基于动态规划算法的增程式电动汽车能量管理策略优化
Energy management strategy optimization of extended-range electric vehicle based on dynamic programming

提出了一种动态规划改进算法,根据约束条件确定未来可达状态序列,通过计算离散状态点间的转移代价,在保证求解精度的同时,降低了离线优化计算量; 利用改进动态规划算法设计了增程式电动汽车能量管理策略,根据能量管理优化问题特点,建立了动力系统模型和适用于全局优化求解的系统状态方程,并确定了以动力电池荷电状态为系统状态量和增程器发电功率为系统控制量; 在迭代计算过程中,将发动机燃油费用和动力电池电能费用之和作为目标函数,构建了基于北京主干道不同行驶里程仿真工况,得到了驱动电机需求功率最优分配结果; 提取了增程器启停状态与动力电池荷电状态和驱动电机需求功率二者之间的控制规则,利用最小二乘法对增程器功率分流比与驱动电机需求功率的分布规律进行拟合,建立了基于优化规则的能量管理策略。仿真结果表明:对于行驶里程为100 km的仿真工况,动态规划改进算法计算时间为7 239 s,与经典动态规划算法相比计算效率提高了78.2%; 基于优化规则的能量管理策略能够获得类似动态规划改进算法的控制效果,2种控制策略的动力电池荷电状态误差小于2.5%; 相比实车电能消耗-电能维持型控制策略,基于优化规则的控制策略能够使整车经济性提高5.4%,使燃油经济性提高7.9%。
A modified dynamic programming algorithm was proposed. A future reachable states array was determined based on the constraints. The transfer costs among discretized states were calculated to guarantee the solving accuracy and reduce the off-line calculation burden. An energy management strategy for an extended-range electric vehicle was designed using a modified dynamic programming algorithm. Based on the energy management problem features, a dynamic system model was constructed, a system state equation for solving global optimization problems was determined, the battery state of charge(SOC)was selected as a state variable, and the extender output power was selected as a control variable. During the iterative calculation process, the cost of engine fuel and the battery energy were added in the objective function. Different driving-distance simulation cycles were constructed based on the Beijing arterial road cycle to obtain the optimal distribution result of required motor power. The control rules of extender start-stop corresponding to the battery SOC and required motor power were extracted, the distributed regulation between extender power split ratio and required power was fitted using the least square method, and the energy management strategy based on the optimal rules was established. Simulation result indicates that for the 100 km driving distance simulation cycle, the calculation time of the modified dynamic programming algorithm is 7 239 s, and the calculation efficiency improves by 78.2% compared to the classic dynamic programming algorithm. The optimal rule-based energy-management strategy has a similar control performance with the modified dynamic programming algorithm. The SOC errors of the two control strategies are within 2.5%. Compared to the charging deplete/charging sustain control strategy, the optimal rule-based control strategy improves the economy performance by approximately 5.4% and the fuel economy by approximately 7.9%. 3 tabs, 13 figs, 26 refs