Research on Pedestrian Head Injury Prediction Based on Machine Learning

In the development of automotive safety performance, predicting early results through CAE (Computer Aided Engineering) simulation models is a crucial stage. However, the efficiency of the CAE simulation development phase is low, resulting in a lengthy overall cycle. To better accommodate the increasingly shortened vehicle development cycles, this paper thoroughly explores the application of machine learning algorithms in predicting head injury in pedestrian protection. Based on 25 different vehicle models, key features at impact points were defined according to finite element simulation models and results, and a head injury database was constructed. Various machine learning algorithms were then used in a comparative study to identify the optimal predictive model. In tests with three randomly selected new vehicle models, the R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> values were approximately 0.7, with a maximum of 0.7456, and prediction times were within seconds, greatly enhancing prediction efficiency. This demonstrates the effectiveness of machine learning algorithms in safety performance development. Finally, the paper outlines further research directions, providing new perspectives and methods for head injury prediction in pedestrian protection and automotive safety performance development.