Scholarlink

<p>Misalignment faults commonly occur in rotor systems due to prolonged operation, severely compromising operational efficiency and performance. This paper proposes a Laplace prior-enhanced sparse Bayesian learning method to identify rotor systems’ misalignment faults and quantify associated uncertainties. First, parallel, angular, and integrated misalignments are analyzed and are equal to nonlinear forces that act on the rotor finite element model. Then, an objective function is established to find misalignment parameters that minimize the measured and theoretical displacements. Given the inherent spatial sparsity of misalignments, the Laplace prior with fewer hyperparameters is considered within the Bayesian framework and sparsely regress the objective function. Finally, the response sensitivity analysis of misalignments is derived to linearize the objective function. This approximate linear regression identifies misalignment parameters. Beyond parameter identification, this method quantifies uncertainties in the identified faults through the posterior probability density function, offering valuable information for supporting system reliability analysis. Numerical simulations and experimental studies demonstrate the method’s accuracy, efficiency, and robustness in identifying misalignments and quantifying uncertainties.</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

Misalignment identification and uncertainty quantification of rotor systems using Laplace prior-enhanced sparse Bayesian learning

  • 期刊论文

  • 发表日期：

    2025年5月17日

查看全文
  • Jie Li,
  • Li Wang,
  • Zuoqiu Liu,
  • Zhong-Rong Lu,
  • Dahao Yang

摘要

Misalignment faults commonly occur in rotor systems due to prolonged operation, severely compromising operational efficiency and performance. This paper proposes a Laplace prior-enhanced sparse Bayesian learning method to identify rotor systems’ misalignment faults and quantify associated uncertainties. First, parallel, angular, and integrated misalignments are analyzed and are equal to nonlinear forces that act on the rotor finite element model. Then, an objective function is established to find misalignment parameters that minimize the measured and theoretical displacements. Given the inherent spatial sparsity of misalignments, the Laplace prior with fewer hyperparameters is considered within the Bayesian framework and sparsely regress the objective function. Finally, the response sensitivity analysis of misalignments is derived to linearize the objective function. This approximate linear regression identifies misalignment parameters. Beyond parameter identification, this method quantifies uncertainties in the identified faults through the posterior probability density function, offering valuable information for supporting system reliability analysis. Numerical simulations and experimental studies demonstrate the method’s accuracy, efficiency, and robustness in identifying misalignments and quantifying uncertainties.