Self-organizing map of Pareto solutions obtained from multiobjective supersonic wing design
Shigeru Obayashi, Daisuke Sasaki · 2002
Self-Organizing Map (SOM) has been applied to analyze 766 Pareto solutions obtained from the four-objective aerodynamic optimization of supersonic wings using Evolutionary Algorithms. Threedimensional Pareto front (tradeoff surface) is mapped onto the two-dimensional SOM where global tradeoffs are successfully visualized. Furthermore, from the clusters obtained in the SOM, the design variables are mapped onto another SOM. This leads to clusters of design variables which indicate the relative importance of design variables and their interactions. SOM is confirmed to be a versatile datamining tool for aeronautical engineering.