A Hybrid Framework For Managing Uncertainty In Digital Twins Information Processing

Abstract Digital Twins (DT) are used as useful elements for adjusting the physical entities in the manufacturing systems based on digital calculations and simulation results. Despite this approach's evident and proven advantages, DTs are affected by uncertainties from the data sensed, biases from data processing and model abstractions. Adequately managing uncertainty in DT is crucial for the efficient operations of manufacturing systems. This paper proposes a hybrid approach for uncertainty management that identifies, quantifies, and reduces uncertainties in DT information processing by means of AI technologies. The proposed approach integrates a continuous-running module that, during the life cycle of the DT, checks for new uncertainties. It acts as a mechanism for tracking the propagation of uncertainties throughout the model, capturing how they influence and alter the resulting outcomes. Also, the proposed hybrid approach helps identify the DT's risky components and allows the human users to manually intervene to facilitate informed and transparent decisions. A case study on predictive maintenance is used to illustrate and validate the proposal.