Developing electric vehicles has become a common solution for the energy security problem and air pollution. In the field of electric vehicles, distributed drive is considered to be one of the cutting-edge technologies. It has advantages of flexible control, fast response and has shown great potential in terms of dynamic performance and energy saving, at the same time it has also brought some hotspot problems, such as torque distribution between drive wheels and dynamic control of each drive wheel. This research focuses on energy efficiency and dynamic response optimization of distributed drive electric vehicles. In terms of optimizing energy efficiency, most of existing researches for torque distribution are based on models, which means the control algorithm needs either accurate models of target vehicle or large quantities of calibration data. In this research, a torque distribution method based on golden ratio search algorithm is adopted, which can realize optimized torque distribution between front and rear axles while don’t need any target models. However, the main problem of this method is torque ripple in the process of searching. In order to solve this problem, an automatic calibration method of optimized torque distribution is proposed, which can solve the torque ripple problem successfully. After that, optimized torque distribution between four wheels when vehicle is turning is also studied, and its energy saving potential comparing with front and rear axle torque distribution method is analyzed. A joint simulation platform between Matlab/Simulink and Carsim has been set up to verify the function of automatic calibration method of optimized torque distribution. Results show that with the proposed torque distribution method, up to 12.5% of energy can be saved in NEDC cycle and 7.4% of energy can be saved in China’s typical cycle of city passenger cars. To further verify the feasibility of the proposed torque control methods in real application, two kinds of HIL (hardware in the loop) simulation platforms are set up. One is based on Matlab xPC platform, the other is based on communication between two controllers. With automatic code generation technology, algorithm models in Simulink are translated into executable code and downloaded to KPV13 rapid prototyping controller. The controller is tested in both of the two HIL simulation platforms. Results show that operation and communication speed of real control systems can satisfy the requirement of real-time search algorithm, the proposed control method can be applied in real control systems.
Keywords distributed electric driving; on-board searching method; torque optimal split; traction motor efficiency model; co-simulation