Answers are grounded in this paper and cite their sources.
Industry 4.0 has driven the integration of physical and virtual systems, transforming simulation from a passive design tool into an active component of machine operation. Digital twins (DT) represent this convergence, enabling bidirectional communication between a virtual entity and its physical counterpart in real time. However, existing DT frameworks for fluid power systems either rely purely on hardware synchronization or on data-driven models that lack physical interpretability. This work presents a five-layer physics-informed hybrid DT framework for a pneumatic actuator that integrates a mixed reality visualization platform, a bidirectional IoT architecture, and a cloud-based data processing layer. The virtual entity combines a lumped quasi-static physics-based model with an extended Kalman filter (EKF) built on the LuGre friction model and a Bayesian ridge regression component. The framework was tested with sinusoidal, ramp, and square-wave position commands at 0.05–0.25 Hz. The live-loop round-trip latency was 150–200 ms. The physics-based model alone produced mean absolute errors (MAE) of 1.0–15.2 mm, largest for square waves at 0.25 Hz. Evaluated on held-out series with one-step-ahead predictions, the hybrid estimator reduced the MAE in eight of nine conditions, with a median reduction of 37% (range 11%–70%). An ablation study shows that the EKF dominates under smooth sinusoidal commands, while the data-driven component dominates under ramp and square-wave commands. The modular five-layer architecture allows each layer to be improved independently, providing a foundation for extension to hydraulic machinery and closed-loop teleoperation.
No takes yet. Share an insight, caveat, or question.
Durango-Cogollo et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: