Overview
1) New Multiobjective design exploration framework that can efficiently extract knowledge useful for design in relation to large-scale multiobjective design optimization problems
To develop new multiobjective design exploration framework that can extract knowledge useful in actual design work related to large-scale multiobjective design optimization problems, we have two specific goals. First, we develop innovative multiobjective design optimization algorithms that uses large-scale computer resources of the “K computer” class to efficiently solve large-scale multiobjective design optimization problems and to obtain Pareto-optimal solutions. Second, we develop innovative data mining techniques to extract useful knowledge for design from the large amount of data contained in Pareto-optimal solutions of large-scale multiobjective design optimization problems.
2) Demonstration of multiobjective design exploration in actual design optimization problems in industries
To demonstrate feasibility and effectiveness of the multiobjective design exploration technology developed in this R&D topic, we will apply it to design and development problems that are actually faced by research institutions and companies.
Software to be used on the “K computer”:
Multiobjective evolutionary computation software (newly developed), LANS3D, software developed in other R&D topics, etc.
Design exploration in a small-scale multiobjective design optimization problem