Research On Configuration Analysis,Parameter Matching And Optimal Control For Hybrid Power System In Electric Vehicles

Developing new energy vehicles is the national strategy in our country.In order to solve the dual requirement between high specific energy and high specific power for power battery and the problem of easy degradation of battery performance,this paper proposes to use a combination of high specific energy lithium ion battery and high specific power super capacitor,the following research work is carried out based on the key issues of system configuration,parameter matching,real-time energy management and algorithm verification existing in PHEV composite power system:1)Aiming at the problem of component modeling and parameter matching in the composite power system,different topologies of composite power are analyzed.Modeling of power battery and supercapacitor,construction of experimental system,parameter identification and model verification,and the research of DC/DC converter are carried out.Based on the analysis of the performance of the target vehicle according to the selected operating conditions,the global optimal algorithm of dynamic programming is applied to optimize the energy management with the goal of minimizing the system energy loss and accurate energy management.The optimal matching results of the parameters laid a theoretical foundation for the PHEV’s energy refinement management strategy.2)Aiming at the problem of coupling between the energy management and the operating temperature and aging of the battery pack,the system parameter optimization and energy management integrated optimization platform established by the global optimization algorithm of dynamic programming,the effect of the operating conditions of the battery on the energy management is investigated.Based on the obtained optimal control trajectory,a new energy management control strategy is extracted between the power ratio of the power system,PHEV demand power and vehicle speed.Compared with the original rule control strategy,the analysis results show that the strategy based on the integrated optimization algorithm platform has superior control performance and energy consumption.3)Aiming at the disturbance of PHEV realtime driving cycles on the energy management of the composite power system,this paper proposes an environment-friendly reinforcement learning method based on environmental feedback.The PHEV vehicle power requirement under real-time operating conditions is described as Markov chain stochastic process,a control strategy based on real-time reinforcement learning algorithm is established.An energy management framework system based on Q-learning online algorithm is constructed.The impact of different forgetting factors and Kullback-Leibler divergence rate is discussed.It is shown that the control strategy can reduce the average discharge current of the power battery pack,make the supercapacitor pack charge and discharge more frequently.The reduction rate of the total energy loss can reach 12%.4)Aiming at the practical application and verification problem of energy management strategy in composite power system,a hardware-in-the-loop simulation platform based on x PC Target is set up.The original rule control strategy is compared with the control strategy based on real-time reinforcement learning and the power battery and supercapacitor are tested separately.The experimental results show that the control strategy based on real-time reinforcement learning can reduce the average discharge current of the power battery,and increase the number of charge and discharge for the supercapacitor pack.The method of combining battery and ultracapacitor proposed in this paper can effectively solve the dual requirements of PHEV.The integrated optimization platform of parameter matching and energy management strategy for the composite power system can effectively slove the configuration optimization,parameter matching and rule control strategy optimization.A new energy management control strategy is put forward by considering the influence of operating temperature and aging of power battery.Energy management based on reinforcement learning can greatly improve the real-time energy refinement management of the composite power system and solve the realtime driving cycle distribution problem of the energy management for the composite power system.Moreover it has good engineering application value.