Study On Predictive Energy Management Strategy For Hybrid Electric Vehicle

Supercapacitors and lithium batteries are two energy storage components with very different properties: supercapacitors have the characteristics of high power density and low energy density,which can release or absorb a lot of energy instantaneously,but they cannot provide stable power output for a long time;Lithium batteries have the characteristics of high energy density and low power density,and can store a lot of electric energy,which can reach hundreds to thousands of times that of super capacitors.However,lithium batteries cannot withstand instantaneous large current charge and discharge,otherwise their life will decay rapidly,and there may be security risks.Compared with single-power electric vehicles with only lithium batteries,the hybrid electric vehicle composed of supercapacitors and lithium batteries has many advantages:1.The supercapacitors can bear the peak power of the power demand for lithium batteries,so that the lithium batteries can reduce the frequency of high current operation and extend the life of lithium batteries;2.The supercapacitors can effectively absorb the instantaneous high-power braking energy of the vehicle braking,avoid the impact of the instantaneous large current on the lithium batteries,protect the lithium batteries,but also can efficiently recover the braking energy;3.The system composed of supercapacitors and lithium batteries can flexibly adjust the power output distribution ratio according to the working condition,so that the supercapacitors can timely consume the braking recovery energy,and ensure that the lithium batteries works in a stable state,so that the driving range of electric vehicles can be improved.This paper takes the hybrid electric vehicle as the research object,studies the energy management strategy of the hybrid electric vehicle,and finds that if the energy allocation strategy of the electric vehicle can take the future power demand into account,it will have great energy saving potential.However,how to get the available future power requirements is a complex matter.However,the advent of deep learning makes it possible to predict future information about vehicles.In this paper,the deep learning method is used to predict the future power demand of vehicles,and the interaction between vehicles and complex traffic flow is considered.The main contents of this paper are as follows:1.Hybrid power systems and dynamic models for electric vehiclesBy analyzing the characteristics of the composite power system with different topologies,the appropriate scheme is selected as the structure form of the DC side of the electric vehicle.The models of lithium battery,supercapacitor,bidirectional DC/DC converter,rectifier/inverter based on SVPWM algorithm and motor based on vector control are established and verified by Matlab/simulink respectively.Based on the relationship between vehicle longitudinal dynamics and transmission system,the relationship between vehicle speed and the required torque and speed of motor is established.2.Speed prediction model under complex traffic flow environmentBased on the idea of vectorized space feature coding,a deep learning model of vehicle speed prediction considering surrounding traffic flow information was established.The structure of the model can be divided into three parts: subgraph,global graph and velocity regression.The velocity regression part adopts the two-stage output method to guide the future vehicle speed with the future vehicle trajectory.The validity of the model is verified on the current mainstream vehicle behavior prediction dataset Argoverse-Ⅱ.According to the future speed,combined with the longitudinal dynamics equation of the vehicle,the future power demand of the vehicle is further obtained.3.Hybrid power energy management strategyFirstly,the power distribution control problem between the supercapacitors and lithium batteries in the energy management strategy is solved.The power distribution controller is established based on the idea of PID.The controller generates the corresponding control signal according to the power expectation,and sends it to the DC/DC converter,so that the output of the supercapacitors or lithium batteries can reach the expected power.This paper introduces the mainstream rule-based energy management strategies in the field of hybrid electric vehicles,and establishes a predictive energy management strategy considering future power based on this.4.Parameter calculation and matching of each component of hybrid electric vehicleThe maximum speed,maximum acceleration,reference mass and other basic parameters of the vehicle are determined,and based on the calculation of the driving motor limit performance,and determined the driving motor parameters,and the motor limit performance of the simulation verification.The total capacity of lithium battery pack and the number of battery series and parallel were calculated with the maximum range of 100 km as the standard.The total capacity of supercapacitor pack and the number of unit series and parallel are calculated based on the braking energy recovery and drive assistance.5.Offline simulation platform building and validationBased on the open source traffic flow simulation software SUMO,a traffic flow simulation environment conforming to the simulation requirements is established.In terms of road structure,considering the simulation requirements and modeling difficulties,the three-ring coupled structure is selected.On this basis,a suitable traffic flow generation and interaction strategy is developed to ensure that the traffic flow at every moment is random and non-repetitive in the simulation environment of traffic flow.Based on the hybrid electric vehicle model and the speed prediction model established above,combined with SUMO simulation conditions,the rule-based energy management strategy and the predicted energy management strategy were simulated and verified respectively,and the differences between them were compared.