Optimization of crash structures
The numerical optimization of vehicle structures involves complex, highly non-linear response systems that have to be analyzed automatically and evaluated efficiently due to the high computing time involved. At the institute, artificial intelligence methods are investigated and adapted for crash optimization. The aim here is to automatically identify the most efficient method for an optimization task and quickly arrive at an initial optimal solution proposal.
Optimizing modern vehicle structures to improve crash safety is a highly complex task. It is based on detailed crash simulations in which the physical processes of an accident are modelled virtually. These simulations are highly non-linear, as material behavior, structural deformations and interactions between occupants, vehicle and environment have to be considered simultaneously. The computing effort is correspondingly high, which makes efficient evaluation and optimization necessary.
At the Institute of Vehicle Concepts, artificial intelligence (AI) methods are therefore being researched and further developed specifically for crash optimization. AI algorithms - including machine learning, surrogate models and Bayesian optimization - learn from derived simulation data in order to recognize patterns and make predictions. This enables them to automatically identify optimized parameter sets for a specific target function and deliver an initial optimal solution proposal in a short space of time.
This intelligent combination of numerical crash simulations and data-driven AI allows development times to be significantly reduced, innovative vehicle concepts to be evaluated more quickly and new safety strategies to be developed in a targeted manner - an important contribution to safe mobility in the future.