develop digital twins

To address the loss of fidelity in digital twins caused by dynamic physical system evolution—such as maintenance, wear, and human intervention—this paper proposes a model-verification-based continuous validation framework. The framework integrates real-time monitoring with historical data comparison to construct an interpretable validation metric system, incorporates a lightweight anomaly detection mechanism, and introduces a data-driven parameter self-adaptation estimation algorithm for online twin diagnosis and closed-loop model updating. Unlike conventional static calibration methods, our approach enables long-term trustworthiness preservation and autonomous evolution of the digital twin. Evaluated on an industrial quay crane use case, the framework accurately detects system deviations and dynamically refines model parameters, reducing modeling error by 37.2% and improving maintenance response timeliness by 52%. These results demonstrate significant enhancements in the representativeness, robustness, and engineering practicality of digital twins.

Digital twins (DTs) suffer from conceptual ambiguity and poor cross-domain reusability due to the absence of a unified definition and standardized reference architecture (RA). To address this, we propose TwinArch—the first domain-agnostic, multi-perspective DT reference architecture. TwinArch innovatively decouples structural and dynamic behavioral views to eliminate semantic ambiguity inherent in monolithic architectures. It adopts the “Views and Beyond” methodology for rigorous architectural modeling and is rigorously validated through three iterative design science research cycles: a systematic literature review, industrial practice feedback, and expert validation involving 20 domain specialists. TwinArch supports DT system design, development, and documentation of legacy systems; its architecture website and reproducibility package are publicly open-sourced. Expert evaluation confirms its completeness and engineering applicability, establishing TwinArch as a plug-and-play architectural foundation for smart manufacturing, smart cities, and other DT-intensive domains.

Digital twins are increasingly critical for real-time monitoring and decision-making in complex systems; however, their dynamic behavior, integration of multi-source heterogeneous data, and requirements for real-time synchronization pose significant challenges to verifying accuracy, reliability, and trustworthiness. To address these challenges, this paper proposes the first comprehensive, lifecycle-oriented TEVV (Testing, Evaluation, Verification, and Validation) framework for digital twins. The framework systematically integrates model-driven engineering, formal verification, simulation-based comparison, data provenance tracking, and uncertainty quantification, augmented with real-time monitoring and feedback mechanisms to enable multi-level, multi-dimensional trust assessment. Designed for cross-domain scalability, the framework has been empirically validated across multiple industrial applications, demonstrating substantial improvements in digital twin model credibility and decision-support effectiveness.

Subsymbolic AI struggles with training under few-shot and low-quality data, while existing virtual simulation approaches lack systematic, standards-aligned frameworks. Method: We conduct a systematic literature review covering 22 state-of-the-art works and propose, for the first time, a unified reference framework for digital twin–driven AI simulation—deeply integrating digital twins with AI agents to establish a closed-loop, cyber-physical data orchestration mechanism. We further achieve systematic alignment of this framework with the ISO 23247 international standard for digital twins. Contribution/Results: We distill key technological evolution trends, identify five core challenges and several open research directions, and deliver a reusable architectural guideline and reference framework. This work provides a standardized, methodology-driven foundation for high-fidelity AI simulation, advancing both theoretical rigor and practical deployability in industrial AI applications.

This study addresses core challenges in integrating Digital Twins (DTs) into Systems of Systems (SoS): model uncertainty, combinatorial complexity, and the absence of rigorous verification and validation (V&V) frameworks. Conducting a systematic literature review (2022–2024), we synthesize semi-formal modeling, simulation-based analysis, and formal verification techniques to establish a structured DT V&V taxonomy. Our analysis reveals that existing composition mechanisms lack formal rigor; V&V remains heavily reliant on simulation and empirical approaches; and cross-layer consistency guarantees and standardized frameworks are notably absent. Consequently, we identify an urgent need for scalable, formal composition methods that ensure cyber-physical consistency across SoS boundaries, alongside a unified, SoS-aware V&V standardization framework. This work provides a methodological foundation for trustworthy DT deployment in complex, heterogeneous SoS environments.

This study addresses the challenges of non-standardized, time-consuming, and resource-intensive digital twin development by proposing a tool-supported framework that, for the first time, enables the automatic derivation of purpose-specific digital twins from existing engineering models. By reusing pre-existing structural and behavioral models of physical assets and integrating lightweight customization and configuration, the approach leverages model-based engineering methods and automated generation techniques to rapidly align digital twins with operational objectives. The framework’s feasibility, generality, and engineering practicality are demonstrated through the successful automatic generation of digital twin instances across four heterogeneous use cases.

This work addresses the challenge that existing cross-domain digital twin approaches struggle to effectively coordinate the intrinsic operational relationships among heterogeneous domains in terms of states, errors, objectives, constraints, and control. To overcome this limitation, the paper proposes a novel cross-domain digital twin framework featuring an original seven-layer conceptual architecture and a cross-domain orchestration core. The framework enables deep multi-domain coordination through shared state alignment, explicit coupling modeling, heterogeneous temporal coordination, joint decision-making, and a feedback-driven adaptive mechanism. It incorporates a single offline training phase with bounded online adaptation and integrates model lifecycle management, runtime safety, and provenance tracking. Compatible with mainstream digital twin and simulation standards, the approach is accompanied by validation criteria, a maturity model, and a deployment architecture, thereby establishing a foundation for benchmarking and real-world implementation.

This study addresses the challenge of effectively validating input model specifications in digital twin simulations, where conventional approaches—relying solely on marginal output distributions—often fail to detect misspecified joint input models. To overcome this limitation, the authors propose a novel statistical validation framework based on sub-trajectory conditioning. By repeatedly restarting simulations from observed system states while conditioning on subsets of random inputs, the method constructs conditional output distributions that enable goodness-of-fit testing of the full joint input model. This approach innovatively transcends the constraints of marginal validation and is complemented by diagnostic tools to pinpoint specific input sources responsible for detected discrepancies. Empirical evaluations on M/M/1 and tandem queueing systems demonstrate the framework’s heightened sensitivity and effectiveness, successfully identifying input model misspecifications that traditional methods overlook.