Energy Management Strategy for Hybrid Electric Vehicle Integrating Ultra-Capacitor and Power Battery
As hybrid electric bus has the characteristics of long mileage and heavy load,it is of great significance to design a reason-able energy management strategy to reduce the overall energy loss of the system.In this paper,an optimal energy management strate-gy for hybrid electric bus,whose power system integrates power battery and ultra-capacitor,is designed based on a deep reinforce-ment learning method,namely deep deterministic policy gradient.Compared with the traditional reinforcement learning control strat-egy,the proposed strategy avoids the discretization of continuous state and action variables in the solution process by introducing neu-ral network,and improves the control accuracy.In addition,dynamic programming algorithm is used to solve the optimal control strategy in advance,and the strategy experience is included in the experience pool,so that the optimal network parameters can be ob-tained quickly according to the optimal experience in the subsequent network parameter training,which accelerates the convergence speed of the network and improves the optimality of the control strategy.The results show that the proposed control strategy can reduce the overall energy consumption by 10.3％compared with the traditional Q-learning algorithm;Compared with the optimal control strategy based on dynamic programming,the energy consumption is only increased by 2.05％,which proves the optimality of the pro-posed control strategy.
Hybrid Electric VehicleUltra-CapacitorPower BatteryEnergy ManagementReinforcement Learning
万方数据
维普
