基于DDPG算法的混动铲运机能量管理策略研究

Energy management strategy of hybrid LHD based on DDPG algorithm

目前混合动力铲运机能量管理策略通常为基于规则的策略,为了进一步优化铲运机的燃油经济性与系统能量稳定性,首次将深度确定性策略梯度(DDPG)算法应用于混合动力铲运机的能量管理策略,针对铲运机高频率循环、短时大功率输出的特点,协调控制系统能量流动,满足整车功率需求的同时,对发动机的工作区域进行“削峰填谷”,在减少油耗的同时保持超级电容SOC的控制目标下,设计奖励函数。利用基于DDPG算法的能量管理策略在铲运机不同循环工况下进行了仿真实验。结果表明,相比于传统的基于规则的能量管理策略,利用基于DDPG算法的能量管理策略,在两种工况下节油效果分别达到了8.36%与8.41%,同时更充分地利用到了超级电容的能量缓冲作用。

Currently,the energy management strategy of hybrid power LHDs is usually based on rule-based strategies.To further optimize the fuel economy and the system energy stability of LHDs,this paper applies the deep deterministic policy gradient(DDPG)algorithm for the first time to the energy management strategy of hybrid power LHDs.Based on the characteristics of the high-frequency cycle and short-time large-power output of LHDs,the coordination control of energy flow in the control system is designed to meet the power requirements of the whole vehicle,while“cutting peaks and filling valleys”in the working area of the engine.To reduce fuel consumption and maintain the control target of the SOC of the supercapacitor,a reward function is designed.Simulation experiments under different operating conditions of the LHD are conducted using the energy management strategy based on the DDPG algorithm.The results show that compared with the traditional rule-based energy management strategy,the energy-saving effect of using the energy management strategy based on the DDPG algorithm reaches 8.36%and 8.41%respectively under two different operating conditions,and the energy buffering effect of the supercapacitor is more fully utilized.