TWI Digital Library

Artificial Intelligence algorithms based on machine learning approaches have recently found application in joining and welding, e.g., for the optimization of process parameters or the prediction of resulting mechanical properties. The importance of using the aforementioned approaches is exacerbated in friction stir welding, where the interaction between process input variables may be of complex nature and, thus, lead to increased experimental efforts needed to find the optimal parameter set. In this research work, a Bayesian optimization technique is therefore employed to optimize test parameters of the welding with exploring and exploiting relation between the input and output parameters during friction stir welding of dissimilar joints between aluminum alloys AA5083 H-111 and AA7020 T-651.

13th International Symposium, 21-23 May 2024, Session 10: Modelling, Paper 01