>
    </item>
    <item>
      <title>Model-based variable speed limit control on wireless charging lanes: Formulation and algorithm</title>
      <link>https://trid.trb.org/View/2618117</link>
      <description><![CDATA[This paper addresses the variable speed limit (VSL) control problem on wireless charging lanes (WCLs). We first introduce a predictive model to describe the evolution of both traffic flow and the state of charge of electric vehicles, considering the impact of VSL control. The model is formulated as a piecewise affine system through various linearization techniques. Subsequently, we propose a control model that accounts for both traffic and charging efficiencies. By employing a hybrid model predictive control approach, the control problem at each stage is cast as a mixed-integer linear programming (MILP) problem. To expedite the MILP problem, we propose an innovative learning-based algorithm, termed Learning from K-nearest Neighbors mode sequences (LKNMS). The algorithm identifies and eliminates predicted inactive system states (each represented by a binary variable) by leveraging historical solutions of the binary configuration. It thereby significantly reduces the size of the resulting MILP problem. We conduct a series of numerical examples on a 10.7 km WCL to test the proposed control model and algorithm. Our simulation results reveal that VSL control can significantly affect the charging efficiency of WCLs, particularly under light traffic conditions. Moreover, an inherent conflict between traffic efficiency and charging efficiency consistently arises on WCLs. The proposed algorithm significantly reduces the computational time of the MILP problem from 46 % of the 60s control cycle to 5∼12%, without compromising closed-loop performance, which implies strong potential for real-time implementation. We further test the proposed algorithm on a 26.75 km WCL to confirm its robust scalability to large-scale networks.]]></description>
      <pubDate>Wed, 17 Jun 2026 16:14:26 GMT</pubDate>
      <guid>https://trid.trb.org/View/2618117</guid>
    </item>
    <item>
      <title>Priority-driven reinforcement learning for multi-aircraft trajectory optimisation under dynamic weather hazards</title>
      <link>https://trid.trb.org/View/2618116</link>
      <description><![CDATA[Reinforcement Learning (RL) has emerged as a state-of-the-art technique for addressing challenges in air traffic control, and weather hazards and flight procedures can contribute to information biases when applying RL to real-world scenarios. This research focuses on the 3D Multi-Aircraft Trajectory Optimisation (3D-MATO) problem under dynamic weather hazards within the Terminal Manoeuvring Area and addresses the aforementioned concerns. We propose an integrated RL-based algorithm incorporating weather avoidance and quick conflict resolution. Given observed weather radar in Flight Information Regions (FIRs), we introduce the Dynamic Fast Marching Method (DFMM) algorithm to reroute flight paths at smaller time intervals, ensuring safer navigation around hazardous regions. To enhance decision-making quality, we develop a Quickest Priority-based Conflict Resolution (QPCR) strategy, which optimises approach sequences and refines available action choices. The RL agent is trained using a Deep Deterministic Policy Gradient (DDPG) framework, and further enhanced with a self-attention mechanism. A numerical study modelled the real-world approach procedures at Hong Kong International Airport involving varying numbers of approach aircraft under dynamic weather hazards. Results demonstrate the high efficiency and effectiveness of the proposed algorithm under traffic mix and weather conditions, highlighting the contributions of its key strategies and individual components.]]></description>
      <pubDate>Wed, 17 Jun 2026 16:14:26 GMT</pubDate>
      <guid>https://trid.trb.org/View/2618116</guid>
    </item>
    <item>
      <title>Tactical planning in blood supply chain: An integrated demand forecasting and optimization approach</title>
      <link>https://trid.trb.org/View/2614770</link>
      <description><![CDATA[Blood Supply Chain management is crucial to ensure blood flow from donors to patients. At the tactical planning level, it is critical to develop strategies that not only guarantee demand fulfillment but also minimize costs and waste. To address this challenge, this research develops a decision support tool, Blood-TAC, that considers demand predictions and the comprehensive tactical planning of the Blood Supply Chain network. This tool comprises a prediction model and a mixed-integer linear programming model within a rolling horizon framework. First, an XGBoost model is used to predict demand for multiple blood products and locations. These predictions serve as input to the second step, which involves an optimization model that simultaneously manages collection, production, storage, and distribution across a network of blood centers. The goal is to provide decision-makers with key insights for setting mid-term targets and strategies that ensure demand fulfillment while minimizing waste and costs. Blood-TAC is tested using the case study of the Portuguese Blood Supply Chain, demonstrating its ability to generate realistic and improved plans and provide valuable insights to decision-makers. Overall, Blood-TAC reduces costs by 16 % when compared to the network’s historical performance while fulfilling demand and reducing waste of blood products to zero.]]></description>
      <pubDate>Wed, 17 Jun 2026 16:14:26 GMT</pubDate>
      <guid>https://trid.trb.org/View/2614770</guid>
    </item>
    <item>
      <title>Towards sustainable shipping: A learning-aided route-speed joint optimization considering energy efficiency and punctual arrival</title>
      <link>https://trid.trb.org/View/2614764</link>
      <description><![CDATA[As the cost-efficient backbone for global supply chains, maritime transportation facilitates the intensive exchange of remote-manufacturing-site products, resulting in an annual global maritime fuel consumption exceeding 200 million tons and greenhouse gas emissions accounting for 2.89 % of anthropogenic emissions. The joint optimization of ship route and speed, a broadly applicable strategy for energy conservation and emission reduction in advancing sustainable shipping, has garnered considerable attention from both academia and industry. The core challenge in this field involve determining optimal route–speed combinations for a given ship and voyage under complex weather conditions, aiming to maximize energy efficiency while ensuring punctual arrival. Traditional heuristic algorithms struggle to efficiently explore the solution space, particularly in modern maritime optimization problems characterized by multiple competing objectives and high-dimensional decision spaces. Recent research has demonstrated that intelligent learning networks can drive the optimization process by extracting and utilizing latent evolutionary knowledge. However, their generated solutions frequently violate established maritime operational practices when deployed without domain-specific constraint integration, rendering them impractical for real-world ship navigation. Consequently, the energy-saving potential of ship route–speed optimization remains largely limited. To overcome these technical barriers, we propose an improved learning network integrated with tailored constraint mechanisms specific to route–speed settings. Not only does the network learn positive evolutionary patterns to achieve high-efficiency, directed optimization of individuals, but it also maximizes the feasibility of sailing plans for transoceanic shipping. Additionally, a series of improvement strategies, designed to facilitate the generation of a comprehensive Pareto-optimal set, are introduced to form an improved learning-aided optimization framework. In real-world case-based analyses, our proposed framework demonstrates superior performance compared to other state-of-the-art algorithms, achieving an average reduction of 9 % in ship energy consumption. Executable codes is available at https://github.com/Ldiper/MCL-EA.]]></description>
      <pubDate>Wed, 17 Jun 2026 16:14:26 GMT</pubDate>
      <guid>https://trid.trb.org/View/2614764</guid>
    </item>
    <item>
      <title>Data-Driven Pattern Analysis of Attributes Influencing Transportation Construction Program Development</title>
      <link>https://trid.trb.org/View/2613575</link>
      <description><![CDATA[There is a growing consensus that transportation construction projects have an influence beyond the provision of reliable infrastructure and significantly impact socioeconomic advancements. This is reflected in recent global infrastructure development initiatives, which encompass diverse objectives related not only to technical aspects but also nontechnical ones. However, it is challenging to synthesize diverse and sometimes competing objectives in program development. Well-established frameworks or formulas supporting the synthesis and delivery of objectives do not always exist. Ex post empirical analysis of program development outcomes, rather than planning documents, is scarce. In this study, the authors applied an explainable machine learning method, specifically, a zero-inflated negative binomial model with two classification parts, supported by a genetic algorithm, the extreme gradient boosting algorithm, and SHapley Addictive explanation values, on the project inventory supported by discretionary transportation programs of Canadian federal governments over an eight-year period. The project records were mapped to the three categories of community attributes corresponding to the three objectives of the Canadian infrastructure initiative: resilient infrastructure; economic growth; and equity promotion. The results reveal a dichotomous reality: Although transportation equity could be a significant factor in rural areas, the communities where most Canadians live host major transportation construction projects because of factors related to transportation assets and local economics. Bridge improvement is a focal theme, and a vibrant local economy is generally favored. Percentages of senior citizens and children might adjust the level of development. It is anticipated that the findings will provide policy decision-makers, transportation asset owners, and community leaders with a realistic understanding of the realities of discretionary transportation program development, thereby informing evidence-based decision-making in the future.]]></description>
      <pubDate>Wed, 17 Jun 2026 16:14:25 GMT</pubDate>
      <guid>https://trid.trb.org/View/2613575</guid>
    </item>
    <item>
      <title>A Federated Learning and Multi-Agent Reinforcement Learning Framework For Sustainable and Fair Smart Parking Systems in Urban Environments</title>
      <link>https://trid.trb.org/View/2678146</link>
      <description><![CDATA[Urban mobility is increasingly challenging due to the rapid growth of the vehicles, exacerbated congestion, energy dissipation, ecological degradation and environmental concerns. The smart parking system had emerged as a promising solution for this concern; however, existing approaches find it difficult to ensure fairness in space allocation, scalability, and adaptability to the dynamic conditions of urban parking systems. The centralized learning based models had raised the challenging aspect of data privacy, communication overhead and the limited generalization process across the diversified environment. This research work addresses this challenge, by proposing the Federated Learning (FL) and Multi-Agent Reinforcement Learning (FL-MARL) method for the fair and sustainable smart parking management in the urban environment. The proposed framework leverages the federated learning to enable the decentralized training model across the various distributed parking zones, ensuring the data privacy, while the multi-agent reinforcement learning model coordinates the vehicles and parking resources for optimizing the space allocation, fairness and minimal search time. The proposed framework achieves a 96.3% prediction accuracy, a 45% reduction in CO₂ emissions, a 31% improvement in parking search time, and a 28% enhancement in resource utilization compared to existing methods. This study’s novelty lies in the synergistic integration of Federated Learning and Multi-Agent Reinforcement Learning to achieve privacy-preserving, fair, and energy-efficient parking management - addressing scalability, fairness, and data security challenges simultaneously.]]></description>
      <pubDate>Wed, 17 Jun 2026 12:23:22 GMT</pubDate>
      <guid>https://trid.trb.org/View/2678146</guid>
    </item>
    <item>
      <title>KL divergence-guided transfer learning for data-driven shield tunneling under distribution shift</title>
      <link>https://trid.trb.org/View/2676155</link>
      <description><![CDATA[Transfer learning (TL) is a pivotal strategy for developing robust data-driven models in engineering applications, particularly in data-scarce scenarios such as clogging predictions in mechanized shield tunneling. However, significant distribution shifts between source and target domains often compromise transfer performance and it remains unclear under what conditions TL can offer tangible improvements. To address this challenge, three key contributions are presented in this study: (1) Multi-dimensional Kullback-Leibler (KL) divergence is introduced as a novel metric to quantify domain discrepancy at the probability distribution level and explain model transferability; (2) an adaptive fine-tuning approach is developed, by incorporating KL divergence as a regularization term in the loss function; (3) a transferability criterion is proposed to assess potential performance gains in cross-project applications. The results demonstrate the effectiveness of the method for clogging prediction tasks in tunnel engineering, while the proposed framework has the potential to extend other TL tasks.]]></description>
      <pubDate>Wed, 17 Jun 2026 12:23:22 GMT</pubDate>
      <guid>https://trid.trb.org/View/2676155</guid>
    </item>
    <item>
      <title>Evolution of research hotspots and future trend prediction of river and lake sediment solidification based on bibliometrics and machine learning</title>
      <link>https://trid.trb.org/View/2679633</link>
      <description><![CDATA[With the increasing frequency of river and lake dredging, large amounts of sediment are continuously produced, posing significant challenges in resource-efficient, low-carbon disposal. Solidification technology, known for its economic viability, adaptability, and environmental friendliness, has become a key research direction for sediment treatment. Based on publications from the Web of Science, this study combines bibliometric analysis and machine learning trend prediction to explore the development trajectory, research hotspots, and future trends in sediment solidification. Publication volume has steadily increased since 2000, with a notable surge in the past five years. Academic influence has shifted from a few core works to a diversified and mature research system. International collaboration and interdisciplinary integration are growing, with China taking a central role in both publication output and academic influence. Research has evolved from focusing on pollutant migration and environmental safety to addressing mechanical properties and engineering applicability, and more recently, expanding to high-performance solidification materials, micro-mechanism analysis, and intelligent methods. Machine learning models, such as ridge regression and Huber regression, are effective in long-term trend modeling, while ensemble methods capture abrupt changes in trends. Predictions suggest that future research will focus on micro-mechanism analysis, low-carbon material adaptability, multi-pollutant co-solidification, and intelligent formulation optimization. This study provides data-driven insights and methodological guidance for future scientific planning and engineering applications.]]></description>
      <pubDate>Wed, 17 Jun 2026 12:23:21 GMT</pubDate>
      <guid>https://trid.trb.org/View/2679633</guid>
    </item>
    <item>
      <title>Multi-factor analysis and machine learning prediction of low-temperature fracture behavior in steel slag asphalt concrete under thermal-oxidative aging</title>
      <link>https://trid.trb.org/View/2677791</link>
      <description><![CDATA[Replacing natural aggregates with recycled Basic Oxygen Furnace Slag (BOFS) in Steel Slag Asphalt Concrete (SSAC) is a sustainable practice. Nevertheless, studies on the long-term thermo-oxidative aging effect on the low-temperature fracture behavior of SSAC, especially under complex mixed-mode loading, are still lacking. To address this, the present study systematically investigates the low-temperature fracture performance of SSAC under thermo-oxidative aging by integrating multifactor coupled experiments, machine learning prediction, and SHapley Additive exPlanations (SHAP) analysis. The results indicate that thermo-oxidative aging leads to a reduction in the Fracture Toughness (FT) and Fracture Energy (FE) of SSAC; however, the high angularity and high roughness of BOFS mitigate the deterioration process effectively through mechanical interlock. Specifically, under different fracture modes, the FT of Full-proportion BOFS Asphalt Concrete (SAC) is 30.2–44.2% higher than that of conventional asphalt concrete. After 7 days of thermo-oxidative aging, the degradation rate of FT for SAC (12.7–24.1%) is generally lower than that of conventional asphalt concrete (17.2–37.4%). In addition, among the 13 evaluated machine learning models, the Gradient Boosting model exhibits the best prediction performance, with R2 values of 0.932 for FT and 0.979 for FE. SHAP interpretability analysis reveals that fracture mode (SHAP value range ±0.3985) and test temperature (SHAP value range ±6.21) are the most influential factors affecting FT and FE, respectively. These findings are expected to provide valuable insights for the design of sustainable SSAC pavements for cold-climate applications.]]></description>
      <pubDate>Wed, 17 Jun 2026 12:23:21 GMT</pubDate>
      <guid>https://trid.trb.org/View/2677791</guid>
    </item>
    <item>
      <title>TuneNSearch: A hybrid transfer learning and local search approach for solving vehicle routing problems</title>
      <link>https://trid.trb.org/View/2673243</link>
      <description><![CDATA[This paper introduces TuneNSearch, a hybrid transfer learning and local search approach for addressing diverse variants of the vehicle routing problem (VRP). Our method uses reinforcement learning to generate high-quality solutions, which are subsequently refined by an efficient local search procedure. To ensure broad adaptability across VRP variants, TuneNSearch begins with a pre-training phase on the multi-depot VRP (MDVRP), followed by a fine-tuning phase to adapt it to other problem formulations. The learning phase utilizes a Transformer-based architecture enhanced with edge-aware attention, which integrates edge distances directly into the attention mechanism to better capture spatial relationships inherent to routing problems. We show that the pre-trained model generalizes effectively to single-depot variants, achieving performance comparable to models trained specifically on single-depot instances. Simultaneously, it maintains strong performance on multi-depot variants, an ability that models pre-trained solely on single-depot problems lack. For example, on 100-node instances of multi-depot variants, TuneNSearch outperforms a model pre-trained on the CVRP by 44%. In contrast, on 100-node instances of single-depot variants, TuneNSearch performs similar to the CVRP model. To validate the effectiveness of our method, we conduct extensive computational experiments on public benchmark and randomly generated instances. Across multiple CVRPLIB and TSPLIB datasets, TuneNSearch consistently achieves performance deviations of less than 3% from the best-known solutions in literature, compared to 6%–25% for other neural-based models, depending on problem complexity. Overall, our approach demonstrates strong generalization to different problem sizes, instance distributions, and VRP formulations, while maintaining polynomial runtime complexity despite the integration of the local search algorithm.]]></description>
      <pubDate>Wed, 17 Jun 2026 12:23:20 GMT</pubDate>
      <guid>https://trid.trb.org/View/2673243</guid>
    </item>
    <item>
      <title>An integrated robust framework for berth, yard, and shore power planning considering vessel arrival uncertainty</title>
      <link>https://trid.trb.org/View/2699408</link>
      <description><![CDATA[Container terminals must make critical berth and yard allocation decisions in advance of vessel arrivals, as outbound containers require pre-stacking in the yard and shore-power-enabled berths need prior commitment. However, the widespread uncertainty in vessel arrival times (VAT) during this pre-arrival planning phase severely disrupts port operations, complicating integrated planning and impeding decarbonization progress due to low utilization of On-shore Power Supply (OPS). To tackle these issues, this paper introduces a robust optimization framework for the integrated Berth and Yard Allocation problem with OPS (BYAOP) under uncertain VAT. The model adopts a min-max robust approach to hedge against worst-case arrival conditions, explicitly incorporating OPS compatibility and carbon tax costs to minimize total expenses. Given the prohibitive computational complexity of this large-scale problem, we design a novel Adversarial Policy Iteration with Hierarchical multi-agent Soft Actor-Critic (API-H-mSAC) algorithm. This framework features a protagonist agent that makes berth and yard decisions and an adversary agent that generates worst-case VAT scenarios. A hierarchical anchor-point mechanism reduces the action space and improves learning efficiency. Experiments show that our robust pre-arrival framework reduces costs by 4–13% compared to deterministic planning under uncertainty, while API-H-mSAC outperforms benchmarks in solution quality and convergence. This work provides a decision-support tool for robust and low-carbon port operations when commitments must be made ahead of vessel arrivals.]]></description>
      <pubDate>Tue, 16 Jun 2026 11:38:33 GMT</pubDate>
      <guid>https://trid.trb.org/View/2699408</guid>
    </item>
  </channel>
</rss>