An Improved ResNet-1d with Channel Attention for Tool Wear Monitor in Smart Manufacturing.

Area of Science:

• Manufacturing Engineering
• Artificial Intelligence
• Machine Learning

Background:

• Tool wear significantly impacts machining efficiency, tool longevity, and workpiece quality.
• Accurate monitoring and diagnosis of tool condition are essential for optimizing machining processes.

Purpose of the Study:

• To propose an improved CaAt-ResNet-1d model for multi-sensor tool wear diagnosis.
• To enhance the accuracy and effectiveness of tool wear monitoring systems.

Main Methods:

• Utilized a ResNet18 architecture with a one-dimensional convolutional neural network (1D CNN) for time-series data feature extraction.
• Integrated channel attention mechanisms (CaAt1 and CaAt5) into the residual network blocks to automatically learn channel features.
• Validated the model on the PHM2010 dataset.

Main Results:

• The CaAt-ResNet-1d model achieved 89.27% accuracy in tool wear diagnosis.
• Demonstrated an improvement of approximately 7% over Gated-Transformer and 3% over Resnet18.
• The model effectively captured and utilized multi-sensor data for diagnosis.

Conclusions:

• The proposed CaAt-ResNet-1d model is effective and capable for real-time tool wear monitoring.
• The integration of channel attention mechanisms enhances diagnostic performance.
• This approach offers a promising solution for improving machining process control and reliability.