基于PID与Q-Learning的混合动力汽车队列分层控制

Hierarchical control of hybrid electric vehicle platooning based on PID and Q-Learning algorithm

提出一种基于PID与Q-Learning的混合动力汽车队列分层控制策略。上层控制器基于车-车通信获得队列中前车的速度和位置信息,采用PID控制器实现队列的纵向控制并获得后车的目标车速;下层控制器根据该目标车速采用Q-Learning进行混合动力汽车队列的能量管理。仿真结果表明:上层控制队列平均车间距保持在14 m左右,确保良好的行驶安全性;下层控制队列平均百公里油耗比DP策略增加了2.57%,离线计算时间减少了23%。该策略在保持与DP基本相同的燃油经济性下,不仅能适应随机工况,也能在线实现。

A hierarchical control strategy based on PID and Q-Learning algorithm for hybrid electric vehicle platooning.In the upper-level controller is proposed in this paper,the speed and position information of the preceding vehicle in the platooning are obtained based on vehicle-vehicle communication,the PID controller to realize the longitudinal control is adopt and the target speed of the following vehicle is obtained.In the lower-level controller,Q-Learning is adopted to distribute the energy of the hybrid vehicle platooning according to the target speed.The simulation results show that the average vehicle spacing in the upper control is maintained at about 14 m,which can ensure good driving safety.The average fuel consumption per 100 kilometers in the lower control is only 2.57% higher than that of DP,and the offline calculation time is reduced by 23%.This strategy can not only adapt to random working condition,but can also be implemented online,which maintains basically the same fuel economy as DP.