Development of the Robust Design Method for Crash Analysis

- Visualization of Conditions for Good Quality by Using Pattern Recognition and Machine Learning -

Abstract

This paper describes a methodology for robust design in crash analysis. We organized a series of robust design processes for crash analysis and showed that it is possible to classify/predict deformation patterns, judge whether they are good or not, identify contributing parameters by applying statistical/machine learning methods in each process. By using a prediction model as a surrogate model, we achieved an extensive search in the design space in a short time.