Artificial neural networks and response surface methodology as a tool for analysis the spindle torque in FSP process

Marek S. Węglowski

Institute of Welding, Bl. Czesława Str. 16-18, 44-100 Gliwice.

DOI:

https://doi.org/10.7494/cmms.2015.1.0504

Abstract:

The article presents the effect of rotational and travelling speeds and down force on the spindle torque acting on the tool in friction stir processing (FSP) process. To find a dependence combining the spindle torque acting on the tool with the rotational speed, travelling speed and the down force, the artificial neural networks (ANN) and response surface methodology (RSM) were applied. Good correlation between experimental set and model was achieved. The best results were gained for the multilayer perceptron type 3-9-1. The results obtained in artificial neural network were compared with those through response surface methodology. Based on achieved results ANN, quadratic and linear models can be recommended to predict the value of spindle torque acting on the tool during FSP process carry out on alloy AlSi9Mg.

Cite as:

Węglowski, M. (2015). Artificial neural networks and response surface methodology as a tool for analysis the spindle torque in FSP process. Computer Methods in Materials Science, 15(1), 65-70. https://doi.org/10.7494/cmms.2015.1.0504

Article (PDF):