PREDICTING THE TENSILE STRENGTH OF FRICTION STIR WELDED DISSIMILAR ALUMINUM ALLOY USING ANN

An Artificial Neural Network (ANN) model was developed to predict the tensile strength of the dissimilar aluminum weld using Friction Stir Welding (FSW) process parameters. The aluminum plates selected for the weld study are AA2014T651 and AA6063T651. The plate AA2014T651 which is the harder metal kept on the advancing side and AA6063T651 kept on the retreating side. The input FSW process parameters such as pin diameter, tool geometry, tool offset and the output parameter -tensile strength as were taken the development of ANN model. Good performance of the ANN model was achieved. Eighteen experiments were conducted and responses of tensile strength of the weld were measured. The 70% of data were used for training purpose, 15% for testing the model and 15% for validation. Levenberg-Marquardt algorithm was used for training the ANN model. The ANN model was able to predict the tensile strength with an accuracy of ± 98 % i.e. within an error of ± 2%. The optimized process parameter based on ANN model are 7mm pin diameter and 4 degree tilt angle in which tool is offset towards advancing side exhibits better tensile strength.