Scholarlink

<p>This study presents a novel modeling framework that integrates artificial neural networks and fuzzy logic systems to predict the maximum post-buckling loads of polymer fiber-reinforced composite tubes manufactured via the filament winding method. The research addresses the challenge of accurately modeling the complex nonlinear mechanical behavior of composite materials, particularly under post-buckling conditions. Four key parameters, reinforcement type, winding angle, wall thickness, and tube length, were evaluated using multilayer perceptron (MLP), learning vector quantization (LVQ), and Mamdani fuzzy inference techniques. The MLP model achieved a prediction accuracy of 99.9%, while the LVQ and SVM classifiers reached 100% classification accuracy. The Mamdani fuzzy model yielded an average absolute percentage error of 5.5%, closely aligning with experimental values. These results confirm the efficacy of hybrid computational approaches in mechanical load prediction. Although promising, the models were trained on a relatively limited dataset constrained to specific geometries and material configurations. Therefore, their broader applicability to other composite systems or manufacturing methods remains to be tested. Furthermore, the high classification accuracy observed, particularly in the LVQ and SVM models, may indicate a degree of overfitting, highlighting the need for validation using larger and more diverse datasets. Overall, the study contributes to safer and more efficient structural design processes in aerospace, automotive, and civil engineering applications and opens new avenues for research and development in both academic and industrial settings.</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

Optimized neural networks and fuzzy logic for predicting post-buckling loads in polymer composite tubes

  • 期刊论文

  • 发表日期：

    2025年7月19日

查看全文
  • Hayri Yıldırım

摘要

This study presents a novel modeling framework that integrates artificial neural networks and fuzzy logic systems to predict the maximum post-buckling loads of polymer fiber-reinforced composite tubes manufactured via the filament winding method. The research addresses the challenge of accurately modeling the complex nonlinear mechanical behavior of composite materials, particularly under post-buckling conditions. Four key parameters, reinforcement type, winding angle, wall thickness, and tube length, were evaluated using multilayer perceptron (MLP), learning vector quantization (LVQ), and Mamdani fuzzy inference techniques. The MLP model achieved a prediction accuracy of 99.9%, while the LVQ and SVM classifiers reached 100% classification accuracy. The Mamdani fuzzy model yielded an average absolute percentage error of 5.5%, closely aligning with experimental values. These results confirm the efficacy of hybrid computational approaches in mechanical load prediction. Although promising, the models were trained on a relatively limited dataset constrained to specific geometries and material configurations. Therefore, their broader applicability to other composite systems or manufacturing methods remains to be tested. Furthermore, the high classification accuracy observed, particularly in the LVQ and SVM models, may indicate a degree of overfitting, highlighting the need for validation using larger and more diverse datasets. Overall, the study contributes to safer and more efficient structural design processes in aerospace, automotive, and civil engineering applications and opens new avenues for research and development in both academic and industrial settings.