Neural networks prediction of different frequencies effects on corrosion resistance obtained from pulsed nanocrystalline plasma electrolytic carburizing

This paper deals with the surface modification of commercially pure titanium by using pulsed nanocrystalline plasma electrolytic carburizing. In order to fully characterize the complex underlying mechanism of this process and evaluate the effects of a thorough range of frequencies, a prediction model is developed using a hybrid of Neural Networks and Genetic Algorithms (GA). Process variables, i.e. time, frequency and corrosion resistance of nanocrystalline carbides, have been experimentally studied. Corrosion resistances were measured by PDS technique for different coated samples. A portion of this dataset is used to train the prediction model, while the rest is set aside to test its predictive performance. This hybrid Neural Networks model uses GA to achieve its optimal architecture for prediction. Finally, it is concluded that the proposed model has an excellent prediction capability of final corrosion resistance of nanocrystalline carbides in the various range of frequencies by comparing the results with the experimental data.