Distributed Electric Drive Vehicle Torque Vectoring Control Considering Driving Style

With the increasing attention to environmental issues,the agreement on the vision of carbon neutrality in the world,and the rapid development of artificial intelligence technology,the automotive industry is undergoing a rapid revolution of electrification,networking,intelligence,and sharing.Compared with the traditional centralized drive electric vehicles,the distributed drive electric vehicles have a short drive chain and high chassis integration,and at the same time,can make full use of the four-wheel adhesion and can respond quickly and execute precisely to the commands of the driverless system,so they gradually become the ideal vehicle for the future Intelligent Connected Vehicle.This paper focuses on the Torque Vectoring control method of distributed drive electric vehicles,and the research topic is "Distributed Electric Drive Vehicle Torque Vectoring Control Considering Driving Style." Since the current yaw movement control of distributed drive electric vehicles mainly focuses on stability control objectives,it cannot realize adaptive control for drivers with different driving styles.Therefore,this paper proposes a Torque Vectoring control strategy considering driving styles,intending to realize personalized and intelligent yaw movement control of distributed drive electric vehicles.The specific contents are as follows:First,a vehicle model that reflects the dynamics characteristics is built based on a four-wheel distributed drive electric vehicle.A joint simulation vehicle model is created based on Carsim-Simulink: the parameter matching of the vehicle body model and chassis model is completed in Carsim,and the wheel motor model and driver model are built in Simulink.After the model was built,the validity and accuracy of the built vehicle model was verified by comparing the simulation with the real vehicle test results under typical conditions such as steady-state steering and serpentine obstacle avoidance test.Secondly,the real-time recognition of driving style was achieved.To collect driver and vehicle state data sets and subsequently verify the real-time and effectiveness of the proposed control strategy,this paper builds a driver-in-the-loop and hardware-inthe-loop driving simulator,and completes the data collection and event classification of drivers of different ages and experience under the composite track conditions that test the vehicle handling stability.The drivers are clustered by style based on various statistical features in the driving events,and the offline and real-time online recognition of driving styles is achieved using BP neural networks.Hardware-in-the-loop testing shows that the recognition algorithm can achieve high recognition accuracy under compound road conditions and ensure good real-time performance.Finally,a Direct Yaw Moment Control strategy is proposed to adapt to different styles of drivers.First,the ideal reference model and control objectives for different driver styles are quantified by further analysis of the 2DOF bicycle model and the clustering results of driving styles;then,the control strategy is further determined,and the stability-based control objectives are adopted for conservative/ moderate driving styles;the control objectives based on professional driver characteristics are adopted for aggressive/ sporty drivers.Subsequently,a hierarchical Direct Yaw Moment Control scheme was built based on sliding mode control.Finally,the steering wheel sinusoidal input simulation,the variable speed Double Lane Change simulation,and the simultaneous driver and controller Hardware In-Loop tests were completed,respectively.The results all show that the controller can quickly switch the control target and respond after recognizing the driving style,guaranteeing the stability of conservative and normal drivers during cornering and also satisfying the demand of aggressive style drivers for path following and high yaw rate gain,reducing the driver’s operational burden.