Research On Hierarchical Control Based On Unidirectional Communication Topology For Intelligent Connected Hybrid Electric Vehicle Platoon

This is the future development trend of vehicles through electrification,intelligence and connectivity to reduce environmental pollution and ease traffic congestion.Under the intelligent transportation system,the information between platforms such as vehicle and vehicle,vehicle and road can be shared,which can effectively improve the decision-making planning ability of vehicle and enhance the intelligence level of vehicle,thus achieving the goal of vehicle formation driving.The platoon can reduce the impact of air resistance and improve the fuel economy,while effectively reducing traffic accidents,alleviating driver fatigue and improving road capacity.This thesis presents a hierarchical control strategy based on unidirectional communication topology for intelligent connected hybrid electric vehicle platoon,which is studied with hybrid electric vehicle homogeneous platoon.Firstly,the single vehicle structure and parameters are introduced and dynamics system model is constructed.According to different state variables and control variables,different platoon longitudinal dynamics models are established.At the same time,the platoon spacing control strategy,the platoon communication topology and the Lyapunov stability theory are described.This provides the theoretical basis for the following research.Secondly,in order to solve the problem that the energy management for hybrid electric vehicle platoon cannot adapt to working condition and online implementation,a hierarchical control strategy based on PF topology of LDMPC+Q-Learning for platoon is proposed.The upper platoon longitudinal controller acquires the location and velocity of the vehicle ahead according to the predecessor following（PF）topology,takes the following vehicle’s spacing error e_s,velocity error e_v,acceleration a_i and velocity v_i as state variables,and the desired acceleration a_des as control variables,establishes a linear third-order platoon longitudinal dynamics model,and the linear distributed model predictive control（LDMPC）algorithm is used to achieve a good platoon longitudinal control performance.The lower energy management controller established a state transfer probability matrix based on WLTC+UDDS typical working condition data,adopts the Q-Learning algorithm for offline optimization,and the optimal output torque of the power source is rationally allocated by online look-up table.What’s more,the dynamic programming（DP）is used to compare with Q-Learning algorithm to verify the effectiveness and adaptability of the strategy.Then,in order to solve the dimensional disaster problem of discrete state reinforcement learning Q-Learning algorithm due to too many state variables,a hierarchical control strategy based on PF topology of NDMPC+DQL for platoon is proposed.The upper platoon longitudinal controller acquires the location,velocity and actual torque of the vehicle ahead according to the PF topology,takes the following vehicle’s position s_i,velocity v_i and actual torque T_i as state variables and the optimum requirement torque T_ireq as control variable,establishes a nonlinear third-order platoon longitudinal dynamics model and adopts nonlinear distributed model predictive control（NDMPC）algorithm to achieve good platoon following,safety and comfortable performance.The lower energy management controller is based on four typical working conditions,ECE＿EUDC,1015,WLTC and UDDS,with minimum fuel consumption and maintain battery state of charge（SOC）balance as the optimization objective,and the Deep Q-Learning（DQL）algorithm is used for iterative optimization training to achieve platoon energy management optimization.The advantage of this strategy is verified by comparing it with the LDMPC+Q-Learning control strategy from the last chapter.Finally,in order to solve the longitudinal control and energy management problems for nonlinear hybrid electric vehicle platoon under different unidirectional communication topologies,a hierarchical control strategy based on different unidirectional topologies of NDMPC+EF-DQL for platoon is proposed.The upper platoon longitudinal controller is based on different unidirectional communication topologies to obtain information about the driving state of the preceding vehicles.The established NDMPC multi-objective function is used to construct the Lyapunov stability function,and the established upper platoon control system is proved to be asymptotically stable by demonstrating that the function is bounded and monotonically decreasing.The lower energy management controller adopts the DQL based equivalent factor optimization（EF-DQL）algorithm,which can better balance the consumption of electrical energy and fuel as well as maintain the SOC balance.The equivalent factor can be automatically adjusted to suit various actual working condition according to the driving state.The platoon control performance under different unidirectional communication topologies is compared and analyzed,which shows that the predecessor-leader following（PLF）topology can achieve both good platoon control and fuel economy,but also has better optimality and real-time performance.Meanwhile,the PLF topology is adopted for subsequent effectiveness and working condition adaptability verification,which further proves that the strategy can realize better platoon longitudinal control and fuel economy.