Related Experiment Video
Updated: Sep 12, 2025
08:47
Mimicking and Measuring Occlusal Erosive Tooth Wear with the "Rub&Roll" and Non-contact Profilometry
Published on: February 2, 2018
12.3K
Grinding wheel wear evaluation with the PMSCNN model.
Sumei Si1,2, Zekai Si1,3, Deqiang Mu4
1College of Electromechanical Engineering, Changchun University of Technology, Changchun, 130012, China.
Scientific Reports
|August 6, 2025
Summary
A new model, PMSCNN, accurately assesses grinding wheel wear using motor current signals and machine learning. This method enhances machining efficiency and quality by predicting wear trends effectively.
Area of Science:
- Manufacturing Engineering
- Materials Science
- Artificial Intelligence
Background:
- Grinding wheel wear critically impacts machining efficiency and product quality.
- Accurate assessment of grinding wheel wear is essential for optimizing manufacturing processes.
Purpose of the Study:
- To develop and validate a novel grinding wheel wear assessment model named PMSCNN.
- To improve the accuracy and reliability of predicting grinding wheel wear using machine learning.
Main Methods:
- A Convolutional Neural Network (CNN) and Transformer model (PMSCNN) were developed for wear assessment.
- Grinding wheel spindle motor current signals were measured and processed using median filtering.
- Feature importance was analyzed using a gradient boosting regressor, selecting the top four features.
- The PMSCNN model's predictive accuracy was validated using these selected features.
Main Results:
- The PMSCNN model demonstrated good similarity between predicted and real wear trends.
- Cross-validated results showed an average Mean Absolute Error (MAE) of 3.028, Root Mean Square Error (RMSE) of 3.938, and R-squared (R²) of 0.919.
- Modular analysis confirmed each component's contribution to the model's performance.
Conclusions:
- The PMSCNN model effectively extracts wear-related patterns from current signals.
- The developed model achieves high prediction accuracy for grinding wheel wear.
- This approach offers a promising solution for real-time monitoring and assessment of grinding wheel wear.
