Prediction of performance parameters in friction stir processing using ANN and multiple regression models

Abstract

Friction stir processing (FSP) is a novel approach for achieving desired surface properties, grain microstructure refinement, fixing cracks, and removing casting imperfections. The quality of desired properties directly depends on the prediction of peak temperature. In this paper, an artificial neural network (ANN) model was developed to predict peak temperature and microhardness and then compared with multiple regression predicted results. The ANN model was developed using Taguchi L16 orthogonal experimental data sets. ANN and multiple regression-based models prediction capability were evaluated using performance criteria such as mean absolute percentage error (MAPE), mean square error (MSE), mean absolute error (MAE) and determinant coefficient (R2). The result shows that the predicted values by multiple regression and the developed ANN model were close to the experimental result. The MAPE, MSE and MAE values of the ANN model predicted results were lower than multiple regression. Also, the R2 value of the ANN model was almost close to 1, indicating the best correlation between experimental and predicted results. Thus, the ANN model predicted the performance parameters more accurately than the multiple regression.