Sensitivity analysis and multiobjective optimization of automotive B-pillar crashworthiness considering mechanical uncertainty of resistance spot welds

Abstract

Unavoidable mechanical uncertainties in engineering structures significantly affect their crash performance. The mechanical uncertainty of Resistance Spot Welds (RSWs) and its effect on structure crashworthiness require a better understanding in vehicle safety and lightweight design. Therefore, this work proposed a feasible multi-objective optimization approach based on Hybrid Response Surface Method (HRSM) to explore the effect of RSWs mechanical uncertainty on B-pillar crashworthiness design. The mechanical uncertainty of RSWs was investigated experimentally and the variation range of RSWs strength under different loading conditions were quantified. Subsequently, a reasonable RSW modeling method and Body-In-White (BIW) subsystem collision simulation approach were demonstrated and validated with experimental results. A HRSM was proposed as surrogate model and the NSGA-III genetic algorithm was employed for multi-objective optimization. A multi-factor analysis based on HRSM was conducted to reveal effects of RSW strength uncertainty on structure crashworthiness design. The performance of HRSM was examined and results show that it can approximate the actual response between inputs and outputs of collision system. Compared with Kriging and Deep Neural Network (DNN), the HRSM can improve computational accuracy while ensuing efficiency. The machine-learning-based HRSM is an alternative for multiobjective optimization and subsequent robustness optimization design for vehicle crashworthiness safety. The numerical modeling method and control mechanisms of RSWs’ mechanical uncertainty on B-pillar crashworthiness can provide reference for BIW safety design.