Scholarlink

<p>The performance of final products in various industries is greatly influenced by the mechanical properties (MPs) of wood. However, if wood is not properly treated, it can develop issues such as moisture damage, decay, and insect infestation. This manuscript presents an innovative technique for forecasting the MP of different wood types by combining guided Lamb wave propagation with a quantum complete graph neural network (QCGNN). The study uses green poplar wood specimens labeled S1–S5, with experimental data including Lamb wave velocities in multiple directions, and density and moisture content. The QCGNN model leverages this rich dataset to accurately predict the modulus of elasticity (MoE) and modulus of rupture (MoR) across varying moisture contents. Guided Lamb waves capture detailed, nondestructive velocity data reflecting wood’s internal state, while the QCGNN models complex nonlinear links between these signals and mechanical properties, including moisture effects. The main contribution of this research lies in integrating physics-based nondestructive testing with advanced graph-based machine learning to improve prediction accuracy and robustness. The proposed method is evaluated utilizing the MATLAB platform and benchmarked against conventional methods such as artificial neural networks, Grey wolf optimizer, and convolutional neural networks. It achieves 99% accuracy with a strong correlation for MoE and MoR, and minimal error, demonstrating superior performance. This research offers a highly accurate, efficient, and scalable method for nondestructive wood property prediction.</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

Enhanced nondestructive testing for mechanical characterization of wood using guided waves and quantum complete graph neural networks

  • 期刊论文

  • 发表日期：

    2025年7月2日

查看全文
  • K. Venkatesan,
  • S. Sivalingam,
  • Srinivasan Rajaram,
  • R. Ashok

摘要

The performance of final products in various industries is greatly influenced by the mechanical properties (MPs) of wood. However, if wood is not properly treated, it can develop issues such as moisture damage, decay, and insect infestation. This manuscript presents an innovative technique for forecasting the MP of different wood types by combining guided Lamb wave propagation with a quantum complete graph neural network (QCGNN). The study uses green poplar wood specimens labeled S1–S5, with experimental data including Lamb wave velocities in multiple directions, and density and moisture content. The QCGNN model leverages this rich dataset to accurately predict the modulus of elasticity (MoE) and modulus of rupture (MoR) across varying moisture contents. Guided Lamb waves capture detailed, nondestructive velocity data reflecting wood’s internal state, while the QCGNN models complex nonlinear links between these signals and mechanical properties, including moisture effects. The main contribution of this research lies in integrating physics-based nondestructive testing with advanced graph-based machine learning to improve prediction accuracy and robustness. The proposed method is evaluated utilizing the MATLAB platform and benchmarked against conventional methods such as artificial neural networks, Grey wolf optimizer, and convolutional neural networks. It achieves 99% accuracy with a strong correlation for MoE and MoR, and minimal error, demonstrating superior performance. This research offers a highly accurate, efficient, and scalable method for nondestructive wood property prediction.