网联混合动力汽车跟驰场景的预测能量管理控制

Predictive energy management control for connected hybrid electric vehicles in car-following scenario

针对网联混合动力汽车跟驰场景下能量管理控制中燃油经济性和驾驶安全性综合优化问题,利用车-车及车-路通信,设计了一种基于前车速度预测-本车速度规划的预测能量管理控制策略。前车速度预测器由长短时记忆神经网络构建,神经网络内部超参数通过粒子群优化算法离线优化确定;基于预测的前车速度,求解以跟车距离、车速度、加速度及直接影响驾驶舒适性的车辆冲击度为成本函数的优化问题获得预测域内本车的速度规划;进一步利用序列二次规划算法求解车辆燃油经济性和驾驶安全性综合优化的能量管理控制问题,得到最优功率分配控制策略。多种驾驶工况下的仿真验证了所提出的预测控制策略的有效性及车辆安全跟驰下较好的燃油经济性。

A predictive energy management strategy is designed based on the preceding vehicle′s speed prediction and the ego vehicle′s speed planning by using vehicle-vehicle and vehicle-road communication to solve the comprehensive optimization problems of fuel economy and driving safety in the following scenario of connected hybrid electric vehicles.The preceding vehicle speed predictor is constructed by long short-term memory network,whose hyper-parameters are optimized by particle swarm optimization.Based on the preceding vehicle′s predictive velocity,the ego vehicle′s speed in the forecasted domain is planned to solve the optimization problem by taking the following distance,velocity,acceleration,and vehicle impact affecting driving comfort as the cost function.Furthermore,the sequential quadratic programming algorithm solves the energy management control problem of the comprehensive optimization of vehicle fuel economy and driving safety to derive the optimal power distribution control strategy.Simulations under various driving conditions verify the effectiveness of the proposed predictive control strategy and better fuel economy during the safe following.