Study On Energy Management Optimization And Auxiliary Power Unit Control Of Plug-in Hybrid Electric Vehicle

Subject to the restrictions on the development level of power battery, plug-in hybrid electric vehicle(PHEV) runs in a limited range in chage-depleting mode.It usally runs in charge-sustainning and blended-driving mode when the charge of power battery is insufficient. It is important to reduce oil consumption for promoting the use of PHEV. Many factors influence the oil consumption of PHEV, among which energy management unit and auxiliary power unit(APU) of hybrid powertrain system impact oil consumption greatly while it runs in charge-sustaining and blended-driving mode. According to driving cycles and driver’s intention, energy management unit assigns required load reasonably to lower implementing agencies and enable hybrid powertrain system to work in high efficient area and improves efficiency of the powertrain componets. APU, one of the lower implementing agencies, is a key componets to realize the energy management strategy and should be controlled quikly, accurately and stably so as to track the required load and work stably at the given point of energy management unit.In this dissertation, aiming at lowering oil consumption of PHEV in its charge-sustaining and blended-driving mode, study on energy management optimization and APU control is carried out. Main contens of this dissertion include following aspects.1) Powertrain parameters should be matched reasonably in advance so as to improve the componets of power source to adapt to driving cycles and expand the componets of power source in more efficient range of driving cycles. The method of matching powertrain parameters according to the dynamic change of torque and required power during satisfiying the power performance process is proposed. After analyzing the dynamic changes of torque and power during satisfying the required performance, required powers are assigned by means of hybridization according to the properties of engine and motor. Rated power of main driving motor is determined according to the statistical prosperities of driving power required. In view of engine efficiency and required power, rated power and speed of integrated starter-generator (ISG) are determined. Engine and motors can work in more efficient range of various vehicle speeds by way of reasonable match of the powertrain parameters.2) In order to provide a verification and evaluation environment for optimizing energy management, the forward model of the hybrid powertrain is established. The models of key components such as engine, motor and power battery are set up using the modeling method of combining theoretical modeling and experimental data modeling. A PI controller is used to model the acts of driver. The forward model is used to simulate the hybrid powertrian under the condition of Kuming driving cycles. The tracking speed error shows the precision of the model is satisfactory.3) Energy mamagement strategies are optimized so that the componets of powertrain can work efficiently which therefore improves the efficiency of powertrain system. The energy management strategy based on logical rules is in wide use because it is easy to implement and has good real-time performance.The frist step is to optimize the control parameters of the energy management strategy based on logical rules.Because genetic algorithm that has many evolutionary operators and parameters evolves slowly, particle swarm optimization(PSO) algorithm is used to optimize the control parameters of the logical rules. However, the energy management strategy based on logical rules can’t get the global optimal performance of desired objective during the whole driving cycles. Danamic programming(DP) is used to solve the energy manage strategy so as to get the global opimal performance, which can be used as the reference comparison of the energy management strategy based on logical rules and furthermore guide the extraction of logical rules. The method named boundary-line fitting is proposed to improve the shortcoming of basic dynamic program on condition of boundaries of the states. Markov property of driving cycles and driver’s operation is studied and the Markov chain of required power is set up. Stochastic dynamic program is carried out offline to determine the best control strategy for each state transition and then the real-time embedded system implements the strategy online in form of table or curve fitting.4) In order to enable APU to track the required load and work stably at the given point of energy management unit, APU should be controlled quikly, accurately and stably. The speed-torque control model of tracking power is put forward. PSO is implemented to optimize the control parameters of classical PID controller offline. However, the parameters optimized by PSO offline are not adaptive online. BP neural network has the advantage of self-learning online. The controller based on BP neural network and classical PID is put forward to realize APU adaptive control. In order to implement the neural network in embedded system online, single neuron PID controller with simpler structure and less calculation is proposed. Those tree types of APU controller are compared in load-tracking and anti-jamming ability. According to the work characteristic of the engine, APU components are designed to work at the area of constant speed and variable qorque while it is working in serial mode so as to reduce fuel consumption and keep bus voltage stable.Vehicle control unit acting as top control unit of powertrain control system is desiged to verify the optimizied energy managent strategy and APU control. Road test shows the optimized energy manage strategy is feasible.