Scholarlink

<p>To address insufficient modal parameter identification accuracy for time-varying signals under underdetermined conditions, this study proposes the Adaptive Signal Window with Forgetting Factor (ASWFF) algorithm. The method integrates a tensor decomposition framework and dynamically adjusts sliding window size using signal change rates to adapt to time-varying dynamics. It further introduces a forgetting factor to optimize weight allocation, suppress noise, and enhance time-varying feature tracking. Modal parameters are progressively identified per window using sensor data within the sliding window, enabling continuous structural tracking via window shifting and parameter updating. Validation on a time-varying three-degree-of-freedom spring-mass oscillator and a liquid-filled cylindrical structure demonstrates that ASWFF significantly outperforms variance contribution rate (VCRWLA) and smoothness-based (SWLA) window length adaptive algorithms in modal consistency (MAC values) and inherent frequency accuracy. Crucially, ASWFF exhibits superior adaptability in time-varying scenarios.</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

Robust Identification of Underdetermined Modal Parameters for Time-Varying Structures Based on Signal Change Rate and Variable Window Length

  • 期刊论文

  • 发表日期：

    2025年6月26日

查看全文
  • Haonan Chen,
  • Cheng Wang,
  • Jin Jiang,
  • Lincong Chen

摘要

To address insufficient modal parameter identification accuracy for time-varying signals under underdetermined conditions, this study proposes the Adaptive Signal Window with Forgetting Factor (ASWFF) algorithm. The method integrates a tensor decomposition framework and dynamically adjusts sliding window size using signal change rates to adapt to time-varying dynamics. It further introduces a forgetting factor to optimize weight allocation, suppress noise, and enhance time-varying feature tracking. Modal parameters are progressively identified per window using sensor data within the sliding window, enabling continuous structural tracking via window shifting and parameter updating. Validation on a time-varying three-degree-of-freedom spring-mass oscillator and a liquid-filled cylindrical structure demonstrates that ASWFF significantly outperforms variance contribution rate (VCRWLA) and smoothness-based (SWLA) window length adaptive algorithms in modal consistency (MAC values) and inherent frequency accuracy. Crucially, ASWFF exhibits superior adaptability in time-varying scenarios.