Prediction of Process Parameters of Friction Stir Welding Using Artificial Neural Network

Gosavi Swapnil Vijaygir; Jaybhaye Maheshwar Dinkar

Abstract

In this paper an artificial neural network (ANN) approach is used to predict the process parameters of friction stir welding (FSW). Initially, the experiments are conducted using the design of experiment (DoE) approach on FSW using L27 orthogonal array. The experiments are conducted using speed, feed, and tool tilt angle as input parameters for DoE and tensile strength, hardness, and ductility as output. ANN is created having 25 neurons and one hidden layer, and an input and output layer in the Matlab® environment. ANN utilizes 75% data for training the model, 15% of the data for validation, and 10% of the data for testing the trained model. ANN predicts the output parameters with R values for training, validation, and test datasets as 0.99884, 0.99488, and 0.98428 while the overall R value is 0.99751. The R values are quite high for training, validation, and test datasets. The ANN is further implemented to predict the FSW parameters. It is observed that ANN predicts the FSW parameters with high accuracy. Thus, ANN can be used for predicting the ANN parameters with accuracy.