Study On Predictive Control Technology For The Multi-power Of Electro-mechanical Transmission

Nowadays Resource shortage and environment pollution are two big challenges in theworld. Hybrid electric vehicle (HEV) is considered as the most promising kind of vehiclefor industrialization because of its advantages, such as low energy consumption, lowemission and long drive distance. Electro-Mechanical Transmission is an important type ofHEV. And the design of energy management strategy of HEV is one of important part forthe vehicle development as well as one of key technology for the energy saving andemission reduction. Therefore, the energy management strategy becomes the research focusfor HEV. There are many studies of energy management strategy based on globaloptimization and transient optimization, and these methods have different advantages anddisadvantages. The control strategy based on model predictive control of HEV can predictthe future operation of the system state in advance optimize control variables, and fullyplay the excellent performance of HEV with low fuel consumption and low emissions. Thispaper will research on model predictive control strategy of the EMT.In this paper, the dual-mode EMT system is studied as the research object, the forwardsimulation model of the EMT system is established and the key technology of modelpredictive control strategy is explored. For the demands of model predictive control, asimplified model for online control needs to be established. The application based ondynamic programming algorithm, quadratic programming algorithm and a one-dimensionalsearch algorithm in the optimization of the energy management system of EMT are studiedrespectively and optimization of the most potential in online usage is determined. Theenergy management strategy based on model predictive control of EMT is established. Themain work and conclusion in this paper are as follows:The mathematical model of engine, motor, battery, driver, transmission and charge anddischarge of battery and vehicle dynamics model are established in detail. ForwardSimulation Model of EMT system based on Matlab/Simulink is established, meanwhile, acombination of Experiments modeling and theoretical modeling is developed. Themathematical models of engine, motor, battery, driver, and transmission are established,respectively. Note that due to the design needs of controller, the corresponding modelsabove are simplified. According to the energy transfer direction of forward Simulation andthe component models above, a simplified vehicle model of EMT system is build and itprovides the basis for verification of energy management strategy. The method of combining the historical data fitting and accelerator pedals is proposedand future operational state of the system is predicted. Prediction model based on rollingoptimization is established, and this model sets vehicle current speed, the accelerator pedalopening position and the battery status as input variables, and sets the next time speed orSOC as output variables. Rolling optimization Control strategy based on model predictivecontrol and stability of Fuel consumption and SOC as an objective function is establishedby combining forecasting model with different scrolling optimization algorithm, throughwhich above we can get optimal control engine speed and torque. The simulation model isused to simulate for different energy management strategies in UDDS, NEDC and heavyvehicles cycle conditions, respectively. The results show that dynamic programming andquadratic programming rolling optimization algorithm can realize good fuel economy andchanges of SOC. However, due to the large amount of computation, simulation response islonger, the real-time is poor. One-dimensional search algorithm has poor performance interms of fuel economy and Changes of SOC, but this algorithm has a good real-timeperformance. Therefore, it is selected as the rolling optimization algorithm of modelpredictive control strategy.Energy optimization management strategy Of Model predictive control based on theone-dimensional search algorithm is proposed. Model predictive control strategy based onFibonacci algorithm is established and optimal power allocation scheme is obtained byrolling optimization. The impact on fuel economy and SOC of the EMT from differentobjects prediction, forecasting domain length, and control field length and weightcoefficients are analyzed, respectively. The control strategy of EMT is simulated andanalyzed under the same situation where has same weight coefficient and simulationplatform, through selecting different prediction target, different weighting coefficientsetting method, different prediction domain time domain and different control time.