基于驾驶循环预测的发动机冷却系统优化控制

Optimization control of engine cooling system based on driving cycle prediction

针对现实中难以获取未来驾驶工况信息的问题,提出了平均意义上的驾驶工况模型及随机动态规划的控制方法。首先,采用基于马尔可夫链的驾驶循环生成器获得驾驶员需求加速度的期望值;然后,设计随机动态规划(SDP)控制器获得对于一簇统计特性相近的驾驶工况均适用的近似最优控制策略;最后,通过与确定型动态规划(DDP)和单步模型预测控制(SSMPC)对比,验证了本文方法在计算时间及控制效果上的优越性。

For engine electrified cooling systems equipped with an electric water pump,fan and thermostat,through optimizing the actuator action accounting for the future driving conditions,the engine economy can be improved. Since it is difficult to obtain the future driving conditions in real driving,this paper proposes a driving condition model in average sense and a Stochastic Dynamic Programming(SDP)control method.First,the Markov chain driving cycle generator is applied to obtain the expectation of driver’s demand acceleration. Then,a SDP controller is applied to obtain an approximate optimal control strategy for a cluster of driving conditions with similar statistical characteristics. Finally, the comparisons of a Determining Dynamic Programming(DDP) controller and a Single-Step Model Predictive(SS-MP)controller show the advantages of proposed SDP controller.