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TruckSim 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%．

Proposed in this paper is a rollover warning control strategy based on the genetic algorithm-optimized BP neural network ( GANN) for hydraulic in-wheel motor hybrid heavy truck.In the investigation,first,a rollover reference model of the heavy truck with three degrees of freedom was established,and the rollover warning indicator was selected.Next,based on the rollover reference model,an observer of rollover warning indicator was presented.Then,a new rollover warning algorithm named GANN-TTR was proposed by introducing genetic algorithm-optimized BP neural network to optimize the traditional TTR ( Time-To-Rollover) algorithm.Moreover,a system model was conducted on the Trucksim platform,a hydraulic system model was presented on the AMESim platform,and a rollover warning algorithm was achieved on the Matlab /Simulink platform.Finally,a co-simulation platform was constructed on the basis of Matlab /Simulink,AMESim and Trucksim platforms to simulate the truck under step steering and hook steering conditions,and the warning precision of the traditional TTR algorithm,the BP algorithm and the proposed GANN-TTR algorithm were compared.Simulated results show that,with the proposed GANN-TTR rollover warning algorithm,the warning precision effectively improves,and the minimum error between the revised warning curve obtained through vertical velocity and driver steering angle and the ideal warning curve is low to 5%.