Wear parametric Optimization of FSW parameters on Al Alloy Using MLP technique

Abstract

The purpose of this study was to identify the suitable Friction Stir Welding (FSW) parameters that would be used for welding scrapped Al alloy plates. On the experimental side, the study used four Factor Three-Level Full Factorial Design of Experiments (DoE) approach. Some of the input parameters incorporated in the analysis were the applied load, the sliding speed, displacement and weight percentage of Al2O3 reinforcement, and some of the output parameters were the specific wear rate and the coefficient of friction. The above said optimum parameters were established using the Minitab software while the above said experimental results was estimated using multilayer perceptron of the feed forward 4–10–1 network. For the actual test data set in the given experiment, the overall performance of the MLP predictions resulted to an R2. This results to a coefficient of determination (R R2 of 0.98474 and a mean squared error (MSE) of 0. 025075. Therefore the high R values, which are near to 1, show that the actual values mean and the predicted values mean are closely matched.