Scholarlink

<p>This paper proposes a bridge structural assessment method based on the regression surface area of the dissipation coefficient (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\eta \)</EquationSource> </InlineEquation>), integrated with Bayesian deep learning (BDL) to optimize vibration signal analysis. With advances in sensor technology, high-precision vibration data can now be collected; however, traditional methods have not fully exploited the potential of these data. The study uses the dissipation coefficient (<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\eta \)</EquationSource> </InlineEquation>) to evaluate structural degradation based on elasticity and viscous damping properties. Instead of relying solely on statistical indicators such as mean and coefficient of variation (CV), this research introduces the surface area (<InlineEquation ID="IEq3"> <EquationSource Format="TEX">\(S\)</EquationSource> </InlineEquation>) of the regression plane as a comprehensive metric, offering a more detailed depiction of structural deterioration. In addition, the paper develops an artificial neural network that converts vibration measurement data into a rating scale from 1 to 10, allowing real-time monitoring of structural conditions. The proposed method was tested on the Saigon Bridge and Song Than Bridge, demonstrating its ability to accurately reflect structural degradation over time and the effectiveness of maintenance interventions. The results indicate that this approach not only provides detailed insights into the structural degradation process, but also facilitates automated monitoring and informed decision making in infrastructure maintenance.</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

Bayesian Deep Learning for Bridge Structural Damage Assessment Based on Dissipation Coefficient Evaluation

  • 期刊论文

  • 发表日期：

    2026年7月5日

查看全文
  • Thanh Q. Nguyen,
  • Dong Phuong Nguyen,
  • Phuoc T. Nguyen,
  • Thuy T. Nguyen

摘要

This paper proposes a bridge structural assessment method based on the regression surface area of the dissipation coefficient ( \(\eta \) ), integrated with Bayesian deep learning (BDL) to optimize vibration signal analysis. With advances in sensor technology, high-precision vibration data can now be collected; however, traditional methods have not fully exploited the potential of these data. The study uses the dissipation coefficient ( \(\eta \) ) to evaluate structural degradation based on elasticity and viscous damping properties. Instead of relying solely on statistical indicators such as mean and coefficient of variation (CV), this research introduces the surface area ( \(S\) ) of the regression plane as a comprehensive metric, offering a more detailed depiction of structural deterioration. In addition, the paper develops an artificial neural network that converts vibration measurement data into a rating scale from 1 to 10, allowing real-time monitoring of structural conditions. The proposed method was tested on the Saigon Bridge and Song Than Bridge, demonstrating its ability to accurately reflect structural degradation over time and the effectiveness of maintenance interventions. The results indicate that this approach not only provides detailed insights into the structural degradation process, but also facilitates automated monitoring and informed decision making in infrastructure maintenance.