Scholarlink

<p>Structural health monitoring (SHM) fundamentally relies on optimal sensor placement (OSP) for accurate damage detection; however, traditional OSP frameworks based on deterministic assumptions often face significant challenges in balancing detection reliability, sensor budget constraints, and robustness against uncertainties arising from real-world damage scenarios. This study introduces a robust optimization framework that integrates criteria derived from structural modal characteristics, the Stochastic Subspace Identification (SSI) method, and a Bayesian inference strategy to overcome these limitations. First, covariance-driven SSI extracts modal parameters from environmental vibration data without the need for known excitations for healthy and damaged states. Next, a novel index is introduced that quantifies the normalized changes in modal strain energy (MSE) under single- and multi-element damage scenarios with stochastic sampling. This index guides the sensor layout optimization via a max–min genetic algorithm (GA) that maximizes the minimum detection capability under a predefined sensor budget. Finally, to enhance damage localization accuracy, the results from the reference-based SSI detection method are integrated with the Bayesian strategy. The process is validated numerically on two 3D truss examples under simulated operational conditions (Gaussian white-noise excitation and additive sensor noise at SNR = 40 and 30&#xa0;dB), demonstrating robust damage detectability and sensor placements.</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 Robust Optimization of Sensor Placement for Truss Structures in Structural Health Monitoring

  • 期刊论文

  • 发表日期：

    2026年6月14日

查看全文
  • Soheil Sarparast Sadat,
  • Amin Gholizad

摘要

Structural health monitoring (SHM) fundamentally relies on optimal sensor placement (OSP) for accurate damage detection; however, traditional OSP frameworks based on deterministic assumptions often face significant challenges in balancing detection reliability, sensor budget constraints, and robustness against uncertainties arising from real-world damage scenarios. This study introduces a robust optimization framework that integrates criteria derived from structural modal characteristics, the Stochastic Subspace Identification (SSI) method, and a Bayesian inference strategy to overcome these limitations. First, covariance-driven SSI extracts modal parameters from environmental vibration data without the need for known excitations for healthy and damaged states. Next, a novel index is introduced that quantifies the normalized changes in modal strain energy (MSE) under single- and multi-element damage scenarios with stochastic sampling. This index guides the sensor layout optimization via a max–min genetic algorithm (GA) that maximizes the minimum detection capability under a predefined sensor budget. Finally, to enhance damage localization accuracy, the results from the reference-based SSI detection method are integrated with the Bayesian strategy. The process is validated numerically on two 3D truss examples under simulated operational conditions (Gaussian white-noise excitation and additive sensor noise at SNR = 40 and 30 dB), demonstrating robust damage detectability and sensor placements.