混联式混合动力汽车工况预测能量管理研究

Driving Cycle Prediction Energy Management Strategy of Series-Parallel Hybrid Electric Vehicles

为了解决混合动力汽车预测能量管理策略中工况预测不准确的问题,探索研究了基于工况预测的能量管理策略。首先基于模型预测的方法分别建立了多阶马尔可夫和神经网络的预测模型,对工况进行预测;在此基础上提出模型预测能量管理策略的方法,利用动态规划作为其滚动优化部分对混合动力整车的能量进行优化分配。通过仿真计算表明,基于神经网络的工况预测方法具有较高的精度,能够满足使用要求;同时基于神经网络的模型预测能量管理方法能够逼近动态规划算法的最优性,而且具有一定的实时应用潜力,为后期实车能量管理策略的实时应用打下基础。

In order to solve the inaccuracy of the driving cycle prediction in the predictive energy management strategy,the research of the predictive energy management strategy were executed.Firstly,the multi-order Markov Chain model and neural network prediction model were established respectively to predict the driving cycles.On this basis,the method about the predictive energy management strategy was proposed,and the Dynamic Programming was used as rolling optimizationpart to optimize and allocate the vehicle energy.The simulation and calculation indicates that the driving cycle prediction based on neural network has high precision and is able to meet the usage requirements.Meanwhile,the energy management strategy of Model Predictive Control(MPC)based on neural network is capable of reaching as close as the best results of Dynamic Programming(DP)algorithm.Moreover,the method has the real-time utilization potentiality and lays a foundation for the later energy management application in real vehicle test.