Predicting surface quality in titanium alloy milling with micro-textured ball-end cutters: A weighted bidirectional temporal learning framework

In high-efficiency milling, vibration behavior reflects the overall machining state, indicating a close relationship between milling vibration characteristics and workpiece surface quality. Monitoring and analyzing vibration amplitude enables real-time evaluation of surface quality. This study proposes a surface quality prediction method for titanium alloy milling using ball-end cutters based on the weighted bidirectional temporal learning framework. A titanium alloy milling experimental platform was constructed to comparatively analyze the vibration signal patterns of different types of ball-end cutters and to explore the correlation between milling vibration behavior and tool wear. Based on vibration behavior, a surface quality prediction model for titanium alloy milling with micro-textured coated tools was developed and experimentally validated. The results show that micro-textured AlSiTiN-coated tools can significantly reduce vibration amplitude, thereby enhancing machining stability and surface quality. MIC-BP neural network analysis further confirmed a strong correlation between milling vibration behavior and tool wear. Using the weighted bidirectional temporal learning framework, the workpiece surface roughness Ra was predicted with a linear correlation of 94.2% and an RMSE of 0.0159 on the training set, and a linear correlation of 83.5% and an RMSE of 0.0265 on the testing set, demonstrating high predictive accuracy and good generalization. Comparative and ablation experiments further validated the superior performance of the proposed framework in terms of prediction accuracy, stability, and generalization. This study not only elucidates the intrinsic relationships among milling vibration behavior, tool wear, and workpiece surface quality but also provides a new approach for intelligent process monitoring and quality control in machining.