A study on the extraction method of critical areas in the design shape of structures using sparse modeling.

Abstract

By using sparse modeling to formulate the relationship between the data of the design image of an automobile body frame and the amount of frame function, we developed a method to extract the features of the image that are important for improving the function. The important areas of the cross-sectional shape visualized by the method and the design guidelines for the cross-sectional shape derived from the visualization are consistent with the findings obtained in previous studies. In addition, we were able to obtain new findings that were not previously available. In other words, this method has the potential to elucidate unknown important factors that are not assumed by designers and to derive general purpose design guidelines by utilizing a large amount of design image data obtained in shape optimization.

Keywords

• machine learning
• sparse modeling
• shape optimization
• hollow flame
• bending strength
• buckling