Scholarlink

<p><?tk 3?>Bridge scour is a primary cause of structural failure worldwide, posing a severe threat to piles of sea-crossing bridges. Frequency-based methods provide a promising approach for scour identification. However, the measured frequencies are highly susceptible to strong background noise, which masks subtle scour-related characteristics and makes accurate identification challenging. To overcome the limitation, this study proposes an innovative adaptive hybrid filtering method that integrates complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), detrended fluctuation analysis (DFA), and singular value decomposition (SVD). CEEMDAN is adopted to iteratively decompose vibration signals into intrinsic mode functions (IMFs). Detrended fluctuation analysis, guided by the Hurst exponent, is then used to distinguish scour-related periodic components, while discarding noise-dominated modes. Finally, residual noise is further suppressed by SVD, which leverages the relative energy difference spectrum of singular values to reconstruct the denoised signal. Numerical simulations and flume experiments on single piles were performed to validate the approach. The results confirm that the method significantly enhances signal-to-noise ratio, accurately recovers frequency shifts caused by noise, and reduces identification errors under strong noise conditions. This work provides a reliable signal-processing strategy for frequency-based scour monitoring and contributes to improving the long-term safety of bridge foundations.</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

A method for scour dynamic identification of single pile foundations based on adaptive noise hybrid filter

  • 期刊论文

  • 发表日期：

    2026年7月23日

查看全文
  • Jian Guo,
  • Ruiqian Li,
  • Chenyu Hu

摘要

Bridge scour is a primary cause of structural failure worldwide, posing a severe threat to piles of sea-crossing bridges. Frequency-based methods provide a promising approach for scour identification. However, the measured frequencies are highly susceptible to strong background noise, which masks subtle scour-related characteristics and makes accurate identification challenging. To overcome the limitation, this study proposes an innovative adaptive hybrid filtering method that integrates complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), detrended fluctuation analysis (DFA), and singular value decomposition (SVD). CEEMDAN is adopted to iteratively decompose vibration signals into intrinsic mode functions (IMFs). Detrended fluctuation analysis, guided by the Hurst exponent, is then used to distinguish scour-related periodic components, while discarding noise-dominated modes. Finally, residual noise is further suppressed by SVD, which leverages the relative energy difference spectrum of singular values to reconstruct the denoised signal. Numerical simulations and flume experiments on single piles were performed to validate the approach. The results confirm that the method significantly enhances signal-to-noise ratio, accurately recovers frequency shifts caused by noise, and reduces identification errors under strong noise conditions. This work provides a reliable signal-processing strategy for frequency-based scour monitoring and contributes to improving the long-term safety of bridge foundations.