Active learning regression quality prediction model and grinding mechanism for ceramic bearing grinding processing

Area of Science:

• Materials Science and Engineering
• Manufacturing Processes

Background:

• Ceramic bearings require high precision grinding for optimal performance.
• Surface roughness is a critical quality indicator in bearing manufacturing.

Purpose of the Study:

• To predict and optimize surface quality in ceramic bearing grinding.
• To investigate the impact of key grinding parameters on surface roughness.

Main Methods:

• Utilized active learning regression for model construction and optimization.
• Employed deep learning models for quality prediction.
• Conducted empirical analysis under varied grinding conditions (wheel speed, depth, feed rate).

Main Results:

• Increased grinding depth to 21 μm reduced surface roughness (Ra) to 0.1624 μm.
• Higher grinding wheel linear velocity (45 m/s) and adjusted depth (0.015 mm) yielded Ra of 0.1876 μm.
• Model training loss decreased to 0.03622 with increased grinding depth.

Conclusions:

• Grinding depth and wheel linear velocity are significant factors in improving ceramic bearing surface quality.
• The developed models offer theoretical support and practical guidance for optimizing grinding parameters.
• Findings have important industrial value for enhancing ceramic bearing production.