Answers are grounded in this paper and cite their sources.
Demonstrates drift detection and correction methods for prediction models in digital twins, suggesting enhanced reliability in edge computing applications.
Digital twin-based control systems are increasingly playing a crucial role in many safety-critical applications across the automotive and industrial sectors, to enable fail-operational architectures. These systems also play a key role in implementing predictive maintenance strategies, which ensure the availability of the system or plant under control. At the core of a digital twin is the virtual entity – a digital representation of the physical entity. This model serves as the foundation for executing control and prediction decisions. The accuracy of these models directly influences the precision of the control signals generated by the digital twin and applied to the physical entity. However, over time, the model and its parameters may drift due to factors such as component aging or changes in environmental and operational conditions. If these drifts go undetected or are detected too late, they can lead to incorrect predictions and unsafe control actions, particularly in safety-critical digital twin systems. This paper explores methods for detecting such model drifts during runtime and proposes techniques for correcting them without disrupting the normal operation of the system throughout its lifecycle.
No takes yet. Share an insight, caveat, or question.
Krishnanurthy et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: