Research On Powertrain Matching Design And Energy Management Strategy Of Dual-Motor Four-Wheel-Drive Electric Vehicle

With the depletion of fossil resources and prominent environmental problems,the development of energy-saving and environmental friendly electric vehicles has become the demand of the times.Electric vehicles are favored by the market due to factors such as no exhaust pollution and national policy subsidies,but the driving range has also restricted its large-scale promotion.The dual-motor four-wheel-drive electric vehicle(DM FWD)can make full use of the difference in the efficiency of the two motors without changing the total driving torque,and make reasonable torque distribution of the front motor and rear motor to expand the energy-saving space.The existing matching studies of dual motor powertrain are mostly focused on the planetary row coupling coaxial drive mechanism,and there are few studies on DM FWD configuration;in addition,in the existing energy management strategy for this configuration,there are few studies on how to achieve battery state of charge(SOC)trajectory planning in the local time domain.Aiming at the above problems,this paper takes DM FWD as the research object,focusing on the multi-power source matching method based on the energy weight of the working condition and the energy optimization strategy with strong working condition adaptability and predictive ability.Firstly,matching design of the DM FWD powertrain is studied.First,combined with the existing matching theory,the total power required is determined by the extreme stable condition and short-term acceleration condition.Then mathematical analysis of common driving conditions based on the energy weight method is performed,which matches the high-efficiency interval of motor with the high-energy weight interval of the driving condition.Finally,the rationality of the matching design is verified by the Cruise simulation software.Secondly,based on MATLAB/Simulink,a forward simulation platform of the DM FWD is established.According to the driver’s control logic,a longitudinal driver model was built.According to the internal principles and characteristics of the components,the front and rear motors and battery models are built,and the vehicle longitudinal model was established.Finally,based on the proportional control strategy,the NEDC condition was used as the test condition and the accuracy of the vehicle longitudinal performance and energy solution performance of the simulation model are compared with the Cruise model.Thirdly,a local SOC trajectory planning method based on driving condition recognition is formulated.First,a global optimization strategy is formulated based on the forward dynamic programming algorithm,which lays the foundation for determining the optimal trajectory for different conditions.Considering the differences of the SOC trajectory in various driving conditions,a driving condition recognizer are designed based on the random forest network,and k-means clustering method is used to screen core feature parameters as the input for driving condition recognizer.Then,by comparing various driving conditions in time domain and spatial domain,it is found that the SOC optimal trajectory in the spatial domain are linear uniform under various driving conditions.Therefore,select the global optimal trajectory in the spatial domain as the linear trajectory,and a fuzzy controller is designed to determine the relationship between the local condition characteristics and the relative slope of the SOC,which serves to correct the linear trajectory.Finally,the effectiveness of the local SOC trajectory planning method is verified through the combination of city conditions.Finally,based on the model predictive control,the vehicle energy optimization problem is constructed,and the SOC trajectory planning in prediction time domain and prediction model are explained.First,based on the condition information in the predicted time domain,the SOC in spatial domain is determined according to the local SOC trajectory planning method,and then combined with the condition information,it is transformed into time domain.Furthermore,based on the radial basis network,a vehicle prediction model is built,and the influence of the prediction time and the spread coefficient on the accuracy of the prediction is explored.The two above parameter are determined based on the lowest root mean square error between the predicted value and the actual value.Combined with the driving condition recognizer,the corresponding driving condition data is used to train the velocity prediction model under various driving conditions.Finally,a comprehensive driving condition is constructed through highway driving condition,suburb driving condition and urban driving condition to verify the collaboration effect of the various modules of the energy strategy.The results show that the recognition accuracy rate under driving conditions is about 88.6% and the maximum velocity single-step prediction error is among [-2.5,+2.5]km/h.What’s more,the above energy strategy is compared with the global optimization strategy and the proportional strategy respectively.The simulation results show that compared with the global strategy,the energy strategy in this paper can achieve about 95.13% of the global optimization effect,and compared with the proportional strategy,it can effectively improve 14.28 % of vehicle economy.