Keiji Sumitani

Research on aerodynamic noise around automobiles

Abstract Along with improved quietness in vehicles in recent years, aerodynamic noise reduction has been demanded strongly. Actions concerning the noise generating source analysis are important. To study the direction for aerodynamic noise reduction, we have theoretically analyzed the characteristics of the sources of aerodynamic noise. As a result, we have succeeded in clarifying the relationships between the flow patterns and noise properties; the resulting aerodynamic noise reduction has been confirmed both in wind tunnel test and on road test.

Application of the neural network to fluctuating aerodynamic noise — Relationship between body shape and sensory evaluation during running

Abstract Because of improved quietness in automobiles, fluctuating aerodynamic noise has become discernible. Although evaluating and reducing noise in the early stage of automobile development (planning or pre-prototype stage) is strongly expected, there are many difficulties in reducing noise by use of computational fluid dynamics or experimental schemes. We therefore adopted an alternative approach, neural network, and succeeded in evaluating and reducing the noise. We verified the effects of neural network by experiment of flow visualization. As a result we propose a new technique in evaluating and designing automobiles by use of neural network.