When Roads Go Underwater: AI-Enhanced Digital Twin–Driven Flood Resilience for Roadway Serviceability Assessment

Emergency Response Resilience to Floods Operationalised with Applied Geoinformatics

Introduction. The development of science and technology has significantly changed approaches to enterprise organization, product manufacturing, and service provision. Currently, the world is transitioning to a new technological paradigm – Industry 4.0. This involves the use of IIoT systems, digital twins, artificial intelligence, cloud computing, edge computing, and data protection technologies. Integrating these technologies into an enterprise allows for the creation of a system that can handle tasks such as optimizing resource utilization, supporting decision-making, predictive maintenance, and improving product quality. Purpose of the Work. The main goal of this article is to form a theoretical and methodological basis for analyzing the impact of Fourth Industrial Revolution technologies on enterprise automation processes, which is realized through the implementation of decision support systems and predictive maintenance of IoT systems. Specifically, this includes: studying the optimization of resource utilization and decision support in IoT systems; analyzing the main problems in solving these tasks; analyzing existing technologies and IT systems, as well as developing an IT system architecture that would have the capabilities to solve these problems and avoid identified issues. Results. During the analysis of articles and existing market systems, it was found that the following problems exist in the implementation of Industry 4.0 technologies: outdated IT systems, equipment compatibility, cost and resource limitations, lack of personnel skills, change management, and cybersecurity issues. To solve the problems of decision support and resource optimization, key technologies were identified: digital twins, which allow obtaining up-to-date information about devices in IoT systems; spatio-temporal graphs and graph neural networks, configured to work with data that have certain patterns and interconnections; reinforcement learning, which allows training a model while interacting with the environment, enabling the model to adapt to changes in enterprise operations. A microservice architecture for an IT system was developed to solve the aforementioned problems, integrating graph neural networks, reinforcement learning, digital twins, and an IoT system. Some of the main advantages of such an IT system are scalability, modularity, and adaptability to business needs. Conclusions. This study demonstrates the relevance, trends, and effectiveness of implementing Industry 4.0 technologies in enterprises based on articles and existing software products. Challenges were identified, and key technologies to overcome them were determined. A microservice IT system architecture based on graph neural networks and digital twins was developed, which has prospects for solving resource utilization optimization and decision support tasks. Areas for additional research on the specific implementation of the proposed architecture's modules were identified, such as: dynamic adaptation to the environment with the combination of pre-known business rules; agent training with minimal performance degradation; digital twin model and data cleaning and transformation mechanisms. Keywords: deep machine learning, reinforcement learning, IoT, IIoT, digital twins, Industry 4.0, graph neural networks.

The occurrence of disasters such as extreme flooding in urban environments has severe consequences, not only on the human population but also on critical infrastructures such as the road networks, which are of vital importance for everyday living and particularly for emergency response. In this article, our main goal is to present-conceptually and in praxis-a model that could be used from the emergency responders for timely and efficient emergency management and response in an urban complex environment. For the city of Cologne in Germany, we aim to indicate possible ways to decrease the emergency response time during an extreme flood scenario through the development of an accessibility indicator, which consists of different components. Therefore, we will investigate the opportunities that occur, in a flood risk scenario, from the use of geographic information in different forms such as Volunteered Geographic Information (VGI) and open-source data in an ArcGIS environment, to increase urban resilience through the decreasing emergency response time. We will focus on network analysis for the fire brigades (first acting emergency responders) during a flood scenario to calculate their emergency response ranges and emergency response routes through flooded road networks, for the assistance of the possibly affected hospitals, refugee homes and fire brigades, which can be flooded. At the end of the paper, we suggest that the vulnerable community of the refugees could be taken into consideration as a new source of VGI, as an additional component that would lead to the decrease in the emergency response time. The geo-located information that could be provided by the refugee community can be very useful in emergency situations, such as those examined in this article where timely information can be forwarded to the proper authorities for a more focused and timely emergency response, increasing the resilience of the urban population and their community.

[First paragraph] The 2001 World Trade Center and 2005 Hurricane disasters, and the 2014 Ebola outbreak were major events that tasked the United States’ public health emergency preparedness and response apparatus. The health and economic cost of these events is huge including over 4000 deaths and damages to infrastructure worth hundreds of billions of dollars. Unfortunately, the U.S. labor force was disproportionately affected (1–4). Similar public health impact of other disasters on workers has also been reported. In the immediate aftermath of the Gulf of Mexico oil spill, for instance, approximately 75% of those who developed acute health effects and sought medical care were clean-up workers (5).

Marine and coastal ecosystems (MCEs) are vital to human well-being, playing a significant role in climate regulation, carbon sequestration, while protecting coastal areas from sea level rise and erosion. However, these ecosystems are increasingly threatened by the combined effects of anthropogenic stressors (e.g., pollution) and climate change-related pressures (e.g., rising sea temperatures and ocean acidification).  Cumulative impacts arising from this complex interplay threaten MCEs' ability to deliver critical ecosystem services, compromising their health and resilience.Machine Learning (ML) has emerged as a valuable tool for assessing ecological conditions under multiple pressures. Algorithms like Random Forest (RF) and Support Vector Machine (SVM) have demonstrated their effectiveness in identifying patterns and predicting changes in ecosystem health. However, these models often fail to account for spatial dependencies between data points, which are crucial for understanding the interconnected nature of marine environments. Graph Neural Networks (GNNs), a more recent advancement in ML, overcome this limitation by explicitly modelling spatial relationships, making them highly suitable for analysing complex MCE dynamics.This study explores the application of GNN-based models to assess the impact of multiple pressures on seagrass ecosystems in the Italian coastal areas. To this aim, a comprehensive dataset was constructed, including key variables influencing seagrass health, such as nutrient concentrations, temperature, and salinity, derived from open-source platforms (e.g., Copernicus CMEMS, EMODnet). Data were synthesized into a 4km raster grid, with each pixel representing seagrass presence or absence. GNNs were constructed by considering each pixel as a node and connecting it to neighbouring pixels to capture spatial relationships. Experiments evaluated different GNN architectures, such as Graph Convolutional Networks (GCNs) and Graph Attention Networks (GATs), alongside traditional ML models like RF, SVM, and Multi-Layer Perceptron.The results showed that GNNs outperformed traditional models in terms of F1-score and accuracy, particularly in spatially complex scenarios. Traditional models often misclassified regions with intricate spatial dependencies, such as boundaries between seagrass patches, whereas GNNs demonstrated superior capability in leveraging spatial context. Despite these advantages, the study faced challenges due to the limited availability of high-resolution, temporal datasets, constraining the full exploration of dynamic ecosystem processes. However, by addressing the challenge of spatial resolution in ecological data, GNNs represents a transformative approach to understanding ocean dynamics. Their integration into a Digital Twin of the Ocean has the potential to transform ecosystem management and significantly advance coastal resilience efforts. This framework would enable detailed simulations and predictions of processes like ocean currents, extreme weather events, and the cumulative impacts of climate change and human activities. Moreover, the combination of GNNs and Digital Twins would provide deeper insights into the complex interplay of factors shaping marine and coastal ecosystems ecological state and processes and their resilience overall. This synergy empowers scientists and policymakers with actionable intelligence, fostering effective decision-making and the development of strategies to mitigate ocean hazards, while safeguarding biodiversity and enhancing the resilience of coastal communities. As future efforts move towards incorporating high-resolution data, this integrated approach holds promise for advancing the sustainable management of MCEs globally.

As the Deepwater Horizon disaster unfolds in the Gulf of Mexico, public health practitioners are having a sinking deja vu feeling. Once again, environmental disaster has struck, and tens of thousands of emergency responders—some professionals, but many more volunteers—have swung into action, potentially risking their health as they work to clean up the worst oil spill in U.S. history. Veterans of similar disasters are wondering if historical lessons learned can help keep the damage to a bare minimum. But a paucity of hard data on emergency responder health makes it difficult even to ask the right questions. “Emergency responders have not been adequately studied,” says Gina Solomon, codirector of the occupational and environmental medicine residency and fellowship program at the University of California, San Francisco, and a senior scientist with the Natural Resources Defense Council. “They tend to be ignored.” Professional emergency responders such as firefighters may not put much emphasis on health effects studies. “They are going to do what they need to do regardless of their own safety,” says Don Donahue, executive director of the Center for Health Policy & Preparedness at the Potomac Institute for Policy Studies, who has firefighters in his extended family. “They have to be a little crazy by definition.” That heroic approach is manna for the people they save, but firefighters may pay a price with disorders such as cancer. Studies of cancers in firefighters have had mixed results, but there is evidence linking this occupation with brain, thyroid, esophageal, bladder, testicular, prostate, and cervical cancers, as well as melanoma and Hodgkin disease (Bates1 and Ma et al.2 are two such studies). Health risks tend to be even greater for the many nonprofessional emergency responders who rush to the scene of crises such as oil spills, terrorist attacks, hurricanes, train derailments, and chemical releases. Those workers often don’t have the training and advanced equipment that protect professional emergency responders to some degree. These and many other factors make it a daunting challenge to protect the health of emergency responders during disasters.

In the digital twin system of mechatronics engineering, the scale and accuracy of models are continually improving. Nevertheless, this growth can hinder real-time interaction and decision-making accuracy within digital twins. The resulting delay impacts the entire system’s reliability by reducing its response speed and real-time decision-making. Consequently, there is an imperative demand for a lightweight approach to tackle the challenges arising from the escalating scale of digital twin virtual entity models. This paper presents a digital twin methodology that is lightweight. The procedure comprises three primary phases: graph data modeling, graph neural network analysis, and hierarchical simplification of virtual entities. Specifically, the graph neural network method proposed in this article is used to classify the graph data of virtual entities. Then, the model is hierarchically simplified based on the classification. Finally, experiments were conducted on factory and robotic arm datasets to evaluate the proposed method. The experimental results indicate that the DTL-GNN method can reduce system redundancy while preserving the essential features of virtual entities.

As the complexity and scale of modern networks continue to grow, the need for efficient, secure management, and optimization becomes increasingly vital. Digital twin (DT) technology has emerged as a promising approach to address these challenges by providing a virtual representation of the physical network, enabling analysis, diagnosis, emulation, and control. The emergence of Software-defined network (SDN) has facilitated a holistic view of the network topology, enabling the use of Graph neural network (GNN) as a data-driven technique to solve diverse problems in future networks. This survey explores the intersection of GNNs and Network digital twins (NDTs), providing an overview of their applications, enabling technologies, challenges, and opportunities. We discuss how GNNs and NDTs can be leveraged to improve network performance, optimize routing, enable network slicing, and enhance security in future networks. Additionally, we highlight certain advantages of incorporating GNNs into NDTs and present two case studies. Finally, we address the key challenges and promising directions in the field, aiming to inspire further advancements and foster innovation in GNN-based NDTs for future networks.

Real-time simulation and the integration of digital twins necessitate the development of computationally efficient surrogate models for contemporary food drying systems, in which complex physical processes must interact dynamically with virtual representations. Traditional modeling approaches for food drying are subject to notable limitations: empirical models often lack physical fidelity, whereas high-fidelity computational fluid dynamics (CFD) simulations require the resolution of tightly coupled heat and mass transfer processes among vapor, liquid water, and air phases within unsaturated porous media, rendering them computationally prohibitive for real-time applications. This work proposes a region-aware graph neural network (GNN) surrogate model for food drying simulation by leveraging mesh-based and region-specific representations of the drying process. The proposed GNN architecture, inspired by MeshGraphNet with specialized masked message passing, handles the heterogeneous domains and sharp discontinuities characteristic of coupled heat and mass transfer in food drying. The GNN model was trained on a limited dataset comprising 28 CFD simulations of hot-air drying of a potato slice under varying ambient conditions, generated using a CFD framework validated against numerical results from the literature and laboratory-scale experimental data. Despite this small training set, the model delivers predictive performance for operating conditions beyond those seen during training. Single-step predictions achieved RMSE values of 0.17 °C for temperature ( T ), 0.26% for relative humidity ( RH ), and 0.0028 for liquid saturation ( S w ). Comparative analysis with standard message passing revealed a threefold reduction in interfacial absolute error (0.003 vs. 0.008) and elimination of nonphysical field propagation across domain boundaries. The model maintained performance on the training domain edges, with moisture ratio curves achieving R 2 > 0.97 for 10-step rollouts (R 2 = 0.989 interior domain; R 2 = 0.979 edge domain) while preserving the characteristic S-shaped drying curves in longer autonomous simulations (R 2 = 0.784 and 0.873 for full rollouts of the interior and edge-of-domain test cases, respectively). This framework is applicable to real-time process control, design optimization, and digital twin implementations in food manufacturing. • Integrated a region-aware GNN into a digital twin for potato drying. • Achieved high accuracy with minimal data in heat and mass transfer prediction. • Region-aware message passing ensured physical consistency across domains. • Model adapted to changes in temperature and geometry with low prediction error. • Fast predictions enable real-time control and food process optimization

A graph theoretic approach for quantitatively comparing digital twin designs is explored in this paper, with a novel methodology being offered to evaluate the structural and functional similarities between these complex systems. Graph theory is leveraged to represent digital twins as mathematical graphs in our approach, with nodes represented by function-flow pairs and the interactions between the functions depicted by edges. The application of graph-based similarity measures, such as NetSimile and DeltaCon distance, for quantitatively assessing the resemblance between digital twin designs is allowed by this representation. Our method addresses the limitations of conventional vector-based metrics of design comparison, which focuses on the connectivity and interactions within these systems, offering a more nuanced and comprehensive analysis. The designs of six digital twins were collected, and functional models were created using the Functional Basis. The functional models were then converted to graphs to perform design similarity analysis. At the end of the study, it was concluded that graph similarity measures such as NetSimile and DeltaCon can be used as quantitative tools for comparing digital twins. However, further investigation is needed to test the effectiveness of such measures for complex designs such as digital twins. The limitations inherent in the current methodology are highlighted, such as challenges in capturing dynamic interactions and the evolving nature of digital twins. Finally, graph neural networks (GNN) are proposed for multiple feature-based similarity analysis and large language models (LLMs) to automate functional model generation.

Abstract: This paper presents a novel AI-powered world modeling framework designed to support smart governance through the integration of traffic, infrastructure, and emergency response data. By constructing a unified spatiotemporal graph that continuously learns from real-time sensor inputs, mobility feeds, and incident logs, the system enables predictive simulation and policy optimization for urban decision-makers. Using a combination of graph neural networks, spatiotemporal transformers, and reinforcement learning, the model captures dynamic interactions among critical urban subsystems. Scenario-based simulations—ranging from infrastructure stress to multi-hazard emergencies—are used to evaluate governance strategies under uncertainty. An interactive GIS-based dashboard allows policymakers to test counterfactuals and visualize the cascading effects of interventions. The proposed architecture enhances operational readiness, decision transparency, and systemic resilience, offering a scalable foundation for AI-augmented public sector planning. Keywords: world models, smart governance, AI for public policy, urban resilience, traffic data, infrastructure monitoring, emergency response, spatiotemporal modeling, graph neural networks, digital twin, reinforcement learning, decision support systems, urban informatics, policy simulation

In this work, we introduce a novel method to develop whole system digital twins (DTs) for advanced nuclear reactors. This method treats a complex reactor system as a heterogeneous graph: with the system components as different types of graph nodes and their physical interconnections as edges. Based on the heterogeneous graph, a graph neural network combining graph convolution and temporal node attention is developed as the DT, facilitating a comprehensive understanding of the system’s dynamic behavior. By utilizing the System Analysis Module (SAM) code for simulating various operational transients, we develop a graph-based database that trains the DT. This DT is characterized by two primary functions: It can infer the entire system’s status using sparse node information, and it can predict the progress of transients based on current and historical system information. Our approach is validated through case studies on the Experimental Breeder Reactor II (EBR-II) system and a generic Fluoride-salt-cooled High-temperature Reactor (gFHR), demonstrating the DT’s accuracy in forecasting operational transients. The DT’s rapid computation capabilities enhance its potential for supporting advanced reactor operations, offering benefits in intelligent simulation, autonomous control, and anomaly detection, paving the way for improved safety analysis and intelligent component health management for advanced reactor systems and reducing their operations and maintenance cost.

In an emergency incident time is of the crucial factor, and the basic philosophy of an emergency response agency is to respond as quickly as possible to minimize the loss of life and property damage. Emergency response operations are very important activities in the oil and gas industrial areas. As an incident in an integrated facility at oil and gas industrial area can have considerable major economic and social impact, response to such an incident must be provided in a very short time. The use of parameters plays an important role in the successful implementation of emergency response. This case study considered configuration emergency response time parameters at an integrated facility. Therefore, to predict the response time consideration must be given to the characteristics of emergency response parameters, its effectiveness and efficiency. The decisions for such emergency responses to integrated facility should consider the available emergency resources, and other factors such as emergency responders’ competency, fire protection and fire detection system, and the characteristics of surrounding affected industrial facilities. Pilot Test results showed that the top five most parameters affecting emergency response time at the case study of oil and gas integrated facility were namely location, competency, road access, emergency vehicle and frequent drill/exercise. Other analyses resulted that emergency drills and exercises parameter shall be critically examined. Drills constitute a simultaneous and comprehensive test of emergency plans, staffing levels, personnel training, procedures, facilities, equipment, and materials. Having this process, the company able to improve emergency preparedness and response management and aligned with crisis management protocol by government authority.

Sixth-generation networks propose integrating multiple networks while ensuring seamless network performance. Hence, networks are becoming increasingly complex while the traditional methods to manage networks are facing significant challenges as the topology sizes, traffic patterns, and network domains are changing. Autonomous network management solutions, which are often built on digital twins, are emerging as possible candidates for addressing these challenges. Machine learning models are widely used for realizing digital twins. Among many neural network structures, graph neural networks are a subclass of promising machine learning methods that perform well in graph-structured data such as network topologies. In this paper, we explore GNN performance on real network data and present our solution to per-flow mean delay prediction which achieves a MAPE of 35.39%, improving the baseline solutions by over 20% together with additional findings and further improved models for Graph Neural Networking Challenge 2023.

Loess deposits in China, covering extensive regions, exhibit distinctive physical and mechanical characteristics, including collapsibility and reduced mechanical strength. These properties contribute to heightened susceptibility to slope-related geological hazards, such as landslides and collapses, in these areas. The widespread distribution and challenging prevention of these geological disasters have emerged as significant impediments to both public safety and economic development in China. Moreover, geological disaster data originates from diverse sources and exists in substantial fragmented, decentralized, and unstructured formats, including textual records and graphical representations. These datasets exhibit complex structures and heterogeneous formats yet suffer from inadequate organization and storage due to the absence of unified descriptive standards. The lack of systematic categorization and standardized representation significantly hinders effective data integration and knowledge extraction across different sources. To address these challenges, this study proposes a novel loess slope stability assessment method employing a dendrogram neural network (GNN-TreeNet) integrated with knowledge graph technology. The methodology progresses through three phases: (1) construction of a multi-domain knowledge graph integrating a large number of loess slopes with historical disaster records, instability factor relationships, and empirical parameter correlations; (2) generation of expressive node embeddings capturing inherent connections via graph neural networks; (3) development and training of the GNN-TreeNet architecture that leverages the graph’s enhanced representation capacity for stability evaluation. This structured framework enables cross-disciplinary data synthesis and interpretable slope stability analysis through a systematic integration of geological, geographical, and empirical knowledge components.