Original title: A digital twin-based probability-interval hybrid uncertainty quantification method
Wensheng Peng, Siyuan Chen, Kai Xue, Kunlong Li · September 2026
Digital twins are virtual models that mirror real-world systems to predict how they behave. To make trustworthy predictions, they must handle uncertainty. Usually, models treat all unknowns as simple probabilities. However, when data is sparse or experts disagree, assuming precise probabilities can give misleading answers. Researchers created a new framework called PI-DTF to solve this. It splits unknowns into two types: natural randomness gets regular probability distributions, while lack of knowledge gets an interval based on a belief-plausibility structure. A double-loop system calculates the boundaries of possible outcomes by pairing random sampling with interval optimisation. Incoming real-time data continuously updates both sides. When tested on critical nuclear power plant piping, the framework produced failure probability ranges instead of single estimates, offering more robust risk metrics.
You are planning a picnic and check the weather forecast. The forecast says a 30% chance of rain, but you know the data behind it is sparse and unreliable. That single number feels too precise. In this paper, the digital twin gives a range of failure probabilities instead of a single number, just like a weather report that says 'between 10% and 50% chance of rain'—more honest about what is not known.
Engineers rely on digital twins to predict failures in critical systems like nuclear power plants. Traditional methods give a single failure probability, which can be dangerously overconfident when data is scarce. This framework provides a range of possible risks, so decision-makers know the true extent of uncertainty and can make safer choices.
The abstract does not state explicit limitations, but the framework is demonstrated on a single nonlinear analytical case and one engineering application (nuclear piping). Its generalizability to other systems is not established. The nested double-loop strategy may be computationally expensive. The method also requires defining belief-plausibility structures, which may rely on expert judgment.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
A digital twin-based probability-interval hybrid uncertainty quantification method — Wensheng Peng, Siyuan Chen, Kai Xue et al. — Digital Twin — doi:10.1080/27525783.2026.2728792
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper.