RC-enhanced CNN-ViT: A model for tool wear state recognition in turbine disk fir-tree slot broaching

Owing to the unique and complex multi-tooth structure of broaching tools and the single-pass feed machining process, direct monitoring methods suffer from inherent limitations. To address this challenge, an RC-Enhanced CNN-ViT model is proposed for tool wear state recognition in turbine disk fir-tree slot broaching. The proposed model integrates the strengths of convolutional neural networks and Transformers, while innovatively incorporating residual connections to fully exploit the efficiency of depthwise separable convolutions and enhance network stability. Subsequently, the Transformer module captures global dependencies to achieve a more comprehensive feature representation. Experimental results demonstrate the superior performance of the proposed method: on the test set, the accuracy, precision, recall, and F1-score reach 98.63%, 98.63%, 98.63%, and 98.62%, respectively, with a single-sample inference time of only 1.4 ms—less than half that of the original ViT model. These results indicate that the proposed approach provides valuable technical support and practical guidance for predictive maintenance of cutting tools in broaching operations.