MULTIOBJECTIVE OPTIMISATION IN INDUSTRIAL DESIGN
Editor: Marjanovic D.
Author: Cappello F., Marchetto M.
Section: DECISION MAKING WORKSHOP
Page(s): 1383 - 1388
Since the mid-1980s, there has been a growing interest in solving multiobjective optimization problems using genetic algorithms because they process a set of solutions in parallel allowing to obtain the Pareto Frontier through a unique run. We propose a new genetic algorithm for multiobjective optimization, named SPLSDCAS, which uses a geographical selection schema integrated with an innovative fitness assignment and an Additive-Sharing technique. The results obtained on a simple test as well as on a complex design problem, the multiobjective shape optimization of a lenticular wheel, suggest that SPLSDCAS can be very effective in sampling the entire trade-off surface, also outperforming the other algorithms involved in the comparison.