Digital Twin Reliability

Aims & Scope

Digital Twin has been considered as a newly emerging technology that benefits the development of many research areas and disciplines. Driven by requirements from the Physics-of-Failure and machine-learning based reliability design and analysis methods, high precise predictions of the failures under dynamical uncertain conditions become the key concern in the synergistic design of products between their reliability and functional performance. This gives birth to a novel multi-disciplinary research area “Reliability Digital Twin (RDT)”, which should be able to fully utilize the multidimensional data collected from the products, including product model data, statistical data of fault events, real-time operational status data, historically environment and load data, etc., to provide more accurate simulation and reliability predictions, by using the Digital Twin technologies. And the related new ideas and solutions have been emerged all over the world in the recent years. To this end, this session is arranged for presenting these innovative researches from both theoretical and application perspectives to academic and engineering circles.

 

Submissions that reflect the session scope and current state of the field are welcome in areas including but not limited to:

▪ Methodologies and application of combing reliability and digital twin

▪ Development of RDT of products in its design and maintenance stages  

▪ AI and LLM for RDT

▪ Uncertainty analysis in RDT

▪ Advanced application researches of RDT

▪ Intelligent predictive maintenance using RDT

▪ PHM on embodied intelligence devices using RDT

▪ PHM on swarm intelligent systems using RDT


------

Your valuable insights are welcome! 

We cordially invite interested researchers to contact us for details and presentation applications at: secretariat@idea-global.net

Session Chairs

Presentations

  • Assistant Professor
    University of Southampton
    Title: Applications of Ship Digital Twins in Maritime Transport: A Systematic Review
    Abstract Ship digital twins (DT) that are high-fidelity, data-driven virtual replicas of vessels are rapidly emerging as transformative tools in the pursuit of intelligent and sustainable maritime transport. By integrating real-time sensor data with physical and simulation models, they support various activities, including but not limited to predictive maintenance, condition-based monitoring, voyage optimisation, and autonomous navigation. This study provides a systematic review of the academic landscape on ship DT, with core literature retrieved from Web of Science (WoS) and Scopus using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol. CiteSpace was used to perform bibliometric analysis, revealing three major thematic clusters: (1) digital twin architecture and modelling frameworks, (2) intelligent technologies and real-time systems, and (3) practical applications. Furthermore, this paper explores six application scenarios of ship DT. The findings serve as a valuable reference for researchers, developers, and maritime stakeholders aiming to harness the potential of DT for building smarter, safer, and more resilient maritime systems.
  • Assistant Professor
    Universidade de Lisboa
    Title: Bridging Physics-Based Modeling and Data-Driven Analytics: Towards Robust Digital Twins for Marine Structures
    Abstract The structural health management of offshore assets faces significant challenges due to harsh environments and operational uncertainties. In this presentation, the strategic transition from traditional physical modeling to sophisticated digital twin frameworks is explored. Based on extensive research in experimental mechanics and finite element analysis, the role of high-fidelity physical insights as the essential "ground truth" for intelligent systems is emphasized. Recent advancements in DT frameworks for subsea pipelines and floating wind platforms are highlighted, with a specific focus on the integration of multi-physics simulations and real-time sensor data. Key technical challenges, such as data standardization and model fidelity, are addressed from a structural integrity perspective. By demonstrating the synergy between domain-specific engineering expertise and data-driven analytics, a comprehensive roadmap for enhancing the reliability of complex marine structures is established.
  • Doctor
    Macquarie University
    Title: Digital Twins for Safer Decision‑Making: Enhancing Risk Awareness and System Reliability
    Abstract Effective decision‑making in complex systems is closely linked to safety, risk management, and situational awareness. This presentation examines how Digital Twins can support safer decision‑making by enabling real‑time monitoring, predictive analysis, and scenario‑based risk assessment. The talk highlights how virtual representations of physical systems allow decision‑makers to explore “what‑if” scenarios, identify potential safety issues, and evaluate the consequences of alternative actions before they are applied in practice. Drawing on academic and teaching experience, the presentation discusses key challenges such as data reliability, model trust, and human–system interaction. The session aims to provide researchers and practitioners with a structured understanding of how Digital Twins can enhance safety‑critical decisions and contribute to more resilient and reliable systems.
  • Associate Professor
    Hokkaido University
    Title: Sparse Bayesian system identification of nonlinear dynamical systems with hysteresis
    Abstract This work presents a Bayesian system identification method for single degree of freedom (SDOF) systems with nonlinear hysteretic behaviour under earthquake excitation. The method introduces an extension of variational sparse dynamics discovery—known as variational identification of nonlinear dynamics (VINDy)—to earthquake-excited structural systems by incorporating representations of hysteretic restoring forces. Specifically, a variational sparse identification strategy is formulated within an augmented state-space representation to capture memory effects in the restoring force. This approach enables the identification of interpretable probabilistic dynamical models directly from seismic response measurements. The proposed framework is designed to enhance robustness against measurement noise, modelling imperfections, and limited data availability, which are commonly encountered in seismic response analysis of real structures. Through numerical investigations and experimental validation, the effectiveness of the method is demonstrated in accurately identifying hysteretic nonlinear dynamics and quantifying associated uncertainties.
  • Research Fellow
    University of Liverpool
    Title: Physics-Informed Structured Latent Learning for Multimodal Plasma State Inference in Fusion Digitalization: A Tokamak Equilibrium Reconstruction Study
    Abstract Reliable operation of fusion systems in the digital world is essential for practical fusion energy deployment. Unlike digital twins (DTs) for mature industrial equipment, fusion DTs face a major challenge: incomplete physical knowledge, because mechanisms such as failure evolution, load transfer, and latent state propagation remain only partially explicit, observable, and computationally tractable. A key task is therefore to infer physically meaningful plasma states from limited multi-modal diagnostic observations. In tokamaks, equilibrium reconstruction is a representative, as it seeks to recover plasma states from heterogeneous measurements for monitoring, control, and operational reliability. To address this, this paper proposes a reliable AI framework for multi-modal tokamak equilibrium reconstruction. A multi-branch convolutional encoder-decoder is designed to fuse diagnostic signals and infer latent plasma states. In addition, a physics-informed regularization constrains the latent space through derivative consistency with an auxiliary dynamics network, improving reconstruction accuracy, robustness, identifiability, and trustworthiness for reliable fusion DTs.
  • PhD Student
    University of Southampton
    Title: From Bayesian sampling to data-driven inference in model updating: A generative AI perspective
    Abstract Reliable engineering digital models depend on a systematic quantification of the underlying uncertainty, which can be achieved by stochastic model updating, a technique that calibrates uncertain parameters against observed responses. Classical Bayesian solvers such as Markov Chain Monte Carlo remain a gold standard for this task, yet they scale poorly when the inverse problem is high-dimensional or must be repeated across many operating conditions. This contribution reframes stochastic model updating as an amortised inference task, in which a conditional generative model is trained once and thereafter maps measurements directly to posterior samples at negligible cost. Three generative model families are examined within a unified distribution-transport framework. Normalising flows offer rapid sampling together with tractable likelihoods, at the price of constraints imposed by invertible architectures. Diffusion models trade iterative sampling cost for high expressiveness and robustness. Conditional flow-matching learns a time-dependent velocity field that transports a simple reference density onto the target posterior along a prescribed path, enabling efficient generation with standard ODE solvers. The three distinct methods are benchmarked of the NASA and DNV UQ Challenge 2025 for comparison in terms of accuracy, sampling efficiency, and the fidelity of their uncertainty representation. Practical guidance on model selection and posterior validation is provided, alongside the open challenges that remain before these methods can be deployed in safety-critical workflows.