基于Elman神经网络的车轮滑移率跟踪控制

Wheel slip tracking control of vehicle based on Elman neural network

针对汽车高速紧急换道避障系统对快速、精确和稳定的车轮滑移率跟踪控制的需求,基于离散滑模变结构控制方法,设计了对系统不确定性具有强鲁棒性特征的车轮滑移率离散积分滑模跟踪控制器,并利用一步延迟估计方法在线估计和补偿系统不确定性,从而抑制了抖振现象.同时,利用Elman神经网络的时间序列预测能力构建了车轮目标滑移率预测模型,用于预估车轮滑移率离散积分滑模跟踪控制器包含的下一个采样时刻车轮目标滑移率,并通过粒子群优化算法实时修正车轮目标滑移率预测模型的未知权重来提高其预估精度.最后,对提出的车轮滑移率离散积分滑模跟踪控制器的可行性和有效性进行仿真验证.

In order to meet the requirement of the high-speed emergency lane-changing obstacle avoidance system for the rapid,accurate and stable tracking control of wheel slip,a wheel slip discrete-time integral sliding mode tracking controller with strong robustness against the system uncertainty was proposed based on discrete-time sliding mode variable structure control method,and one-step delay estimation method was used to on-line estimate and compensate the system uncertainty to suppress the chattering phenomenon.Meanwhile,a desired wheel slip prediction model was constructed based on Elman neural network to predict the desired wheel slip at the next sampling tim e,which was included in the wheel slip discrete-time integral sliding mode tracking controller,and particle swarm optimization algorithm was used to modify the unknown weight of the desired wheel slip prediction model to improve the prediction accuracy.Finally,the feasibility and effectiveness of the proposed wheel slip discrete-time integral sliding mode tracking controller were verified by simulation.