Research On PHEV Energy Management Strategy Based On Short-term Speed Prediction

Under the requirements of the development of green energy and environment-friendly society,hybrid electric vehicles,fuel cell vehicles,battery electric vehicles and other new energy vehicles have gradually become the main development direction of automotive technology research.The Plug-in Hybrid Electric Vehicle(PHEV)has been widely recognized as the one most meets the objective conditions in the near-term and is widely developed in worldwide.The energy management strategy is essential to balance the performance of the vehicle power,emissions,and fuel economy.This paper proposes a mixed logical dynamical model predictive control strategy based on a short-term prediction for driving cycles.The main contents are as follows:First,the vehicle powertrain structure and vehicle parameters are determined.Then the mathematical models of the controlled PHEV are established,i.e.,the longitudinal dynamic model of the vehicle,including the dynamic model of the wheel and transmission system,the engine fuel consumption rate model,the motor equivalent fuel consumption rate model and the battery SOC changing rate model.In order to find a reasonable SOC reference trajectory formulation method,the PHEV global optimization energy management strategy based on dynamic programming is studied,and the fuel economy simulation is performed under the NEDC condition.The SOC trajectory obtained by the global optimization can be considered to be quasi-optimal.The regression analysis method is used to fit the linear regression equation of the SOC reduction and cycle characteristic parameters over a period of time.This method is used for formulating the SOC reference trajectory,and it is applied in the MLD-MPC control strategy based on short-term driving cycle prediction to determine the terminal SOC constraints of each prediction horizon.This paper mainly focuses on the MLD model predictive control strategy of PHEV based on the vehicle speed prediction.First of all,the prediction of the future short-term driving cycles is need to be realized,and it is used to calculate the future short-term vehicle demand torque based on the longitudinal dynamic model in chapter 2.As the driving intention directly affects the change of vehicle speed,this paper proposes a NAR neural network short-term driving cycle prediction method based on the driving intention analysis.The driving intention is identified through fuzzy inference and is used for the training of NAR neural networks together with historical vehicle speed data.Finally,the prediction accuracy of NAR neuralnetwork prediction method is analyzed.The MLD model is constructed for the typical hybrid system of PHEV.First,the engine fuel consumption model and the battery SOC changing rate model are linearized;then,the state transition equation and the evaluation index equation of the energy management strategy are established,also their constraint equations are set according to the six PHEV operating modes.Finally,the established MLD model is utilized combined with the concept of model predictive control.Taking the reference trajectory of SOC into account,the terminal SOC in the prediction horizon can be limited,and by solving the MILP problem converted from MLD-MPC model,the local optimal distribution of driving torque in the prediction horizon can be calculated,and this local optimization is repeated in every control step which realize the rolling optimal control of MLD-MPC control strategy under the total driving cycle.Through simulation and comparison analysis,the MLD-MPC energy management strategy proposed in this paper based on short-term driving conditions prediction has high solving efficiency,and can achieve near global optimization control effect.