Performance evaluation of friction stir welding using machine learning approaches

The aim of the present study is to evaluate the potential of sophisticated machine learning methodologies, i.e. Gaussian process (GPR) regression, support vector machining (SVM), and multi-linear regression (MLR) for ultimate tensile strength (UTS) of friction stir welded joint. Three regression models are developed on the above methodologies. These models are projected to study the incongruity between the experimental and predicted outcomes and preferred the preeminent model according to their evaluation parameter performances. Out of 25 readings, 19 readings are selected for training models whereas remaining is used for testing models. Input process parameters consist of rotational speed (rpm), and feed rate (mm/min) whereas UTS is considered as output. Two kernel functions i.e. Pearson VII (PUK) and radial based kernel function (RBF) are used with both GPR and SVM regression. It is concluded that the GPR approach works better than SVM and MLR techniques. Therefore, GPR approach is used successfully for predicting the UTS of FS welded joint. The aim of the present study is to evaluating the friction stir welding process using sophisticated machine learning methodology, i.e. Gaussian process (GP) regression, support vector machining (SVM) and multi-linear regression (MLR). Three models are projected to study the incongruity between the experimental and predicted outcomes and preferred the preeminent model according to their evaluation parameter performances. Two kernel functions i.e. Pearson VII (PUK) and radial based kernel function (RBF) are used with both GPR and SVM regression. GPR approach works better than SVM and MLR techniques. Therefore, GPR approach is used successfully for predicting the UTS of FS welded joint. (C) 2018 The Authors. Published by Elsevier B.V.