Scholarlink

<p>Hybrid fiber metal laminates (FMLs) containing open holes are promising for aerospace structures owing to their high stiffness-to-weight ratios and superior damage tolerance. However, their design remains challenging because of complex interactions among metal–fiber ratios, fiber species, and geometric discontinuities. In this investigation, twenty-seven configurations were generated using Taguchi’s L27 orthogonal array, combining Ti6Al4V with carbon, glass, and Kevlar–epoxy systems in a 2/1 stacking sequence. The optimization objectives involved simultaneous minimization of total deformation and von Mises stress and maximization of strain energy. A stacked ensemble surrogate integrating random forest, gradient boosting, and extra-tree regressors with a ridge meta-learner achieved high predictive accuracy (R<sup>2</sup> = 0.8873 for deformation, 0.8928 for stress, 0.8282 for strain energy). Embedding this surrogate in a multi-objective grey wolf optimization (MOGWO) framework produced a well-distributed Pareto front with a hypervolume of 0.917 and a generational distance of 0.072. The knee-point solution, corresponding to a metal-dominant Ti/carbon laminate with a circular cut-out, exhibited a deformation of 0.294&#xa0;mm, stress of 328.3&#xa0;MPa, and strain energy of 11.98&#xa0;mJ. Relative to the least favorable configurations, these correspond to approximately 34 percent lower deformation, 43 percent lower stress, and 10 percent higher strain energy. Experimental validation of the optimized configuration showed deviations below 5 percent from predictions. The scanning electron microscopy confirmed matrix shear, fiber pull-out, and delamination consistent with simulated patterns. The integrated FE–ML–MOGWO framework provides a validated, data-driven route for stiffness–toughness optimization of open-cutout FMLs in aerospace applications.</p> Graphical Abstract <p></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

Finite-element, ensemble learning and grey-wolf optimization of open-cutout titanium fiber metal laminates

  • 期刊论文

  • 发表日期：

    2026年2月28日

查看全文
  • P Hariharasakthisudhan,
  • Mohamed Fazil Kaja Mydeen,
  • Sridhar raj Subramaniam,
  • Vishnu Chitravel,
  • K Logesh

摘要

Hybrid fiber metal laminates (FMLs) containing open holes are promising for aerospace structures owing to their high stiffness-to-weight ratios and superior damage tolerance. However, their design remains challenging because of complex interactions among metal–fiber ratios, fiber species, and geometric discontinuities. In this investigation, twenty-seven configurations were generated using Taguchi’s L27 orthogonal array, combining Ti6Al4V with carbon, glass, and Kevlar–epoxy systems in a 2/1 stacking sequence. The optimization objectives involved simultaneous minimization of total deformation and von Mises stress and maximization of strain energy. A stacked ensemble surrogate integrating random forest, gradient boosting, and extra-tree regressors with a ridge meta-learner achieved high predictive accuracy (R2 = 0.8873 for deformation, 0.8928 for stress, 0.8282 for strain energy). Embedding this surrogate in a multi-objective grey wolf optimization (MOGWO) framework produced a well-distributed Pareto front with a hypervolume of 0.917 and a generational distance of 0.072. The knee-point solution, corresponding to a metal-dominant Ti/carbon laminate with a circular cut-out, exhibited a deformation of 0.294 mm, stress of 328.3 MPa, and strain energy of 11.98 mJ. Relative to the least favorable configurations, these correspond to approximately 34 percent lower deformation, 43 percent lower stress, and 10 percent higher strain energy. Experimental validation of the optimized configuration showed deviations below 5 percent from predictions. The scanning electron microscopy confirmed matrix shear, fiber pull-out, and delamination consistent with simulated patterns. The integrated FE–ML–MOGWO framework provides a validated, data-driven route for stiffness–toughness optimization of open-cutout FMLs in aerospace applications.

Graphical Abstract