Scholarlink

<p>This work describes and verifies a conceptually novel, experimentally informed, and AI-based design tool for assessing Honeycomb Sandwich Panels (HSPs) retrofits for protecting concrete gravity dams against underwater explosion (UNDEX) pressure. Centrifugal-scale underwater explosion (C-NDEX) experiments were performed to simulate incident and transmitted pressure waves, as well as energy absorption characteristics for a number of HSP configurations. A peak pressure reduction of 35–45% and a maximum principal strain reduction of 30–52% were observed for a number of tested HSP configurations across their design space to confirm their ability for shock pressure mitigation of protected concrete dams. The data generated via C-NDEX experiments were used to train multi-output machine learning models for fast processing of response variables for untested but similar design points. The machine learning models were then coupled with a multi-objective evolution optimization tool for unearthing novel designs of minimized masses with superior pressure mitigation efficiency for a number of HSP retrofits.</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

Integrating machine learning and metaheuristics with centrifugal underwater explosion (INDEX) model tests for evaluating honeycomb sandwich panels in protecting concrete gravity dams

  • 期刊论文

  • 发表日期：

    2025年12月16日

查看全文
  • Tahani Hameedat

摘要

This work describes and verifies a conceptually novel, experimentally informed, and AI-based design tool for assessing Honeycomb Sandwich Panels (HSPs) retrofits for protecting concrete gravity dams against underwater explosion (UNDEX) pressure. Centrifugal-scale underwater explosion (C-NDEX) experiments were performed to simulate incident and transmitted pressure waves, as well as energy absorption characteristics for a number of HSP configurations. A peak pressure reduction of 35–45% and a maximum principal strain reduction of 30–52% were observed for a number of tested HSP configurations across their design space to confirm their ability for shock pressure mitigation of protected concrete dams. The data generated via C-NDEX experiments were used to train multi-output machine learning models for fast processing of response variables for untested but similar design points. The machine learning models were then coupled with a multi-objective evolution optimization tool for unearthing novel designs of minimized masses with superior pressure mitigation efficiency for a number of HSP retrofits.