基于MPC算法的PHEV发动机转矩预测控制策略研究

Investigation of MPC-based Engine Torque Prediction Control Strategy for a Parallel Hybrid Electric Vehicle

针对一款并联式混合动力系统的结构特点,以提高其燃油经济性为目标,开展局部工况下发动机转矩预测的控制策略研究。采用规则与优化算法相结合的方法,在模式切换层上选定电池荷电状态(SOC)与表示负荷大小的加速踏板开度为控制参数。当系统进入联合驱动模式,采用优化算法,即以燃油消耗为优化目标价值函数,发动机转矩为控制变量,基于模型预测控制(MPC)算法建立局部工况时域预测滚动优化策略。为确定MPC的预测时域和权衡因子两个关键参数,设计了相应的正交仿真实验方法。确定预测时域改变权衡因子,选用预测时域S为3s,权衡因子分别为0.0001、0.00001、0.000001。确定权衡因子改变预测时域,则权衡因子选用0.0001,预测时域S分别为3s、5s、10s。结果表明,因子λ为0.0001,预测时域S为3s为最佳组合,折合百公里等效综合油耗为6.179 L/100 km。另外,选用最佳组合参数所制定的策略与普通规则策略相比,其燃油经济性提升9.3%。通过以上对比实验,进一步说明基于MPC算法的局部工况发动机转矩预测控制策略可有效改进燃油经济性。

To improve the fuel economy of a parallel hybrid power vehicle,the torque distribution strategy based on engine prediction is investigated.The proposed approach combines the rule logic and optimization algorithm.The‘state-of-charge’and acceleration rate are selected as the control parameters for the mode transition.When the system is working in the torque-split mode,the model predictive control(MPC)of the optimization algorithm is implemented to establish the receding horizon predictive control optimization strategy.The fuel consumption is defined as the target cost function and the engine torque serves as the control variable.The orthogonal simulation experiment is carried out to examine the two MPC parameters,which are the prediction horizon and weight coefficient.Specifically,the prediction time domain variable tradeoff factor was determined.The prediction time domain S was chosen as 3 s,and the tradeoff factors were 0.0001,0.00001,and 0.000001,respectively.In the prediction time domain of constant tradeoff variable,the tradeoff factor is 0.0001,and the prediction time domain S are 3 s,5 s,and 10 s respectively.The results indicate that the best solution is the group consisting of the prediction horizon S=3 s and weight coefficientλ=0.0001;the equivalent integrated fuel consumption for 100 km is 6.179 L/100 km.In addition,comparison of the proposed strategy and the rule-based strategy showed that the former increased fuel economy by 7.8%.Through the above comparative experiments,the effectiveness of the MPC algorithm for improving fuel economy is further demonstrated.