Scholarlink

<p>Glass and carbon fiber composite materials are extensively utilized in wind turbine blades due to their superior mechanical properties. This study presents a novel expert system for crack diagnosis and prognosis in glass/carbon fiber composite structures, aiming to enhance the safety and reliability of wind turbine blades. The proposed approach integrates experimental and numerical modal analysis to determine the first three natural frequencies of glass and carbon fiber composite beams and plates (GCCBP) under fixed free boundary conditions, considering various crack configurations. The average error between experimental and numerical modal analyses is found to be 8.64%. Additionally, experimental and numerical fatigue analyses are conducted to estimate fatigue stress and remaining useful life (RUL) for cracked GCCBP under the same boundary conditions, with a deviation of only 4.31% between experimental and numerical results. A comprehensive dataset, comprising crack length, crack depth, the first three natural frequencies, fatigue stress, and RUL derived from numerical analyses, is used to train a Gaussian process regression based expert system developed in Python. The system achieves an R-squared score of 96.0, indicating high predictive accuracy. Validation with unseen experimental data confirms the system effectiveness, demonstrating an average error of 3.82% in crack location identification, severity assessment, and RUL estimation. This work highlights the potential of machine learning for advancing structural health monitoring in composite materials.</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

Machine Learning-Enabled Crack Diagnosis and Prognosis in Glass/Carbon Fiber Composite Structures

  • 期刊论文

  • 发表日期：

    2025年3月25日

查看全文
  • S. Rama Krishna,
  • J. Sathish,
  • M. Tarun,
  • T. Rahul Mani Datta,
  • S. Raghu Vamsi,
  • S. Janu Sree

摘要

Glass and carbon fiber composite materials are extensively utilized in wind turbine blades due to their superior mechanical properties. This study presents a novel expert system for crack diagnosis and prognosis in glass/carbon fiber composite structures, aiming to enhance the safety and reliability of wind turbine blades. The proposed approach integrates experimental and numerical modal analysis to determine the first three natural frequencies of glass and carbon fiber composite beams and plates (GCCBP) under fixed free boundary conditions, considering various crack configurations. The average error between experimental and numerical modal analyses is found to be 8.64%. Additionally, experimental and numerical fatigue analyses are conducted to estimate fatigue stress and remaining useful life (RUL) for cracked GCCBP under the same boundary conditions, with a deviation of only 4.31% between experimental and numerical results. A comprehensive dataset, comprising crack length, crack depth, the first three natural frequencies, fatigue stress, and RUL derived from numerical analyses, is used to train a Gaussian process regression based expert system developed in Python. The system achieves an R-squared score of 96.0, indicating high predictive accuracy. Validation with unseen experimental data confirms the system effectiveness, demonstrating an average error of 3.82% in crack location identification, severity assessment, and RUL estimation. This work highlights the potential of machine learning for advancing structural health monitoring in composite materials.