Tool wear prediction based on multifeature fusion and deep learning

Recent advancements in artificial intelligence and industrial big data have revolutionized tool wear monitoring. This study proposes a BO-Bi-LSTM model leveraging multi-sensor feature fusion for wear prediction, integrating cutting force, vibration, and acoustic emission data to overcome single-sensor limitations in accuracy and robustness. The framework employs streamlined feature-level fusion for optimal information retention and Bayesian-optimized bidirectional LSTM networks for enhanced temporal pattern learning. Validated under complex machining conditions, the solution enables reliable real-time wear monitoring while advancing intelligent manufacturing system capabilities.