Neural network modeling of pulsed-laser weld pool shapes in aluminum alloy welds

Citations Over Time

Abstract

A model was developed to predict the weld pool shape in pulsed Nd:YAG laser welds of aluminum alloy 5754. The model utilized neural network analysis to relate the weld process conditions to four pool shape parameters: penetration, width, width at half-penetration, and cross-sectional area. The model development involved the identification of the input (process) variables, the desired output (shape) variables, and the optimal neural network architecture. The latter was influenced by the number of defined inputs and outputs as well as the amount of data that was available for training the network. After appropriate training, the best network was identified and was used to predict the weld shape. A routine to convert the shape parameters into predicted weld profiles was also developed. This routine was based on the actual experimental weld profiles and did not impose an artificial analytical function to describe the weld profile. The neural network model was tested on experimental welds. The model predictions were excellent. It was found that the predicted shapes were within the experimental variations that were found along the length of the welds (due to the pulsed nature of the weld power) and the reproducibility of welds made under nominally identical conditions.

Related Papers

• → Natural head posture--considerations of reproducibility(2002)69 cited
• → Short- and long-term reproducibility of radioisotopic examination of gastric emptying(1990)20 cited
• The reproducibility of reported height and body weight in repeated questionnaire surveys.(1995)
• The reproducibility of scores from three video formats.(1988)
• Comparisons of probing depth measurements with Florida probe and conventional periodontal probe(2005)