Scholarlink

<p>Erosion damage in TC4 titanium alloy poses a significant threat to the performance and reliability of aero-engine compressors, making accurate prediction methods crucial for preventive maintenance. This study proposes an integrated prediction framework combining improved Harris Hawks Optimization (iHHO), Extreme Gradient Boosting (XGBoost), and SHapley Additive exPlanations (SHAP) for erosion damage prediction. Gas flow erosion experiments were conducted on TC4 titanium alloy specimens (6 × 50 × 50&#xa0;mm) to generate high-quality datasets, complemented by a validated numerical simulation model to ensure accuracy and reliability. Leveraging this dataset, an enhanced iHHO-XGBoost model was developed for precise erosion damage prediction, while SHAP analysis was employed to interpret feature contributions and model decision-making. The proposed approach achieved an average <i>R</i><sup>2</sup> of 0.97228 and RMSE of 1.1751 × 10<sup>−8</sup> across three seeds, outperforming traditional models (BP, RF, GRNN) and their iHHO variants by improving <i>R</i><sup>2</sup> by up to 45.173% and reducing RMSE by up to 70.868%. These findings demonstrate the effectiveness of integrating iHHO optimization with machine learning for predicting erosion damage in high-performance titanium alloys, offering valuable insights for health monitoring and intelligent maintenance strategies in aero-engine applications.</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

Modeling and Prediction of Erosion Rate for Aero-Engine Compressor Blades Using an Integrated Framework Based on Improved Harris Hawks Optimization with XGBoost and SHAP

  • 期刊论文

  • 发表日期：

    2025年10月15日

查看全文
  • Hao Wang,
  • Ruijie Xu,
  • Zeshang Zhang,
  • Shilin Jiang

摘要

Erosion damage in TC4 titanium alloy poses a significant threat to the performance and reliability of aero-engine compressors, making accurate prediction methods crucial for preventive maintenance. This study proposes an integrated prediction framework combining improved Harris Hawks Optimization (iHHO), Extreme Gradient Boosting (XGBoost), and SHapley Additive exPlanations (SHAP) for erosion damage prediction. Gas flow erosion experiments were conducted on TC4 titanium alloy specimens (6 × 50 × 50 mm) to generate high-quality datasets, complemented by a validated numerical simulation model to ensure accuracy and reliability. Leveraging this dataset, an enhanced iHHO-XGBoost model was developed for precise erosion damage prediction, while SHAP analysis was employed to interpret feature contributions and model decision-making. The proposed approach achieved an average R2 of 0.97228 and RMSE of 1.1751 × 10−8 across three seeds, outperforming traditional models (BP, RF, GRNN) and their iHHO variants by improving R2 by up to 45.173% and reducing RMSE by up to 70.868%. These findings demonstrate the effectiveness of integrating iHHO optimization with machine learning for predicting erosion damage in high-performance titanium alloys, offering valuable insights for health monitoring and intelligent maintenance strategies in aero-engine applications.