Amir Sidiq

Extreme weather conditions can cause significant damage to transport infrastructure, including railways, and disrupt its services. In this context, rainfall-induced landslides have been extensively studied with respect to failure probability; however, predicting the probability of damage to railway infrastructure remains challenging due to inadequate documentation of historical damage records from past extreme events. To address such data gaps, a hybrid data approach is proposed to quantify the physical vulnerability of railway embankments to rainfall-induced landslides, accounting for multiple damage states. In this study, the volume of a landslide is considered an appropriate damage indicator, and cumulative rainfall is used to assess the impact of rainfall-induced landslides on a railway embankment. The proposed integrated method includes maximum likelihood and best-fit regression to derive the fragility function, and Monte Carlo simulation is employed to estimate uncertainty in soil capacity. Data from a real case scenario in the Sri Lankan railways is used to demonstrate the application of the proposed method. Key parameters influencing rainfall-induced slope failure are identified to predict the fragility of rainfall-induced landslides in natural slope railway embankments using cumulative rainfall, with particular reference to small-scale events where landslide volume is less than 1000 m ³ . The application can be extended to assess the vulnerability of slope structures and earthworks and, ultimately, to support the development of resilient transportation infrastructure systems, provided that standardized data documentation and appropriate local data representation are adopted.

This study examines the implementation and application of Artificial Intelligence (AI) methodologies for estimating the Remaining Useful Life (RUL) of civil infrastructure assets, with the aim of supporting more effective civil infrastructure maintenance and management practices. A total of 90 publications were reviewed. Although not all were directly related to civil infrastructure RUL, the overall body of work reveals a continuous research activity since 2014, reflecting growing interest in data-driven deterioration forecasting. A key motivation for this review is to identify AI approaches that have been successfully applied to structured datasets and that demonstrate potential for practical integration into civil engineering asset-management environments. While advanced AI techniques exist, their adoption in civil infrastructure engineering and maintenance management remains limited, and many real-world systems require methods that balance predictive capability with interpretability, robustness, and compatibility with existing workflows. This study discusses a range of AI approaches—including Deep Learning (DL), Machine Learning (ML) ensemble regression, and hybrid models—highlighting their ability to capture complex degradation processes and their potential to enhance durability predictions. Challenges such as data quality, generalisability and interpretability of AI models, and the difficulty of embedding advanced analytics into current maintenance systems are identified. Opportunities for future research include improving model stability against noise, leveraging diverse data sources, addressing class imbalance, quantifying predictive uncertainty, exploring alternative degradation models, and integrating maintenance actions within RUL prediction frameworks. Overall, the findings underscore the increasing role of AI in asset-life prediction and highlight the need for approaches that remain technically sound while being feasible for implementation in civil real infrastructure-management settings.

The concept of structural health monitoring (SHM) of civil infrastructure has evolved rapidly with the incorporation of artificial intelligence (AI). Although available literature mainly focuses on supervised and unsupervised learning to detect and classify damages, both are inherently limited in addressing sequential decision‐making, optimization of long‐term maintenance, adaptive sensing, and dynamic uncertainty handling. In comparison, reinforcement learning (RL) offers a policy‐based framework that can learn optimal actions by interacting with changing structural and environmental circumstances. Although there is an increasing use of RL applications in SHM, there has not been a comprehensive and critical review of RL applications. This review paper critically evaluates RL approaches used in civil infrastructure SHM, which spans across sensor network optimization, energy management of internet of things (IoT)–based wireless sensor networks (WSNs), updating finite element models (FEMs), contact and noncontact damage detection, unmanned vehicle‐based inspection, predictive maintenance planning, autonomous SHM and structural control, etc. The review traces the methodological development from tabular Q‐learning to complex deep reinforcement learning (DRL), multiagent systems, hierarchical designs, and combined physics‐informed systems. In each of these areas, RL shows a high potential in adaptive resource scheduling, enhanced damage localization, large‐scale network maintenance planning, integration of digital twins, federated learning, edge intelligence, and smart infrastructure ecosystems. Nevertheless, a few inherent challenges that prevent large‐scale applications, such as high‐dimensional state‐action spaces, complexity in reward engineering, sample inefficiency, lack of real‐world validation, and lack of transferability to heterogeneous infrastructure types, are discussed. Most of the existing implementations are simulation‐based or applied to lab‐scale models and lack standardized validation procedures for field deployment. Future research should focus on physics‐informed and risk‐sensitive formulations of RL in safety‐critical systems, as well as on scalable multiagent and hierarchical learning designs, transfer and federated learning to alleviate data scarcity and privacy limitations, uncertainty‐aware safe exploration strategies, and enhanced interchangeability with IoT‐enabled digital twin ecosystems. Through this review, academicians, researchers and industrial experts will get critical insights on RL‐based SHM to transform into an intelligent, resilient, sustainable, and autonomous infrastructure monitoring.

Continuous monitoring is significant to ensure the safe operation of infrastructure systems despite the high costs of traditional methods. The current study presents the development of a real-time digital twin of a laboratory-scaled bridge that can assist in the infrastructure monitoring process. Initially, the bridge model was instrumented with strain gauges, and a script was developed to conduct an inverse structural analysis and subsequently, run a finite element analysis to visualize the overall structural response. Three main loading scenarios were tested, and observations highlighted that the digital twin model emulated the actual structural behavior with a high accuracy. Also, the magnitude and the location of the applied loads on the real structure were correctly identified and a linear elastic behavior was identified in the digital model as expected from the actual structure. Further, the rates of change in the strain values and deflections were also evaluated while discussing the significance of digital twin development.

The structural health monitoring (SHM) of bridge infrastructure has become essential for ensuring safety, serviceability, and long-term functionality amid aging structures and increasing load demands. SHM leverages sensor networks to enable real-time data acquisition, damage detection, and predictive maintenance, offering a more reliable alternative to traditional visual inspection methods. A key challenge in SHM is optimal sensor placement (OSP), which directly impacts monitoring accuracy, cost-efficiency, and overall system performance. This review explores recent advancements in SHM techniques, sensor technologies, and OSP methodologies, with a primary focus on bridge infrastructure. It evaluates sensor configuration strategies based on criteria such as the modal assurance criterion (MAC) and mean square error (MSE) while examining optimisation approaches like the Effective Independence (EI) method, Kinetic Energy Optimisation (KEO), and their advanced variants. Despite these advancements, several research gaps remain. Future studies should focus on scalable OSP strategies for large-scale bridge networks, integrating machine learning (ML) and artificial intelligence (AI) for adaptive sensor deployment. The implementation of digital twin (DT) technology in SHM can enhance predictive maintenance and real-time decision-making, improving long-term infrastructure resilience. Additionally, research on sensor robustness against environmental noise and external disturbances, as well as the integration of edge computing and wireless sensor networks (WSNs) for efficient data transmission, will be critical in advancing SHM applications. This review provides critical insights and recommendations to bridge the gap between theoretical innovations and real-world implementation, ensuring the effective monitoring and maintenance of bridge infrastructure in modern civil engineering.

Over recent decades, the implementation of Artificial Intelligence (AI) across various industrial fields from automation to cybersecurity has been transformative. Whilst the implementations of linking AI and data sciences remain complex and thus limited, they both aim to harness data for actionable insights and future predictions. A research focal point in the application of AI in maintenance is crucial for the sustainability and efficiency of assets. Typically, in the civil infrastructure, there are significant benefits to be gained from AI-driven applications. This study reviews the implementation of the AI in bridge maintenance decision-making by conducting a review of literature on major works undertaken by researchers and analysing 102 scientific articles published from 2010 to 2023. Our literature review revealed an emerging trend in recent studies, focusing on the exploration of defect prognosis in bridge maintenance. However, upon further analysis, it becomes evident that there is a notable gap in the existing literature, in the studies related to performance-based prognostic maintenance strategies for bridges. This gap presents an opportunity for future research, one that could yield valuable insights in the field of bridge maintenance and asset management. The review also reveals the focus of the existing literature on defect identification by using the bridge imagery processing. While the AI’s potential in damage detection using bridge imagery is evident, challenges persist including the computational processing and data availability. This review of the literature includes a comprehensive overview of the current implementation of AI in bridge maintenance, highlighting limitations, challenges, and prospective directions.

The concept of digital twins (DT)s enhances traditional structural health monitoring (SHM) by integrating real-time data with digital models for predictive maintenance and decision-making whilst combined with finite element modelling (FEM). However, the computational demand of FE modelling necessitates surrogate models for real-time performance, alongside the requirement of inverse structural analysis to infer overall behaviour via the measured structural response of a structure. A FEM-based machine learning (ML) model is an ideal option in this context, as it can be trained to perform those calculations instantly based on FE-based training data. However, the performance of the surrogate model depends on the ML model architecture. In this light, the current study investigates three distinct ML models to surrogate FE modelling for DTs. It was identified that all models demonstrated a strong performance, with the tree-based models outperforming the performance of the neural network (NN) model. The highest accuracy of the surrogate model was identified in the random forest (RF) model with an error of 0.000350, whilst the lowest inference time was observed with the trained XGBoost algorithm, which was at approximately 1 millisecond. By leveraging the capabilities of ML, FEM, and DTs, this study presents an ideal solution for implementing real-time DTs to advance the functionalities of current SHM systems.

Civil infrastructure assets’ contribution to countries’ economic growth is significantly increasing due to the rapid population growth and demands for public services. These civil infrastructures, including roads, bridges, railways, tunnels, dams, residential complexes, and commercial buildings, experience significant deterioration from the surrounding harsh environment. Traditional methods of visual inspection and non-destructive tests are generally undertaken to monitor and evaluate the structural health of the infrastructure. However, these methods lack reliability due to the need for instrumentation calibration and reliance on subjective visual judgments. Digital twin (DT) technology digitally replicates existing infrastructure, offering significant potential for real-time intelligent monitoring and assessment of structural health. This study reviews the existing applications of DTs across various sectors. It proposes an approach for developing DT applications in civil infrastructure, including using the Internet of Things, data acquisition, and modelling, together with the platform requirements and challenges that may be confronted during DT development. This comprehensive review is a state-of-the-art review of advancements and challenges in DT technology for intelligent monitoring and maintenance of civil infrastructure.

The concept of digital twins (DT)s enhances the traditional structural health monitoring (SHM) by integrating real-time data with digital models for predictive maintenance and decision-making whilst combined with finite element modelling (FEM). However, the computational demand of FE modelling necessitates surrogate models for real-time performance, alongside the requirement of inverse structural analysis to infer overall behaviour via measured structural response of a structure. A FEM-based machine learning (ML) model is an ideal option in this context as it can be trained to perform those calculations instantly based on FE-based training data. However, the performance of the surrogate model depends on the ML model architecture. In this light, current study investigates three distinct ML models to surrogate FE modelling for DTs. It was identified that all models demonstrated a strong performance, with the tree-based models outperforming the performance of the neural network (NN) model. The highest accuracy of the surrogate model was identified in the random forest (RF) model with an error of 0.000350 whilst the lowest inference time was observed with the trained XGBoost algorithm which was at approximately 1 millisecond. By leveraging the capabilities of ML, FEM and DTs, this study presents an ideal solution for implementing real-time DTs to advance the functionalities of current SHM systems.