Prediction of the Mechanical Properties of Titanium Alloy Castings Based on a Back-Propagation Neural Network

The mechanical properties of titanium alloy castings are very important for their wide applications in high-end equipment and engineering; however, testing and characterization of the mechanical parameters of titanium alloy castings are complicated and costly. Therefore, the present work proposed a novel method based on a back-propagation (BP) neural network to predict the mechanical properties of a TC4 titanium alloy casting, specifically, the presence of shrinkage cavities at a given location. It was found that the statistical error between predicted values of the BP neural network and experimental results was less than 10%, indicating that the proposed model is suitable for predicting the presence of shrinkage cavities in TC4 titanium alloy castings. Moreover, the BP neural network model was also used to predict the grain size and hardness of the titanium alloy casting. The correlation between predicted and experimental results was r = 0.99485, thus indicating that the proposed model could effectively predict the grain size and hardness of TC4 titanium alloy castings.