基于自适应时域MPC的平行泊车跟踪控制研究

Research on parallel parking tracking control based on adaptive time domain MPC

针对固定预测时域模型预测控制算法(Model Predictive Control,MPC)在泊车时难以满足大转向工况下路径跟踪精准性要求的问题,提出一种基于自适应预测时域MPC的泊车路径跟踪控制算法。根据车辆运动学模型以前轮转角与车速为控制目标设计MPC泊车跟踪控制器,并通过模糊控制算法计算当前最优预测时域,实现对预测时域的自适应优化,从而降低横向与航向误差;CarSim/Simulink仿真试验表明,在平行泊车工况下,基于自适应预测时域MPC的泊车路径跟踪控制算法对期望轨迹与车身航向角的跟踪效果优于传统固定时域MPC算法。

In view of the difficulty of traditional fixed time domain model predictive control(MPC) to meet the requirement of path tracking accuracy under large turn conditions during parking,a parking path tracking control algorithm based on adaptive time domain MPC was proposed.Firstly,the MPC parking tracking controller with front wheel angle and vehicle speed as the control targets was designed based on vehicle kinematics model.Then,the optimal prediction time domain was calculated by fuzzy control algorithm to achieve adaptive optimization of prediction time domain to reduce lateral and heading errors.The CarSim/Simulink simulation results showed that under parallel parking conditions,the parking path tracking control algorithm based on adaptive predictive time-domain MPC was better than the traditional time-domain MPC algorithm in tracking the expected trajectory and vehicle heading angle.