Scholarlink

<p>The present study deals with the experimental and machine learning assisted investigations on flexural behaviour hybrid jute/hemp fibre reinforced epoxy composite developed through hand lay-up technique. To understand the notch sensitivity, Single-edge notch bending (SENB) test was conducted and differences in the obtained results were reported. Flexural testing was carried out according to ASTM D790 (3-point bending) and ASTM D5045 (Single Edge Notch Bending—SENB). Flexural strength of the composite was between 112 and 114&#xa0;MPa and Young’s moduli of the composite was between 5.6 and 7.2&#xa0;GPa for both regular and SENB samples respectively. The fractured surface was investigated through Scanning Electron Microscopy (SEM) in order to confirm good fibre dispersion and interfacial bonding, and also, Energy Dispersive X-ray Spectroscopy (EDS) and elemental mapping confirm the organic-dominant nature of the fibre with uniform distribution of carbon and oxygen, and the presence of small amounts of minerals that are inherent in natural fibres. In addition to the experimental investigations, machine learning (ML)-based predictive framework was established to model the flexural behaviour of the developed hybrid composites. The trained models include random forest (RF), XG-Boost, and support vector regression (SVR) models using the experimental load–deformation data so as to achieve better predictive capability. The Random Forest model had the lowest prediction error and shows stable generalisation behaviour with the highest R<sup>2</sup> value (0.809) and cross-validation (CV) R<sup>2</sup> value (0.828 ± 0.014) among all the models. The ensemble and kernel-based machine learning frameworks are shown to be effective in accurately predicting mechanical performance of hybrid jute/hemp composite, which can be applied in the sustainable semi-structural applications.</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

Predictive Modelling and Experimental Investigation of Flexural Properties of Hybrid Natural Fiber Reinforced Epoxy Composites

  • 期刊论文

  • 发表日期：

    2026年7月26日

查看全文
  • G. M. Manjunatha,
  • K. S. Lokesh,
  • M. Santhosha,
  • Swapnil Sunil Pai,
  • B. R. Vatsala,
  • Raghavendra Pai

摘要

The present study deals with the experimental and machine learning assisted investigations on flexural behaviour hybrid jute/hemp fibre reinforced epoxy composite developed through hand lay-up technique. To understand the notch sensitivity, Single-edge notch bending (SENB) test was conducted and differences in the obtained results were reported. Flexural testing was carried out according to ASTM D790 (3-point bending) and ASTM D5045 (Single Edge Notch Bending—SENB). Flexural strength of the composite was between 112 and 114 MPa and Young’s moduli of the composite was between 5.6 and 7.2 GPa for both regular and SENB samples respectively. The fractured surface was investigated through Scanning Electron Microscopy (SEM) in order to confirm good fibre dispersion and interfacial bonding, and also, Energy Dispersive X-ray Spectroscopy (EDS) and elemental mapping confirm the organic-dominant nature of the fibre with uniform distribution of carbon and oxygen, and the presence of small amounts of minerals that are inherent in natural fibres. In addition to the experimental investigations, machine learning (ML)-based predictive framework was established to model the flexural behaviour of the developed hybrid composites. The trained models include random forest (RF), XG-Boost, and support vector regression (SVR) models using the experimental load–deformation data so as to achieve better predictive capability. The Random Forest model had the lowest prediction error and shows stable generalisation behaviour with the highest R2 value (0.809) and cross-validation (CV) R2 value (0.828 ± 0.014) among all the models. The ensemble and kernel-based machine learning frameworks are shown to be effective in accurately predicting mechanical performance of hybrid jute/hemp composite, which can be applied in the sustainable semi-structural applications.