Optimization of airbag release algorithms using evolutionary strategies

Increasing cost pressure for airbag systems brings up the need for automated design of release algorithms. Simple threshold release is not possible when using modern single point sensing, i.e. crash sensing in the passenger compartment. This paper presents a method for tuning airbag release algorithms using evolutionary strategies for optimization. A quality function for evaluating algorithm performance is introduced and a modified evolution strategy, based on biologic evolution theory, is used to optimize the continuous parameters of a simple release algorithm. Results are presented for a typical vehicle that confirm the feasibility of evolution strategies for optimization of strongly ill-conditioned parametric optimization problems.