Neural Computing for Determining the Accuracy of Ultimate Tensile Strength of Friction Stir Welded Joints by Using Various Activation Functions

Abstract

Activation functions in a particular Artificial Neural Network (ANN) architecture plays a vital role. It imparts non-linear properties to our Neural Networks. There is a complicated and Non-linear complex functional mapping between the inputs and response variable. In our present work, we have focussed on the accuracy of the UTS of the dissimilar Friction Stir Welded joints obtained by the training and testing the Artificial Neural Network architecture on Sigmoid activation function, Rectified Linear Unit (ReLu) activation function and Hyperbolic tangent activation function. Tool Rotational Speed (rpm), Welding speed (mm/min) are the inputs and Ultimate Tensile Strength (MPa) is the output in our neural network architecture.

Keywords: Artificial Neural Network, Friction Stir Welding, Activation Functions, Google Colaboratory

JEL Classification: C53, C63