Hierarchical Multiobjective Optimization Methods for Deeper Understanding of Design Solutions and Breakthrough for Optimum Design Solutions

1. Abstract In order to make most effective use of an optimization problem formulation within actual design scenarios, the following three goals should be met: (1) acquisition of optimum solutions from wider viewpoints, (2) deepest possible understanding of the achieved optimum design solutions, and (3) breakthrough of the optimum design solutions. This study presents a product design optimization method that can accomplish these goals. A multiobjective optimization problem having a variety of performance characteristics such as accuracy, operating efficiency, manufacturing cost, and energy consumption during use, is reconstructed as a set of hierarchical multiobjective problems comprised of decomposed characteristics that are simplified or idealized characteristics of higher characteristics. In the hierarchical optimization structure, the characteristics existing at the lower levels are grouped into optimization units based on the input and output relationships among the characteristics. The optimizations are processed sequentially from the bottom hierarchical level to the higher levels, and at each level, Pareto optimum solutions are transferred to the next highest level. Finally, the Pareto optimum solution set for the problem’s performance characteristics at the highest level is achieved. The proposed method offers the following advantages: 1) effective obtainment of the optimum solution from a global viewpoint for product designs having complex evaluation characteristics; 2) effective evaluation of the validity of the optimization formulation, based on hierarchical optimization strategies, and effective reformulation of the optimization problem; and 3) effective realization of improved design proposals, based on detailed examination of hierarchical optimization structures when seeking breakthrough improvements in initially obtained optimized solutions. 2.