Quality Prediction of GH4169 Alloy Turbine Discs Based on CatBoost Feature Selection and SHAP Validation

As a critical component of aircraft engines, the manufacturing quality of turbine discs directly affects engine performance. However, due to fluctuations in raw material properties and the complexity of the manufacturing process, traditional mechanistic model prediction methods are time-consuming and heavily reliant on the experience of designers, making it difficult to quickly and accurately predict the quality performance of forgings. To address this issue, this paper proposes a feature selection method based on CatBoost, validated through the SHAP framework, and combines multiple ensemble learning models to construct a predictive model for the quality performance of turbine disc forgings. Through validation with actual production data, the proposed method significantly improves prediction accuracy, particularly demonstrating strong robustness and generalization capability in handling complex process parameters and multi-dimensional feature selection. This approach offers an efficient and accurate solution for quality control in turbine disc manufacturing processes and is applicable to real industrial production scenarios.