具有实时运算潜力的并联混合动力汽车模型预测控制

摘要

The control strategy of the hybrid electric vehicle （HEV） plays an important role in the fuel economy and emission performance. This paper presents a model predictive control （MPC） torque-split strategy for parallel hybrid electric vehicle powertrains based on a simplified first-order diesel engine transient model. A torque request prediction method based on the driver＇s accelerator pedal is presented for the MPC strategy. With consideration of the motor＇s compensation to the engine＇s dynamic torque response delay, the optimization constraints and objectives were designed with the non-linear HEV torque-split problem simplified to a linear programming problem. In the simulation the MPC torque-split strategy, which is potentially real-time-capable, operates in less than 150 ms. Simulation results show that the MPC strategy can improve the fuel economy by 2.5% compared with a rule-based strategy.%混合动力车辆（HEV）的控制策略对提高燃油经济性和排放性能十分重要。该文基于简化过的一阶柴油发动机动态模型，提出了一种并联混合动力汽车模型预测控制（MPC）转矩分配策略。该策略提出了基于驾驶员加速踏板位置的需求转矩预测方法，在考虑用电机补偿发动机转矩响应延迟的情况下，设计了优化问题的约束条件以及目标函数，把非缌陛的混合动力汽车转矩分配问题转化成为线性规划问题并进行求解。该模型预测控制转矩分配算法在计算机仿真中的运算时间小于150ms，具有实时运算的潜力。仿真结果表明：与一套基于规则的控制策略相比，该模型预测控制策略能将混合动力车辆的燃油经济性提高2．5％。