Process Quality Prediction for Piston Ring Based on Improved Wavelet Elman

Quality prediction and control methods are the key to obtaining safe and reliable operation in process quality control. To address the complex factors and strong correlation of the piston ring nitride hardening process, sliding window kernel principal component (KPCA) analysis is used to eliminate the secondary factors and extract the primary factors. The quality prediction model using wavelet output-hidden-input feedback Elman (OHIF Elman) neural network was successfully applied to the nitriding process, and the network model was trained using a hybrid algorithm of the Chicken Swarm Optimization (CSO) algorithm to coordinate the prediction accuracy and prediction efficiency. Simulation and experimental results show that the quality model can effectively predict the eigenvalues of the process quality, and also identify the abnormal change rule to improve the process control accuracy.