Multi-surrogate-assisted metaheuristics for crashworthiness optimisation

This work proposes a multi-surrogate-assisted optimisation and performance investigation of several newly developed metaheuristics (MHs) for the optimisation of vehicle crashworthiness. The optimisation problem for car crashworthiness is posed to find the shape and size of a crash box while the objective function is to maximise the total energy absorption subject to a mass constraint. Two main numerical experiments are conducted. Firstly, the performance of different surrogate models along with the proposed multi-surrogate model is investigated. Secondly, several MHs are applied to tackle the proposed crashworthiness optimisation problem by employing the best obtained surrogate model. The results reveal that the proposed multi-surrogate model is the best performer. Among the several MHs used in this study, sine cosine algorithm is the best algorithm for the proposed multi-surrogate model. Based on this study, the application of the proposed multi-surrogate model is better than using one particular traditional surrogate model, especially for constrained optimisation.