Abstract:

This paper aims to overcome the difficulty in the modelling of nitride hardening of piston rings. In the in- vestigation, the feature parameters of nitridation process are extracted using the principal component analysis method to reduce the dimension of input samples in the quality model. Then, a quality prediction model of the key process for piston ring manufacturing is built based on the wavelet Elman neural network. The proposed model helps to pre- dict the process quality fluctuation and lays a foundation for further process optimization and quality improvement. Experimental results show that the proposed method effectively improves the quality control of nitride hardening, and that the proposed prediction model predicts more accurately and converges more quickly than the normal Elman neural network, showing an accuracy of output-quality characteristic value of 89%.

Key words: piston ring, nitride hardening, principal component analysis, Elman neural network, wavelet neural network, quality prediction