Machine learning enhanced optimisation of crash box design for crashworthiness analysis

Abstract The primary goal of the current study is to optimise crash box designs for automobiles. Machine learning (ML) techniques are used build an intelligent and reliable ML framework. With help this framework, design can be optimised crashworthiness analysis. optimisation resource‐intensive due its intricate geometric design, use a variety materials, extensive dynamic simulations determine ideal structural parameters through simulations. A reinforcement learning‐based (RL) optimization technique developed reason. built with nonlinear first‐order shear deformation shell elements provide necessary simulation data, elastoplastic material model impact process. RL agents tuned using finite element (FE) synthetic data generated by generative adversarial network (GAN) select optimal vehicle To estimate correct parameter required fulfil specified metrics, agent learns optimises automatically. This method effective in terms resources could helpful early stages development.