Prediction of mechanical properties and optimization of process parameters in friction-stir-welded dissimilar aluminium alloys

Abstract

<jats:sec> <jats:title content-type="abstract-subheading">Purpose</jats:title> <jats:p>This study aims to predict the mechanical properties such as equivalent tensile strength and micro-hardness of friction-stir-welded dissimilar aluminium alloy plates AA 6063-O and AA 2014-T6, using artificial neural network (ANN).</jats:p> </jats:sec> <jats:sec> <jats:title content-type="abstract-subheading">Design/methodology/approach</jats:title> <jats:p>The ANN model used for the experiment was developed through back propagation algorithm. The input parameter of the model consisted of tool rotational speed and weld-traverse speed whereas the output of the model consisted of mechanical properties (tensile strength and hardness) of the joint formed by friction-stir welding (FSW) process. The ANN was trained for 60% of the experimental data. In addition, the impact of the process parameters (tool rotational speed and weld-traverse speed) on the mechanical properties of the joint was determined by Taguchi Grey relational analysis.</jats:p> </jats:sec> <jats:sec> <jats:title content-type="abstract-subheading">Findings</jats:title> <jats:p>Subsequently, testing and validation of the ANN were done using experimental data, which were not used for training the network. From the experiment, it was inferred that the outcomes of the ANN are in good agreement with the experimental data. The result of the analyses showed that the tool rotational speed has more impact than the weld-traverse speed.</jats:p> </jats:sec> <jats:sec> <jats:title content-type="abstract-subheading">Originality/value</jats:title> <jats:p>The developed neural network can be used to predict the mechanical properties of the weld. Results indicate that the network prediction is similar to the experiment results. Overall regression value computed for training, validation and testing is greater than 0.9900 for both tensile strength and microhardness. In addition, the percentage error between experimental and predicted values was found to be minimal for the mechanical properties of the weldments. Therefore, it can be concluded that ANN is a potential tool for predicting the mechanical properties of the weld formed by FSW process. Similarly, the results of Taguchi Grey relational analysis can be used to optimize the process parameters of the weld process and it can be applied extensively to ascertain the most prominent factor. The results of which indicates that rotational speed of 1,270 rpm and traverse speed of 30 mm/min are to be the optimized process parameters. The result also shows that tool rotational speed has more impact on the mechanical properties of the weld than that of traverse speed.</jats:p> </jats:sec>