Key Quality Characteristics Identification Based on GSCV-RF-RFE

Identifying key quality characteristics (KQCs) in manufacturing processes plays an important role in quality control. This paper proposes a method called GSCV-RF-RFE, which combines Grid Search Cross-Validation (GSCV), Random Forest (RF), and Recursive Feature Elimination (RFE), to identify KQCs in imbalanced production data. In the proposed method, the Multiple Imputation by Chained Equations (MICE) algorithm is employed for multiple interpolation of the dataset, while the parameters of the RF algorithm are optimized using the GS method. Comparative analysis demonstrates that the GSCV -RF model outperforms six algorithms, and achieves higher model accuracy than Multinomial Naive Bayes (MNB). Furthermore, the optimized RF algorithm is combined with a RFE algorithm, and adaptive synthetic sampling is employed to balance the input samples. Spearman correlation analysis is conducted to eliminate strong correlations. Experimental result reveal that the proposed combined strategy achieves a lower Class II error rate and a higher accuracy rate in quality feature identification, indicating its excellent overall performance.