Scholarlink

Accurate fatigue crack evaluation is vital for structural life management. Long-term aircraft monitoring faces challenges from heteroscedasticity—changing uncertainty distributions under varying operational conditions—which degrades conventional diagnostics’ accuracy. This study develops a heteroscedastic Gaussian process model combined with guided-wave SHM, using quantile regression to estimate crack evaluation distribution. The median regression output replaces traditional mean estimation, while quantile-specific confidence intervals explicitly represent heteroscedastic uncertainties. Validated through aircraft attachment lug fatigue tests, the method enhances evaluation quality and uncertainty quantification capabilities.

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

Heteroscedastic Gaussian Process Modeling for Crack Evaluation Under Stochastic Loads

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Wang Hui

摘要

Accurate fatigue crack evaluation is vital for structural life management. Long-term aircraft monitoring faces challenges from heteroscedasticity—changing uncertainty distributions under varying operational conditions—which degrades conventional diagnostics’ accuracy. This study develops a heteroscedastic Gaussian process model combined with guided-wave SHM, using quantile regression to estimate crack evaluation distribution. The median regression output replaces traditional mean estimation, while quantile-specific confidence intervals explicitly represent heteroscedastic uncertainties. Validated through aircraft attachment lug fatigue tests, the method enhances evaluation quality and uncertainty quantification capabilities.