Scholarlink

<p>For safety in dynamic environmental and load conditions, big dome trusses need real-time structural strength prediction. Traditional Internet of Things (IoT) monitoring systems are delayed or inaccurately responding to structural changes due to centralized processing, redundant sensing, and inadequate physical connection. To address these constraints, the paper proposed a networked, reinforcement learning–driven system for real-time strength estimates and optimal sensor allocations utilizing IoT, civil mechanics, and intelligent analytics. In the recommended architecture, S-PHIST reconstructs dome modal states using graph-spectral processing and physical equilibrium criteria. It powers DiSCo-meta for adaptive, edge-level strength prediction with formal uncertainty bounds. These uncertainty maps enable RIGEL-RL optimize sensor placement and scheduling using reinforcement learning under energy and reliability constraints. Streaming Koopman operator with physics residuals (SKOOP-R) predicts in real time using linear Koopman dynamics and deep residual corrections for sub-40 ms inference sets. In validation-oriented robustness and transfer examination, counterfactual load injections and coverage audits ensure model validity. The integrated system reduces sensors by 35%, strength RMSE by 34%, and bandwidth by 60% over earlier systems. Using a self-validating, physics-consistent IoT framework for civil infrastructure, dome trusses become intelligent, adaptable, and verifiably reliable structures that can learn and operate under uncertainty.</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

An integrated IoT–AI framework for intelligent structural performance prediction and real-time strength forecasting in large-scale dome trusses

  • 期刊论文

  • 发表日期：

    2025年10月31日

查看全文
  • Rashmi Keote,
  • Shilpa Katre,
  • Minal Keote,
  • Sujata Chiwande,
  • Alaka Das,
  • Aseel Smerat,
  • Priti Golar,
  • Princy Diwan,
  • Shailesh Kediya

摘要

For safety in dynamic environmental and load conditions, big dome trusses need real-time structural strength prediction. Traditional Internet of Things (IoT) monitoring systems are delayed or inaccurately responding to structural changes due to centralized processing, redundant sensing, and inadequate physical connection. To address these constraints, the paper proposed a networked, reinforcement learning–driven system for real-time strength estimates and optimal sensor allocations utilizing IoT, civil mechanics, and intelligent analytics. In the recommended architecture, S-PHIST reconstructs dome modal states using graph-spectral processing and physical equilibrium criteria. It powers DiSCo-meta for adaptive, edge-level strength prediction with formal uncertainty bounds. These uncertainty maps enable RIGEL-RL optimize sensor placement and scheduling using reinforcement learning under energy and reliability constraints. Streaming Koopman operator with physics residuals (SKOOP-R) predicts in real time using linear Koopman dynamics and deep residual corrections for sub-40 ms inference sets. In validation-oriented robustness and transfer examination, counterfactual load injections and coverage audits ensure model validity. The integrated system reduces sensors by 35%, strength RMSE by 34%, and bandwidth by 60% over earlier systems. Using a self-validating, physics-consistent IoT framework for civil infrastructure, dome trusses become intelligent, adaptable, and verifiably reliable structures that can learn and operate under uncertainty.