Scholarlink

<p>Delamination is one of the most critical failure mechanisms in composite structures, significantly impacting their mechanical properties and service performance. This study presents a vibration-based non-destructive testing (NDT) approach, combined with finite element analysis (FEA) and machine learning, for detecting and characterizing delaminations in glass fiber-reinforced polymer (GFRP) composites. The methodology utilizes changes in natural frequencies, extracted experimentally and numerically, as input features for a Random Forest regression model to predict the size and location of delaminations. Experimental validation of FEA results showed maximum errors of 4.29% for undamaged plates and 5.37% for delaminated plates. Additionally, the Random Forest model achieved an error rate of less than 14% in most scenarios, highlighting its robustness and practical utility in real-world conditions. While traditional NDT techniques like ultrasonic or thermographic testing are highly effective, the vibration-based approach offers a low-cost, accessible alternative suitable for rapid assessments, particularly in settings where extensive equipment and specialized expertise may not be feasible. The findings highlight the potential of integrating machine learning with vibration analysis for structural health monitoring of composite materials, providing a scalable framework for detecting damage efficiently.</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

Vibration-based delamination evaluation in GFRP composite plates using random forest

  • 期刊论文

  • 发表日期：

    2025年3月6日

查看全文
  • Omer N. Saleh,
  • Alaa Abdulhady Jaber,
  • Rasha Mohammed Hussein

摘要

Delamination is one of the most critical failure mechanisms in composite structures, significantly impacting their mechanical properties and service performance. This study presents a vibration-based non-destructive testing (NDT) approach, combined with finite element analysis (FEA) and machine learning, for detecting and characterizing delaminations in glass fiber-reinforced polymer (GFRP) composites. The methodology utilizes changes in natural frequencies, extracted experimentally and numerically, as input features for a Random Forest regression model to predict the size and location of delaminations. Experimental validation of FEA results showed maximum errors of 4.29% for undamaged plates and 5.37% for delaminated plates. Additionally, the Random Forest model achieved an error rate of less than 14% in most scenarios, highlighting its robustness and practical utility in real-world conditions. While traditional NDT techniques like ultrasonic or thermographic testing are highly effective, the vibration-based approach offers a low-cost, accessible alternative suitable for rapid assessments, particularly in settings where extensive equipment and specialized expertise may not be feasible. The findings highlight the potential of integrating machine learning with vibration analysis for structural health monitoring of composite materials, providing a scalable framework for detecting damage efficiently.