Intelligent tool wear prediction based on deep learning PSD-CVT model.

Area of Science:

• Manufacturing Engineering
• Artificial Intelligence
• Signal Processing

Background:

• Reliable machining quality depends on accurate tool wear prediction.
• Existing methods may not fully capture complex signal characteristics for wear monitoring.

Purpose of the Study:

• To propose a novel deep learning model, PSD-CVT, for enhanced tool wear prediction.
• To leverage the strengths of power spectral density (PSD), convolutional neural networks (CNN), and vision transformer models (ViT) in a unified framework.

Main Methods:

• Developed the PSD-CVT model integrating PSD maps for spectral analysis, CNN for local feature extraction, and ViT for global dependencies.
• Utilized two fully connected layers with a ReLU activation function for predicting tool wear values.
• Evaluated the model on the PHM 2010 dataset.

Main Results:

• The PSD-CVT model demonstrated higher prediction accuracy compared to standalone CNN or ViT models.
• The proposed model outperformed several existing methods in tool wear prediction accuracy.
• Experimental results validate the effectiveness of the integrated approach.

Conclusions:

• The PSD-CVT model offers a robust and accurate solution for tool wear prediction.
• This novel deep learning approach can be applied to various machining fields for improved quality control.
• The synthesis of PSD, CNN, and ViT provides a powerful tool for predictive maintenance.