Scholarlink

<p>In large-scale robotic assembly of flexible structures, timely stress feedback serves as the critical prerequisite for ensuring structural integrity and operational safety. However, the prohibitive computational cost of High-Fidelity Finite Element Analysis (FEA) typically induces severe simulation latency, forcing critical safety monitoring to be absent from the cycle-to-cycle supervisory loop. To resolve this critical fidelity-latency conflict, this paper proposes an integrated Digital Twin system that integrates real-time sensor streams into the physics loop for cycle-to-cycle supervisory decision-making. Unlike passive simulation tools, this system enables online estimation of unobservable full-field states. First, to bridge the temporal gap, an asynchronous event-driven architecture is constructed using ROS 2, establishing a non-blocking data pipeline from physical sensors to the digital solver. Second, to enable the system within the industrial cycle time, a physics-aware discretization kernel (Mesh Scale Ratio strategy) is implemented. Experimental validation on a complex five-component aircraft assembly sequence demonstrates that the proposed method achieves a 5.86 × speedup over the monolithic baseline, compressing the end-to-end computation time to approximately 10.7&#xa0;s while retaining prediction accuracy within 3.3% of the high-fidelity reference. This advancement not only aligns simulation speed with the industrial assembly cycle but also restores the cycle-to-cycle supervisory safety feedback loop, shifting the assembly paradigm from empirical trial-and-error to physics-based, quantitative safety decision-making.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Quasi-real-time digital twin for robotic assembly: a physics-based multi-fidelity framework

  • 期刊论文

  • 发表日期：

    2026年6月2日

查看全文
  • Kaixiang Yao,
  • Hao Chi,
  • Dingye Chen,
  • Jiwen Zhang

摘要

In large-scale robotic assembly of flexible structures, timely stress feedback serves as the critical prerequisite for ensuring structural integrity and operational safety. However, the prohibitive computational cost of High-Fidelity Finite Element Analysis (FEA) typically induces severe simulation latency, forcing critical safety monitoring to be absent from the cycle-to-cycle supervisory loop. To resolve this critical fidelity-latency conflict, this paper proposes an integrated Digital Twin system that integrates real-time sensor streams into the physics loop for cycle-to-cycle supervisory decision-making. Unlike passive simulation tools, this system enables online estimation of unobservable full-field states. First, to bridge the temporal gap, an asynchronous event-driven architecture is constructed using ROS 2, establishing a non-blocking data pipeline from physical sensors to the digital solver. Second, to enable the system within the industrial cycle time, a physics-aware discretization kernel (Mesh Scale Ratio strategy) is implemented. Experimental validation on a complex five-component aircraft assembly sequence demonstrates that the proposed method achieves a 5.86 × speedup over the monolithic baseline, compressing the end-to-end computation time to approximately 10.7 s while retaining prediction accuracy within 3.3% of the high-fidelity reference. This advancement not only aligns simulation speed with the industrial assembly cycle but also restores the cycle-to-cycle supervisory safety feedback loop, shifting the assembly paradigm from empirical trial-and-error to physics-based, quantitative safety decision-making.