Multi-objective Optimal Control Strategy Of Plug-in Hybrid Electric Vehicle

Selective catalytic reduction technology is considered to be one of the necessary technique to reduce NOx emissions.However,SCR catalytical efficiency is very sensitive to temperature.Especially in the process of multi working mode switching of Plug-in diesel electric hybrid vehicle,the frequent start and stop of the engine causes the great fluctuation of the exhaust temperature of the engine,which leads to the decrease of SCR efficiency and the deterioration of emission performance.Therefore,it is of great significance to develop an optimized vehicle control strategy to realize the comprehensive optimal control of engine fuel consumption and NOx emission of SCR’outlet.In this thesis,a P2 plug-in diesel electric hybrid vehicle is investigated.The fuel consumption and NOx emission at SCR outlet are taken as the optimization objective.The multi-objective optimal control strategy of plug-in diesel electric hybrid vehicle is presented.The main content is as follows:(1)Through the detailed analysis of the whole vehicle system architecture of plugin diesel electric hybrid vehicle,a forward vehicle dynamic model including engine,motor,battery,SCR,transmission,driver and vehicle longitudinal dynamics are established based on MATLAB / Simulink,which is regarded as the basis of control strategy simulation analysis and hardware in the loop test.(2)Based on the Pontryagin’s maximum principle,the energy management problem of plug-in diesel electric hybrid vehicle is transformed to a constrained optimization problem.Taking the weighted sum of fuel consumption and NOx emission of SCR outlet as the objective function,a multi-objective optimal control strategy based on the PMP is proposed.The equivalent factor of battery charge and discharge as well as the weight factor of SCR temperature change rate are introduced to ensure the maximum use of battery energy and fast start-up of SCR.The results show that the fuel consumption and emissions of the multi-objective optimal control strategy based on the minimum principle are significantly reduced compared with the rule-based strategy.(3)Deep reinforcement learning is applied to the energy management of plug-in diesel electric hybrid vehicle.The multi-objective optimal control strategy based on depth Q-learning algorithm and depth certainty strategy gradient algorithm is proposed.The nonlinear relationship between vehicle environment state and optimal control action is fitted by using depth neural network.The proposed control strategy takes the demand power,SOC and SCR temperature as state variables and the optimal power of the motor as output variables.It is an end-to-end control strategy.The simulation results show that the multi-objective optimization control strategy based on deep reinforcement learning algorithm can achieve good fuel saving and emission performance in offline scenarios and has the potential for real-time online application.(4)Based on the d SPACE/Autobox real-time simulation system and the vehicle controller developed with MC9S12XS128 MCU,a hardware-in-the-loop test platform is built to test the basic functions of the vehicle controller software designed in this paper.The test results show that the different state switching logic of the vehicle controller application layer meets the design requirements,and the torque allocation strategy under normal driving state is basically consistent with the simulation results,which verifies the validity of the proposed control strategy and the feasibility of its application in embedded system.