Power Management And Components Sizing Of Dual-motor Drive System Used In Battery Electric Vehicles

Compared with single-motor drive system used in battery electric vehicles,dual-motor drive system contains more working modes,which makes it has the ability to choose different working modes according to driving patterns.For example,when the power requirement is low,the dual-motor system can drive the vehicle by a small motor,which can increase the load rate and promote motor working efficiency.When the power requirement is high,the system can work in two-motor mode.As the performance of dual-motor driving system mainly depands on the system working mode,the power allocation between two motors and system's parameters,this paper carried out a detailed study on these aspects.To analyze the dual-motor system's working property,the effeicinecy model of the dual-motor driving system has been built including motor power loss,planet system power loss,wet clutch power loss and bearing power loss.To verify the dual-motor system efficiency model,the test bench for the system is established.The NI test system is applied to record the efficiency distribution and the comparision between the test data and mode data indicates that the dual-motor system's efficiency model can well express the system's effiency property.This study proposed a control strategy design method based on driving blocks classification for dual-motor drive system.Based on 20 typical drive cycle,298 driving blocks have been abstracted and these driving blocks are classified into different types according to a classification parameter.Then dynamic programming algorithm is applied to optimize the control strategy for different types of driving blocks and then the driving pattern identification method is proposed.The combination of different control strategies and driving pattern recognition method make up an adaptive control strategy.To determine the classification parameter,ten parameters are tried to be applied in the adaptive strategy,and after comparing their performances,the classification parameter and its updating cycles are finally determined.Simulation results indicate that,compared with original control strategy,the proposed new control strategy shows better efficiency performace with different driving cycles.The hardware in loop experiment bench based on MC9S12DP256B?dSPACE AutoBox and so on are constructed to verify the performance of new strategy.A predictive power management for the dual motor driving system is applied.To make sure control performance of predictive power management,a velocity predictor with three coefficients are developed,and compared with conventional markov theory based velocit y predictor and neural network based velocity predictor,the new predictor shows better prediction performance.As dynamic programming algorithm is applied in our predictive power management,the heavy calculation burden makes it hard to be applied online.So we developed netual network based power management,which will output the rotation speed of M1 directly according to history vehicle speed.Simulation results indicate that though the neural network based power management's improvement is lower than predictive power management,the discrepancy is small and the neural network based power management is easy to be applied online.Finally hard ware in loop experiment is carried out to verify the neural network based power management.Experiment results indicate that this strategy have good online control performance and can satify the vehicle motivation requirement.By analyzing the optimal control strategy of dual motor driving system,we added two modified dual-motor system topologies.The scaling efficiency model for the dual-motor driving system is developed: The motor efficiency model is build according to Willans line mode;the planet model is buil by meshing power method;the wet clutch dragloss mode is built according to viscous fluid mechanics principle and the size of the wet clutch,which is obtained by some assumptions.Based on the scaling models,two integrated optimization methods are developed: QP-DP(quadratic programming and dynamic programming)based mehod and PSO-DP(particle swarm optimization)based method.The optimization target is to minimize the system energy loss and system power level.Simulation results indicate that the system energy loss will increase with the increasing of kp and the system energy loss will decrease with the increasing of ix.When the system's power level is given,the efficiency performance of topology 1 is better than that of topology 2 and topology 3.On the other hand,the simulation results also show that the QP-DP based optimization method is easy to be traped into local optimal while the PSO-DP can effectively avoid this condition and get global optimal design.