2008 年 2 巻 1 号 p. 268-280
The hybrid method between multi-objective particle swarm optimization and adaptive range multi-objective genetic algorithm has been developed and its performance has been measured by using three test functions with noise. Moreover, it was applied to a large-scale and real-world engineering design problem. The performance measurement was carried out under the conditions of a small number of population size and generations to apply the practical problem which it needed large computational time for the evaluation. The convergence metric and the cover rate were employed as the measurement manners. Consequently, it revealed that the present hybrid method had similar performance for a simple three-dimensional test problem compared with genetic algorithm in a small number of generations. Moreover, it had the best performance for the test functions with noise. Therefore, the present hybrid method was applied to the wing design of the silent supersonic technology demonstrator. As a result, the efficient design exploration was performed and obtained 75 non-dominated solutions revealed the beneficial knowledge to decide a compromise solution.
