Artykuł

Vol: 33 , Nr: 4 , Art.: 5
Tytuł / Title:
Prediction of spindle torque acting on the tool during FSP using neural networks and RSM
Autorzy / Authors:
WĘGLOWSKI Marek Stanisław
Instytut Spawalnictwa w Gliwicach
Słowa kluczowe / key words:
friction stir processing, neural networks, response surface methodology, aluminium alloys
Streszczenie / Summary:

An artificial neural network (ANN) and response surface methodology (RSM) models were
developed for the analysis and simulation of the correlation between parameters of the friction stir
processing process (FSP) and the torque acting on a tool during modification of cast aluminium
alloy AlSi9Mg. The input parameters were: rotational speed, travelling speed and down force. The
output parameter of the models was the torque of the spindle. Good correlation between the experimental
set and the model was achieved. The best results were obtained for the multilayer perceptron
type 3-6-1. Results obtained in artificial neural network were compared with those through
response surface methodology. Based on results achieved, ANN and linear model can be recommended
to predict the spindle torque value acting on the tool during FSP process carried out on
alloy AlSi9Mg.