Scholarlink

<p>Foundation voids, resulting from improper installation or subsurface water erosion, induce bending stresses in the pipe wall and excessive joint rotation, thereby significantly increasing the failure risk of buried water pipelines. The lack of visibility of buried pipelines complicates the assessment of structural conditions and the detection of foundation defects. While distributed fiber optic sensors facilitate real-time monitoring of longitudinal bending strains, these measurements do not directly reveal the state of the pipe–soil system. This study proposes a novel methodology based on physics-informed neural networks (PINNs) for the joint identification of foundation voids and structural deformations using distributed strain data. An analytical pipe–soil interaction model is developed for jointed pipes subject to foundation voids at arbitrary locations, yielding theoretical expressions for deflection, slope, and curvature. By parameterizing the foundation voids, a novel PINN-based approach enables the identification of void characteristics from strain measurements. The joint rotation is subsequently reconstructed by incorporating the identified void parameters into the analytical deflection model. A series of physical model tests validate the proposed method, demonstrating accurate identification of void location and length, with maximum errors of 0.2 m and 0.27&#xa0;m, respectively. The reconstructed pipeline deformations closely match experimental observations. The proposed PINN-based framework enables accurate, real-time assessment of foundation voids and associated bending deformations, offering a promising tool for early warning and risk management in buried water pipeline systems.</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

Physics-informed neural network-based methodology for joint identification of foundation voids and structural deformations of buried water pipelines

  • 期刊论文

  • 发表日期：

    2025年9月24日

查看全文
  • Juju Bai,
  • Minghao Li,
  • Xin Feng,
  • Ziyu Wu,
  • Yang Liu,
  • Yifan Zhang

摘要

Foundation voids, resulting from improper installation or subsurface water erosion, induce bending stresses in the pipe wall and excessive joint rotation, thereby significantly increasing the failure risk of buried water pipelines. The lack of visibility of buried pipelines complicates the assessment of structural conditions and the detection of foundation defects. While distributed fiber optic sensors facilitate real-time monitoring of longitudinal bending strains, these measurements do not directly reveal the state of the pipe–soil system. This study proposes a novel methodology based on physics-informed neural networks (PINNs) for the joint identification of foundation voids and structural deformations using distributed strain data. An analytical pipe–soil interaction model is developed for jointed pipes subject to foundation voids at arbitrary locations, yielding theoretical expressions for deflection, slope, and curvature. By parameterizing the foundation voids, a novel PINN-based approach enables the identification of void characteristics from strain measurements. The joint rotation is subsequently reconstructed by incorporating the identified void parameters into the analytical deflection model. A series of physical model tests validate the proposed method, demonstrating accurate identification of void location and length, with maximum errors of 0.2 m and 0.27 m, respectively. The reconstructed pipeline deformations closely match experimental observations. The proposed PINN-based framework enables accurate, real-time assessment of foundation voids and associated bending deformations, offering a promising tool for early warning and risk management in buried water pipeline systems.