Mode Switching Characteristic Analysis And Optimal Control Of A Multi＿mode Power Split Hybrid Electric Vehicle

Automotive electrification has become a key solution to social issues such as fossil fuel consumption and global warming,among which hybrid electric vehicle(HEV)remains an important direction for the current development of vehicle powertrain towards a higher level of electrification.The power split HEV using the double planetary power coupling device can decouple the engine torque and speed from the wheel torque and speed.By integrating multiple clutches into the double planetary drive system,the power split HEV can achieve richer modes in complex driving scenarios.For a multi-mode power split HEV with integrated multi clutches,the transient mode switching process often involves collaborative switching of two clutch states.The difference in switching time between different clutch states will lead to rapid changes in the system’s mechanical freedom and coupling behavior between different power sources in a short period of time,increasing the complexity of mode switching control.Therefore,it is of great significance to carry out the analysis of the mode switching characteristics and the optimal control of the multi-mode power split HEV.This paper takes a double planetary four clutches power coupling system as the research object,and conducts mode switching behavior analysis and optimal control research for the switching process from pure electric mode to hybrid mode involving the cooperative switching of two clutches.The main research contents include:Firstly,a dynamic modeling of the multi-mode power split HEV transmission system was conducted.Based on the structural characteristics of the double planetary with multi-clutch HEV,the mode switching behavior from the pure electric mode driven by two motors to the hybrid mode is analyzed.According to the torque and speed constraints of the planetary components at different stages,based on the matrix method,the system dynamic model of each switching stage is established,and the key component model of the transmission system is established according to the response characteristics of the engine,motor and clutch,which lays the foundation for the analysis and control research of mode switching behavior.Secondly,the analysis and optimization research on the mode switching behavior of the multi-mode power split HEV including dual-clutch coordinated switching is carried out.A dual-clutch synergy model for transient mode switching is established,and the influence of clutch slipping sequence and slipping duration on the engine speed and different mode switching evaluation indicators is explored.On this basis,the clutch cooperative behavior optimization problem in the transient mode switching process is constructed.Based on the simulated annealing algorithm,the clutch torque response parameters and its timing parameters are optimized.The results show that the optimized clutch response parameters can effectively improve the quality of transient mode switching.Then,an optimization strategy for dynamic coordinated control based on deep reinforcement learning algorithms is investigated.Based on the optimized dual-clutch timing parameters and clutch response parameters,a deep reinforcement learning algorithm DDPG is introduced to coordinate the motor torque control.By analyzing the impact of the power source response and clutch response on the jerk degree,the state variables and action variables are determined,and a reward function with the jerk degree as the target is designed,trained the deep reinforcement learning neural network through Matlab/Simulink and Python co-simulation,and obtained the compensation torque of the drive motor during the transient mode switching process.The research results show that the deep reinforcement learning algorithm reduces the jerk of mode switching by 34.4%,and by comparing the reward values before and after compensation control,the proposed control scheme can increase the reward by about84.4%,effectively improving the quality of mode switching.Finally,the hardware-in-the-loop test platform is built,and the hardware-in-the-loop test research of dynamic coordinated control is carried out.Taking the observation state of deep reinforcement learning as the controller input and the motor compensation torque as the controller output,based on the hardware-in-the-loop test platform composed of real-time simulator and D2P controller,the controlled HEV model and control strategy are respectively imported into NI The real-time simulator and D2P controller verify the effectiveness of the dynamic coordination control strategy based on deep reinforcement learning.The results show that due to the addition of the motor compensation torque,the maximum jerk during the mode switching process of the vehicle is reduced from 14.75 m/s~3 to 11.03 m/s~3,the maximum jerk and the fluctuation range of the jerk are significantly reduced,and the mode switching quality is effectively improved.