CONCEPTUAL DESIGN OF TIRES USING MULTI-OBJECTIVE DESIGN EXPLORATION

The importance of the rule of simulation at a conceptual design stage is pointed out recently. The methodology using multi-objective design optimization and data mining, which is called MODE (Multi-Objective Design Exploration), is helpful to find out design knowledge. In this study, we show that the design knowledge found by MODE is fruitful for decision making at a conceptual design stage of real world example i.e. contour design of a fuel efficient tire. Our procedure of MODE consists of nonlinear FEA (Finite Element Analysis) to predict tire performance, evolutional computation to get desirable solutions including Pareto solution, and data mining to find out design knowledge. SOM (Self-Organizing Map) and decision tree are used to obtain appropriate design variables and their threshold. In this study, SOM with MLS (Moving Least Square) filter is proposed to obtain the structured smooth data of design variables in an objective function space. We can overlook the whole causation with the objective functions and the design variables using this map. On the other hand, the decision tree gives a rule to get the aimed performance balance. In order to demonstrate the validity of the knowledge, we produced prototype tires. The experimental results of them guaranteed the validity and the effectiveness of the knowledge and MODE. Furthermore, the latest fuel efficient tire, which was designed based on the obtained knowledge, was released in July 2013.