Cross-Material Adversarial Neural Network Tool Wear Monitoring for Drilling Aerospace Stacks

In the aerospace sector, multi-material stacks are widely used. However, it is harder for monitoring tool wear in drilling the stacks, due to the non-uniformed properties of materials leading data distribution shifts for the models. To address the inconsistent feature distribution, this work proposed a cross-domain adaption network to examinate tool wear in drilling CFRP/AI stacks. The adversarial-based adaptation is established and incorporated channel attention mechanism to promote learning domain invariant features. It adapts to data distribution shifts, offering a robust tool condition monitoring for the complete drilling process. The experiment validated the performance of network, showing the improvements in tool wear monitoring accuracy. The cross-material adversarial neural network contributes to improving tool wear monitoring for multi-material stacks.