3rd ICAI 2024

International Conference on Automotive Industry 2024

Mladá Boleslav, Czech Republic

Optimization of Airbag Design by AI Support Jan Korouš 1 , Martin Kubíček 2 Škoda Auto a.s. 1 tř. Václava Klementa 869, Mladá Boleslav, 293 01 Czech Republic UptimAI s.r.o. 2 Hybernská 1613/38, Praha 1, 110 00 Czech Republic e-mail: jan.korous@skoda-auto.cz 1 , martin.kubicek@uptim.ai 2

Abstract Passenger safety is a very important and challenging topic in the automotive industry. The key components for reaching safety targets are the restrain systems like seat belts and airbags. Their development is becoming more complex due to increasing safety requirements. The current trend is to use virtual development methods (FEM simulations) massively to reduce the number of real tests. As many simulations must be performed to obtain an optimized restrain system, an idea was arisen to use an AI approach for this problem. From many topics, a side pole crash was selected with the aim of a side airbag design optimization the reduce the crash test dummy loading. The airbag geometry was parametrized, and these parameters together were used as the input parameters for surrogate model, which describes the dependency between the input and output values. The surrogate model was prepared using an algorithm which used existing simulation results or prepared new additional simulation runs. This procedure found an optimized solution which reduces the dummy loading. Keywords: airbag design, artificial intelligence, FEM, surrogate models, passenger safety JEL Classification: R41, L74, Q55 1. Introduction The development of a car is a complex activity in which many requirements must be considered. If we focus on the development of the car body, one of the challenging tasks is to comply with the regulations for the car safety in the case of traffic accident. These regulations are constantly tightening and as well as their scope is extending. It has been resulted in the state that the current cars have to withstand much more severe tests today than 10 years ago. At the same time, great importance is placed on weight reduction, both for economic and environmental reasons. It is obvious that these requirements are contradictory and meeting them requires the massive deployment of simulation methods. These allow the examination of many variants of the car design and the assessment of the most suitable solution. Real tests are already carried out on an optimised design state and serve to confirm and validate simulation results, or to detect a problem in areas where the simulations have not yet

93