Study of Optimization Strategy for Vehicle Restraint System Design 2019-01-1072
Vehicle restraint system design usually requires occupant injuries to be minimized to achieve high safety rating. The optimization formulation often involves restraint feature options (e.g., some specific features on/off) as discrete design variables, as well as continuous restraint design variables, such as airbag firing time, airbag vent size, inflator power level, etc. The optimization problem is constrained by injury criteria involve HIC, chest deflection/acceleration, neck tension/compression, etc., which ensures the vehicle meeting or exceeding all Federal Motor Vehicle Safety Standard (FMVSS) No. 208 requirements. For this type of problems, Genetic Algorithms (GA) are usually applied as optimizer because of their capability of handling discrete and continuous variables simultaneously, and jumping out regions of multiple local optima, particularly for this type of highly non-linear problems. However, the computational time for the GA based optimization is often lengthy because of the relatively slow convergence compering to derivative based algorithms, due to the nature of GA.
In this study, using the design optimization software modeFRONTIER, GA and multi-strategy optimization algorithms are compared in terms of the easiness of algorithm parameter tuning, the quality of optimal solution, and the convergence speed, on a driver side frontal impact simulation as a benchmark example. MADYMO models are developed to simulate the full frontal 90-degree rigid barrier vehicle crash scenarios at different impact speeds with a belted or an unbelted occupant. The multi-strategy optimization algorithms are sophisticated combinations of GA, gradient-based algorithms, and Response Surface Modeling (RSM).Conclusions and suggestions to design engineers are made, based on the comparison of optimization performance of aforementioned algorithms.
Guosong Li, Zhendan Xue, Ching-Hung Chuang, Kevin Pline
Ford Motor Company, ESTECO North America