Crash Performance Prediction and Knowledge Discovery from Crash Simulation Using Data Mining

Abstract

The extensive use of computer-aided design (CAE) is a standard approach in car crash simulation. As high-performance computation gets more available at a reasonable cost, it is becoming possible to produce huge amount of intermediate data such as the displacement of individual nodes. However, little is known about how to extract useful insights from the intermediate data. This paper proposes a data mining approach to CAE-based crash simulation. Using a dimensionality reduction technique, we demonstrate that the proposed method can automatically extract useful features from the data.