Research On Torque Distribution Control Of Distributed Drive Electric Vehicles

The decentralized drive electric vehicle(EV)is known as the best kind of EVs regarding its advantages in arrangement,dynamic and handling performance.However,electronic stability control(ESC)now widely used is not suitable for a decentralized drive EV.Supported by International Cooperation Project of Nanjing University of Science and Technology “Optimal torque distribution method based on nonlinear tire model for decentralized drive EVs”,this paper proposes a novel torque distribution control algorithm for decentralized drive EVs based on direct yaw moment control(DYC).First,this paper builds a 14 degree-of-freedom vehicle model consists of tire models,vehicle body planar motion,suspension and rotation model,and wheel rotation models in Simulink.Second,this paper designs estimators for tire forces and vehicle side slip angle considering features of decentralized drive EVs.The tire forces estimator operates on the wheel rotation model and planar vehicle body dynamics.The vehicle side slip angle estimator is based on unscented kalman filter(UKF).The tire forces used in UKF comes directly from the tire forces estimator and therefore there is no need for complicated tire model,which simplifies the vehicle side slip angle estimator.Third,this paper proposes a hierarchical torque distribution controller.The upper yaw moment controller calculates the desired yaw moment using sliding mode control(SMC).The state variable in SMC is a combined variable of yaw rate and side slip angle.The torque distribution algorithm in the bottom is based on optimization theory.It aims at minimizing the yaw moment generated by longitudinal tire forces,considering yaw moment generated by lateral tire forces.Finally,this paper uses hardware-in-the-loop(HIL)simulation to validate the torque distribution control algorithm.Calculation of optimization problem usually takes too much time,which is inappropriate for real-time application.To solve this problem,the nonlinear programming problem is transformed into an input-output fitting problem based on feedforward neutral network in this paper.