Deep learning-based tool wear prediction and its application for machining process using multi-scale feature fusion and channel attention mechanism

Tool wear is a key factor in the cutting process, which directly affects the machining precision and part quality. Accurate tool wear prediction can make proper tool change at an early stage to reduce downtime and enhance product quality. However, traditional methods can not meet the high requirements of the intelligent manufacturing. Therefore, a novel method based on deep learning is proposed to improve the prediction accuracy of tool wear. The multi-scale feature fusion was implemented by the developed parallel convolutional neural networks. The channel attention mechanism combined with the residual connection was developed to consider the weight of the different feature map to enhance the performance of the model. The different tool wear prediction experiments were implemented to verify the superiority of the developed method, and the prediction results of tool wear are more robust and accurate than current methods. Finally, a tool wear monitoring system was developed and applied to the tapping process of the engine cylinder to ensure the quality of the engine cylinder and the stability of the machining process.