2015 年 9 巻 5 号 p. JAMDSM0060
This study presents a new approach, group inching fortification (GIF) method, to deal with multiobjective optimization problems found in various mechanical designs. The GIF method is exemplified by a four-factor porous air bearing design. In the GIF method the initial group of designs in Pareto rank 1 is used as the basis to inch up the formation of Pareto solution set, which is fortified over the search process by uniting superior or non-dominated solutions from base-point exploration moves. In this study, a comparison of the GIF method with genetic algorithm (GA) and hyper-cube dividing method (HDM) for the same air bearing design is presented. The results show that the Pareto solution set obtained by the GIF method has more design selections with a wider coverage (breadth). Equally important, the number of objective-function calls required (179, 736, and 3600 for GIF, HDM, and GA, respectively) in the GIF method is significantly reduced. In this study, the GIF method is suggested to be terminated when all the designs are non-dominated by each other, a criterion which is very difficult to achieve by using the GA and HDM. This study proposes an effective design tool which is easy-to-implement for solving the problems with multiple objectives.