Scholarlink

<p>Aircraft icing can significantly degrade aerodynamic performance and compromise flight safety, highlighting the need for efficient prediction of icing characteristics and critical influencing factors. In this study, a structured set of icing simulation parameters was defined by identifying key variables and constraining their ranges in accordance with airworthiness requirements of intermittent maximum icing conditions. According to the simulation results, it concludes that heavier ice accretion leads to increased surface roughness and consequently more severe aerodynamic degradation under the intermittent maximum icing conditions. Within this admissible domain, representative samples were generated and applied to FENSAP-ICE to simulate ice accretion on the reference aircraft airfoil, thereby constructing a comprehensive icing dataset with the iced mass as the output response. Based on this dataset, a surrogate model employing the Random Forest algorithm was developed, enabling rapid and reliable prediction of icing outcomes. Furthermore, sensitivity analysis was conducted to identify the dominant parameters governing the icing process, offering valuable insights into the core factors that critically influence aircraft icing behavior.</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

Research on a surrogate model and sensitivity analysis for aircraft icing under intermittent maximum icing conditions based on random forest

  • 期刊论文

  • 发表日期：

    2026年2月4日

查看全文
  • Chao Xi,
  • Tingting Cheng,
  • Shumin Pu,
  • Xiaoliang Wang

摘要

Aircraft icing can significantly degrade aerodynamic performance and compromise flight safety, highlighting the need for efficient prediction of icing characteristics and critical influencing factors. In this study, a structured set of icing simulation parameters was defined by identifying key variables and constraining their ranges in accordance with airworthiness requirements of intermittent maximum icing conditions. According to the simulation results, it concludes that heavier ice accretion leads to increased surface roughness and consequently more severe aerodynamic degradation under the intermittent maximum icing conditions. Within this admissible domain, representative samples were generated and applied to FENSAP-ICE to simulate ice accretion on the reference aircraft airfoil, thereby constructing a comprehensive icing dataset with the iced mass as the output response. Based on this dataset, a surrogate model employing the Random Forest algorithm was developed, enabling rapid and reliable prediction of icing outcomes. Furthermore, sensitivity analysis was conducted to identify the dominant parameters governing the icing process, offering valuable insights into the core factors that critically influence aircraft icing behavior.