1. ANN-Based wear performance prediction for plasma nitrided Ti6AI4V Alloy
- Author
-
Hülya Durmuş, Fatih Kahraman, and Süleyman Karadeniz
- Subjects
Materials science ,Surface treatment ,chemistry.chemical_element ,Different treatments ,Diffusion layer ,Neural techniques ,Diffusion layer thickness ,Deep neural networks ,Performance prediction ,Titanium alloys ,General Materials Science ,Dry sliding wear test ,Experimental conditions ,Mechanical Engineering ,Metallurgy ,Nitrogen plasma ,Titanium alloy ,Rotational speed ,Plasma ,Artificial neural network models ,chemistry ,Mechanics of Materials ,Statistical performance ,Surface modification ,Plasma applications ,Tin ,Thickness and hardness ,Nitriding ,Neural networks ,Forecasting - Abstract
Surface modification of a Ti6Al4V titanium alloy was made by the plasma nitriding process. Plasma nitriding was performed in a constant gas mixture of 20% H2–80% N2 at temperatures between 700 and 1000° C and process times between 2 and 15 h. Samples nitrided at different treatment times and temperatures were subjected to the dry sliding wear test using the pin-on-disc set up under 80N normal load with rotational speed of counter face disc of 0.8 m/s at room conditions. An artificial neural network (ANN) model of was developed for prediction of wear performance of the plasma nitrided Ti6Al4V alloy. The inputs of the ANN model were processing times and temperatures, diffusion layer thickness, Ti2N thickness, TiN thickness and hardness. The output of the ANN model was wear loss. The model is based on the multilayer backpropagation neural technique. The ANN was trained with a comprehensive dataset collected from experimental conditions and results of authors. The model can be used for the prediction of wear properties of Ti6Al4V alloys nitrided at different parameters. The ANN model demonstrated the best statistical performance with the experimental results.
- Published
- 2012