Artificial Neural Network Modelling for Polyethylene FSSW Parameters

Abstract

In friction stir spot welding (FSSW) process the welding parameters (the tool rotational speed, the tool plunge depth and the stirring time) effect the nugget formation in high density polyethylene (HDPE) sheets. The size and the microstructure of the nugget determine the resistance of the joint to outer forces. The optimization of these parameters are vital to obtain high quality welds. Feed forward back propagation artificial neural network models are developed for the optimization of the FSSW parameters for HDPE sheets. Input variables of these models are the tool rotation speed (rpm), the plunge depth (mm) and the stirring time (s) affect to lap-shear fracture load (N) output. Prediction performance of 6 models in different specifications are compared. These models differ in terms of the training dataset used (80%-100%) and the number of neurons (5-10-20) in hidden layer. Best prediction performances are obtained with using 20 neurons in hidden layer in both training dataset. There is a good agreement between developed models' predictions and the experimental data.