基于遗传算法优化的BP神经网络侧翻预警算法

发布时间：2018-02-09 12:25

本文关键词： 重型车辆侧翻预警算法预警精度遗传算法 BP神经网络　出处：《华南理工大学学报(自然科学版)》2017年02期 　论文类型：期刊论文

【摘要】：针对轮毂液压混合动力重型商用车,引入基于遗传算法优化的BP神经网络算法建立侧翻预警控制策略.首先,建立重型车辆3自由度侧翻参考模型,选取车辆侧翻预警算法的侧翻指标,并结合参考模型建立侧翻指标观测器;然后,在传统TTR(Time-To-Rollover)侧翻预警算法研究的基础上,引入遗传算法优化的BP神经网络(GANN)对传统的TTR预警算法进行优化,建立基于GANN-TTR的侧翻预警算法;最后,利用Truck Sim仿真软件建立整车模型,利用AMESim仿真软件建立轮毂液压系统模型,在Matlab/Simulink环境下实现侧翻预警算法,并通过Matlab/Simulink、Trucksim和AMESim三软件搭建联合仿真平台,选取阶跃转向和鱼钩转向两种典型工况进行仿真,对比传统TTR、传统BP神经网络以及基于GANN-TTR的侧翻预警算法的预警精度.仿真结果表明,基于GANNTTR的侧翻预警算法能够有效提高预警精度,通过方向盘转角和纵向车速进行算法修正后得到的曲线与理想预警曲线误差最小达5%.
[Abstract]:The BP neural network algorithm based on genetic algorithm is introduced to establish the early warning control strategy for the heavy commercial vehicle with hydraulic hybrid electric vehicle hub. Firstly, the 3 DOF rollover reference model of heavy duty vehicle is established. Selecting the rollover index of vehicle rollover early warning algorithm and combining with the reference model to establish the rollover index observer. Then, based on the traditional TTRT Time-To-Rollover early warning algorithm, Genetic algorithm optimized BP neural network was introduced to optimize the traditional TTR early warning algorithm, and the rollover early warning algorithm based on GANN-TTR was established. Finally, the whole vehicle model was established by using Truck Sim simulation software. The hydraulic system model of wheel hub is established by using AMESim simulation software, and the early warning algorithm of rollover is realized under the environment of Matlab/Simulink. Through Matlab / Simulink Trucksim and AMESim software, a joint simulation platform is built, and two typical working conditions, step steering and fishhook steering, are selected to simulate. Compared with the traditional TTR-based BP neural network and the GANN-TTR based rollover early warning algorithm, the simulation results show that the rollover early warning algorithm based on GANNTTR can effectively improve the early warning accuracy. The minimum error between the curve and the ideal warning curve is up to 5 by the algorithm correction of steering wheel angle and longitudinal speed.
【作者单位】：吉林大学汽车仿真与控制国家重点实验室;一汽解放青岛汽车有限公司;
【基金】：国家自然科学基金资助项目(51575221,51675214) 吉林大学研究生创新研究项目(2016083)~~
【分类号】：U461.6;TP183

【相似文献】