Surface Roughness Prediction of Bearing Ring Precision Grinding Based on Feature Extraction

Grinding, as the most crucial finishing process for bearing rings, influences the surface integrity of bearings through the roughness of the ground surface. In order to improve the surface roughness of bearing ring grinding under multiple working conditions, a prediction model of bearing ring surface roughness based on feature extraction was proposed. Firstly, the signal was decomposed using the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) algorithm, and the sensitive components were selected based on the correlation coefficient between Intrinsic Mode Functions (IMFs) and the original signal. The time-domain, frequency-domain, and entropy-domain features of the selected IMF components were extracted. Then, Principal Component Analysis (PCA) was employed for signal feature fusion, and a feature set was constructed in combination with grinding parameters. A prediction model based on Support Vector Regression (SVR) was established to achieve regression prediction of the grinding surface roughness. The proposed method for predicting the surface roughness of precision cylindrical grinding of bearings demonstrated that the determination coefficient (R2), mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE) were 0.9953, 0.0020, 0.0050, and 0.0187, respectively. The results indicate that the incorporation of entropy features and grinding parameters in the model provide more information pertinent to grinding surface roughness, thereby effectively enhancing the predictive accuracy.