Yusuke Kageyama

Response surface method using hierarchical clustering

In the field of the numerical analysis, response surface method is one of the most practical optimization methods. The optimum design variable is obtained from response surface approximation under constraint condition in the optimization using response surface method. The authors have proposed hierarchized response surface method (H-RSM) using hierarchical clustering analysis in order to make high-precise response surface approximations. Response surface approximation is made in the every hierarchy cluster by the classification of the design pattern. Therefore, hierarchized response surface approximation is high accuracy because of approximate by the local pattern.

DesignSpace Embossing Method for Basic Design Plans of a Vehicle Suspension System

In general, vehicle structure design processes, like a suspension system, are performed in three stages, the concept design, the basic design, and the detailed design processes. There is incessant demand to shorten the design time required for all three of these stages. An abundance of CAE technologies have been developed for the detailed design process, serving as powerful tools to improve design quality and efficiency. However, the concept design process is still performed using classic approaches, such as experience and know-how. By implementing a quantitative analysis method to evaluate design information, design engineers can improve their design concept or basic design plans. Unfortunately, the early design stages involve numerous design factors and extensive design ranges, making it difficult for design engineers to fully understand the extent of their design capacity. This paper introduces a new approach called the Design Space Embossing Method. This method uses CAE technology to efficiently search for promising design spaces for complicated basic design processes. Having this information will enable engineers to visually comprehend the design capabilities in the early phases of the design process. When this method is implemented, the number of necessary samples can be drastically reduced while providing a satisfactory result for even complicated multiple maximum problems. In this paper, the proposed method has been applied to the basic design process of a vehicle suspension system. The results show that the design spaces generated by this method could help design engineers understand the basic behavior of the system.© 2004 ASME