Research On Energy Management Strategy Of Plug-In Hybrid Electric Vehicle Considering Battery Thermal Effects

Plug-in hybrid vehicles(PHEV)combine the advantages of electric vehicles and hybrid electric vehicles,in the context of traditional internal combustion engine vehicles facing technological replacement,and electric vehicles still have some issues with safety and range anxiety,the PHEV which equipped with two sets of drive components could be an important solution for those transition period.Hybrid technology breakthroughs will play a significant impact in the low-carbon energy transition.Energy management technology is the key of plug-in hybrid technology,and is also the premise and foundation for ensuring that the fuel economy and other performance to achieve maximum efficiency.A maturity energy management system could fully coordinate the power distribution between two sets of power units,so that engine,electric motor and other components could operate in the optimal operating range.While improving the vehicles fuel efficiency,energy management strategy(EMS)also needs to consider the life cycle costs of battery and pay attention to the environmental temperature of battery.Therefore,this thesis focuses on a parallel PHEV and aims to conduct the research on multi-objective EMS,which considering the battery thermal effects,so as to achieve the objectives of enhancing the fuel economy and limiting the battery temperature rise.The main accomplished contents of this thesis are listed as follows:(1)Modeling and model validation of PHEV.The forward simulation models of hybrid power system model,vehicle dynamics model,driver model and powertrain model based on numerical model are set up.To evaluate the battery heat exchange dynamics during the ride,a calculating model of cells operating temperature is then established.According to the test of speed following simulation,the result verifies the rationality of forward simulation models,those models provide a simulation platform for the study of control strategies.(2)Construction and verification of typical high-speed driving cycle.By selecting low frequency data as the database to complete the preprocessing of driving cycles,then considering the correlation coefficient between the different kinematic sequences.Through the PCA of characteristic parameters,the dimension reduction of characteristic parameters is completed.The K-means clustering algorithm is applied to cluster kinematic sequences to obtain four kinds of driving cycles.Pick out those kinematic sequences based on the principle of cluster selection and then select the representative sequences to construct a typical high-speed driving cycle.By means of comparing the characteristic parameters with several driving cycles,the rationality of the constructed driving cycle is proved.This work lays the foundation for the following prediction and simulation.(3)The simulations of the battery temperature are conducted based on several driving cycles.In order to observe the temperature change of the battery during the driving,the battery temperature rise simulations are conducted based on the constructed driving cycle,NEDC and WLTC.Accordingly,we hereby propose the solutions to resolve the problems of battery thermal accumulation during the simulations,i.e."power demand prediction" and "multi-objective EMS".(4)Establishment and verification of the short-term driving cycle prediction model.Take the constructed driving cycle data as the sample data and propose a driving cycle prediction model based on stochastic Markov process.The velocity and acceleration of constructed driving cycle data are processed with the discrete way,and the state transition probability matrix of acceleration is estimated based discrete velocity grid.Based on this,Markov chain is used as the forecast method for the driving cycle prediction model under the entire driving cycle.Through the analysis of RMSE,the accuracy of the prediction model with 3,4,5,6s prediction horizon is proved.(5)Abstraction of balanced optimal time-varying equivalent factors.The multi-objective EMS is proposed,which takes the fuel economy and the battery temperature rise into consideration.Due to the shortcomings of the weighted method that converts multi-objective EMS into single-objective EMS,by using Pareto front to solve the problems of multiple objects.The multi-objective optimization task is introduced into the equivalent consumption minimization strategy(ECMS)by selecting the equivalent factors as decision variables,in order to get the balanced optimal time-varying equivalent factors,an iterative optimization method of equivalent factors based on NSGA-II is proposed.(6)The multi-objective optimization-oriented ECMS based on vehicular power demand prediction model is built.To solve the problem of control hysteresis caused by passive adjustment of equivalence factor(EF),the power demand prediction model is incorporated into the equivalence factor abstraction algorithm.This method aims to adjusted EF by using the forecast information to adapt the power balancing requirements.The EMS effect of the proposed method,including the fuel economy,battery temperature control effect and battery SOC sustaining effect,is simulated under the constructed driving cycle.The results prove the rationality and effectiveness of proposed method.By comparing the control effect of proposed method using the forecast information and actual information,the robustness of the proposed method against the forecast error is demonstrated.Furthermore,the adaptability of the method under different driving cycles is analyzed and verified.