Research On Hierarchical Energy Management Control Strategy Of Intelligent Connected Hybrid Electric Vehicle

Due to the dwindling global resources and increasingly prominent environmental problems,energy-saving and environmentally friendly hybrid vehicles have been widely concerned and become a major development trend.In recent years,the rapid development of intelligent connected vehicle technology has added new directions and new forces to achieve the goal of energy saving and environmental protection.In this paper,a hierarchical energy management strategy for hybrid electric vehicles was proposed to improve the fuel economy,comfort,vehicle traffic efficiency and safety of single planetary row Series-parallel hybrid Hybrid Electric Vehicles(HEV)in the connected urban road environment.The main works of the full text were summarized as follows:(1)The hierarchical control framework of HEV energy management in the networked environment was given;The Simulink communication model of vehicle to vehicle and vehicle to traffic infrastructure was established to simulate the network environment;The four working modes of electric drive,parallel drive,hybrid drive,and braking energy regeneration of single planetary HEV were analyzed,and the mathematical models were established respectively;Based on Simulink,the longitudinal dynamics model and driver model were established,the main component models of the vehicle were established,including the engine model,generator model,motor model,and power battery model.(2)Aiming at the driving scene of a single vehicle and the interference vehicle in front of it in the networked environment,the upper controller was designed,and the optimal target speed sequence was given.Among them,considering the traffic signal light information and the front vehicle information,an algorithm for solving the allowable range of target speed based on pre-deceleration control under different driving conditions was proposed;In order to improve the fuel economy,comfort,and safety of vehicle driving,a multi-objective optimization function considering fuel consumption per unit mileage,vehicle distance,the difference between the actual speed and the initial target speed of the vehicle,the longitudinal acceleration of the vehicle,and the adaptive weight of each performance index was given,the Fast Model Predictive Control(F-MPC)algorithm is used to solve the objective function to optimize the target speed,and the optimal target speed sequence is obtained.The simulation results show that the upper controller improves the traffic efficiency and ensures the safe distance of the following vehicle,and the calculation time cost of the F-MPC algorithm can be reduced to 7.2% of the Model Predictive Control(MPC).(3)Based on the optimal target speed obtained by the upper controller,a lower energy management method based on the Adaptive Equivalent Consumption Minimization Strategy(A-ECMS)combined with the logic threshold was proposed.Among them,the global energy optimization model of power split HEV was established,and the optimal control quantity was solved by Dynamic Programming(DP),and the global minimum fuel consumption was obtained,which provided a benchmark for the evaluation of energy management strategy below;For the energy management strategy based on instantaneous optimization,the Equivalent Consumption Minimization Strategy(ECMS)was improved,and the A-ECMS combined with the logic threshold was proposed,the strategy set the switching value of the working mode according to the engine demand torque and SOC,and ensured the electric drive mode under the conditions of low speed,low engine demand torque,and high SOC value,which avoided the low efficiency of the engine;The simulation results showed that the A-ECMS strategy combined with the logic threshold proposed in this paper obtained better fuel economy than the ECMS strategy,and the fuel consumption per hundred kilometers was only increased by 3.08% compared with the dynamic programming strategy.(4)Based on the optimal target speed obtained by the upper controller,the energy management strategy based on Deep Reinforcement Learning(DRL)was studied,and an improved Twin Delayed Deep Deterministic Policy Gradient(TD3)energy management algorithm was proposed.Among them,based on the characteristics of HEV energy management,the key elements of TD3 such as environmental state,action,and reward were given;Aiming at the problem of high training time cost caused by unreasonable exploration action in agent training,rules of action exploration area were proposed to limit the optimal action exploration area under different working conditions;Aiming at the problems that the TD3 ignored the difference between the advantages and disadvantages of the empirical samples and the catastrophic forgetting of the neural network,an experience replay buffer emphasizing the priority was proposed,and the exponential weighted average was used to distinguish the advantages and disadvantages of the empirical samples.Then,the TD3 algorithm was improved.The simulation results showed that the improved DRL algorithm converged faster and more stably,and the fuel consumption per hundred kilometers was only3.34% higher than that based on the DP algorithm.