Scholarlink

<p>The aero-engine load spectrum is the basis for fatigue life analysis. It is highly challenging to model the fatigue load characteristics in the load spectrum for aero-engines with complex maneuvering loads and numerous extreme operating conditions. In this paper, a parameterized modeling method of the fatigue load characteristics for aero-engines is proposed based on a novel mixture distribution. The novel mixture distribution is composed of hybrid-type components including Weibull–Normal distribution, Gaussian mixture distribution, and Normal–Normal distribution. An improved Expectation–Maximization algorithm is employed for parameter estimation of the novel mixture distribution model. The results demonstrate that the method proposed improves the fitting accuracy significantly and avoids overfitting. The multi-correlation coefficient reaches 0.9920 and the Kolmogorov–Smirnov errors of marginal distributions are only 0.0090 and 0.0072. Both indicators are superior to those of the mixture distribution composed of single-type components. With the method, the model parameters of the fatigue load characteristics of aero-engines are obtained, providing a basis for the compilation of the aero-engine life test spectrum.</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

Fatigue Load Characteristic Modeling Method for Aero-engines Based on the Novel Mixture Distribution

  • 期刊论文

  • 发表日期：

    2025年4月18日

查看全文
  • Shunyu Yao,
  • Xuming Niu,
  • Zhigang Sun,
  • Yingdong Song

摘要

The aero-engine load spectrum is the basis for fatigue life analysis. It is highly challenging to model the fatigue load characteristics in the load spectrum for aero-engines with complex maneuvering loads and numerous extreme operating conditions. In this paper, a parameterized modeling method of the fatigue load characteristics for aero-engines is proposed based on a novel mixture distribution. The novel mixture distribution is composed of hybrid-type components including Weibull–Normal distribution, Gaussian mixture distribution, and Normal–Normal distribution. An improved Expectation–Maximization algorithm is employed for parameter estimation of the novel mixture distribution model. The results demonstrate that the method proposed improves the fitting accuracy significantly and avoids overfitting. The multi-correlation coefficient reaches 0.9920 and the Kolmogorov–Smirnov errors of marginal distributions are only 0.0090 and 0.0072. Both indicators are superior to those of the mixture distribution composed of single-type components. With the method, the model parameters of the fatigue load characteristics of aero-engines are obtained, providing a basis for the compilation of the aero-engine life test spectrum.