Targeted Digital Twin via Flow Map Learning and Its Application to Fluid Dynamics

2025-10-08

AI Summary

Full digital twins (DTs) incur prohibitive computational overhead and struggle with long-term online prediction. Method: This paper proposes a flow-map learning (FML) framework tailored to quantities of interest (QoIs), enabling lightweight target digital twins (tDTs). Using cylinder wake flow as a case study, the tDT is trained offline solely on short-duration QoI time series (e.g., lift/drag forces), without requiring access to or simulation of the full flow field. It integrates memory-enhanced recurrent modeling and nonlinear dimensionality reduction to directly learn a compact dynamical system governing QoI evolution. Contribution/Results: Experimental results demonstrate that the tDT achieves over an order-of-magnitude reduction in hydrodynamic force prediction error and >90% lower computational cost compared to conventional methods, while maintaining superior long-term forecasting accuracy. This work establishes, for the first time, a high-fidelity, long-horizon QoI modeling paradigm independent of full-field simulations—pioneering a new lightweight DT paradigm for complex physical systems.

Technology Category

• Machine Learning: Learning with Manifolds
• Multiagent Systems: Multiagent Learning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We present a numerical framework for constructing a targeted digital twin (tDT) that directly models the dynamics of quantities of interest (QoIs) in a full digital twin (DT). The proposed approach employs memory-based flow map learning (FML) to develop a data-driven model of the QoIs using short bursts of trajectory data generated through repeated executions of the full DT. This renders the construction of the FML-based tDT an entirely offline computational process. During online simulation, the learned tDT can efficiently predict and analyze the long-term dynamics of the QoIs without requiring simulations of the full DT system, thereby achieving substantial computational savings. After introducing the general numerical procedure, we demonstrate the construction and predictive capability of the tDT in a computational fluid dynamics (CFD) example: two-dimensional incompressible flow past a cylinder. The QoIs in this problem are the hydrodynamic forces exerted on the cylinder. The resulting tDTs are compact dynamical systems that evolve these forces without explicit knowledge of the underlying flow field. Numerical results show that the tDTs yield accurate long-term predictions of the forces while entirely bypassing full flow simulations.

Problem

• Modeling dynamics of quantities of interest efficiently
• Learning flow maps from short trajectory data bursts
• Predicting long-term fluid forces without full simulations

Innovation

• Memory-based flow map learning for data-driven modeling
• Offline construction of targeted digital twins from trajectory data
• Compact dynamical systems bypassing full flow simulations