An ANN and Taguchi algorithms integrated approach to the optimization of CO 2 laser welding

Abstract

Nowadays several numerical methods are widely used for either modelling or optimizing the performance of the manufacturing technologies. That has been advanced due to the large diffusion of the personal computer and the numerical algorithms. The knowledge of those methods and the ability in integrating their functions can make both the manufacturing engineer and the researcher ace their duties. In this paper, two of those methods have been employed, the backpropagation artificial neural network and the Taguchi approach to the design of the experiment. They were applied to find out the optimum levels of the welding speed, the laser power and the focal position for CO 2 keyhole laser welding of medium carbon steel butt weld. The optimal solution is valid in the ranges of the welding parameters that were used for training the neural networks. Extrapolation over those limits would restrict the applicability of the found solution. The proposed approach would be extendable to other keyhole laser welding processes for different materials and joint geometries.