A Mixture Design of Experiments Approach for Genetic Algorithm Tuning Applied to Multi-objective Optimization
This study applies mixture design of experiments combined with process variables in order to assess the effect of the genetic algorithm parameters in the solution of a multi-objective problem with weighted objective functions. The proposed method allows defining which combination of parameters and weights should be assigned to the objective functions in order to achieve target results. A study case of a flux cored arc welding process is presented. Four responses were optimized by using the global criterion method and three genetic algorithm parameters were analyzed. The method proved to be efficient, allowing the detection of significant interactions between the algorithm parameters and the weights for the objective functions and also the analysis of the parameters effect on the problem solution. The procedure also proved to be efficient for the definition of the optimal weights and parameters for the optimization of the welding process.
KeywordsGenetic algorithm tuning Mixture design of experiments Multi-objective optimization Global criterion method
- 14.Fernandez-Prieto, J.a., Canada-Bago, J., Gadeo-Martos, M.a., Velasco, J.R.: Optimisation of control parameters for genetic algorithms to test computer networks under realistic traffic loads. Appl. Soft Comput. J. 12(4), 1875–1883 (2012). https://doi.org/10.1016/j.asoc.2012.04.018CrossRefGoogle Scholar
- 17.Myers, R., Montgomery, D., Anderson-Cook, C.: Response Surface Methodology, 3 edn. (2009)Google Scholar
- 19.Gomes, J.H.F., Paiva, A.P., Costa, S.C., Balestrassi, P.P., Paiva, E.J.: Weighted multivariate mean square error for processes optimization: a case study on flux-cored arc welding for stainless steel claddings. Eur. J. Oper. Res. 226(3), 522–535 (2013). https://doi.org/10.1016/j.ejor.2012.11.042CrossRefGoogle Scholar