A Prediction Model Based on Simulation (FEA) and Response Surface Methodology (RSM) for Sintered Al–TiB2 Preforms

The current study centers on two techniques: neural network (NN) and response surface methodology (RSM), applied to predict the final density (FD) of sintered aluminum preforms. In this work, the load, the aspect ratio and the initial preform density were taken as input parameters and the response (output) variable measured was FD. Prediction for the response variable FD was obtained with the help of empirical relation between the response and the input variables using RSM’s (RSM) Box–Behnken design of experimental technique and also through Neural Network (NN). Predicted values of the response by both the techniques, i.e., RSM and NN were compared with the experimental values and their closeness with the experimental values was determined. Moreover, it has been discovered that the aspect ratio has minimal impact on densification and that the FD of the preform rises with both the load applied and the initial preform density of the sintered preforms. The authors were able to predict the FD of sintered preforms of Al–TiB₂ for different initial preform and aspect ratio conditions by using NN and RSM techniques.