基于BP神经网络的气囊点火算法模型

通过建立完整的车体-乘员-约束系统仿真分析模型,将BP神经网络模块嵌入到仿真分析流程中,在产品开发阶段实现了气囊点火算法的验证。本算法模型在对网络进行训练的基础上,以不同碰撞速度的正面刚性墙碰撞分析模型为研究对象,建立了车体加速度与乘员位移之间的数学模型,实现了随机输入车体加速度曲线即可获取乘员位移曲线,计算结果与仿真分析结果吻合,为进一步研究气囊点火优化算法奠定了基础。

Setup of BP ANN-based crash sensing algorithm

A vehicle crash sensing algorithm combining simulation model and BP artificial neural network module was developed and validated at vehicle developing stage.The simulation model is composed of vehicle,occupant and restraint system(VOR).The neural network training of multi-velocity frontal crashing into fully rigid wall was conducted first.Then the mapping model of vehicle acceleration and occupant displacement was built.Using this mapping model,the random vehicle acceleration input can be easily translated to the occupant displacement.The calculation results are in good agreement with the simulation results.It provides the basis for further development of optimized algorithm of vehicle crash sensing system.