Optimal Torque Distribution Towards Power Saving for Distributed-driven Electric Vehicles

This paper presents an optimal torque allocation strategy for distributed-driven electric vehicles based on linear active disturbance rejection control (LADRC) strategy and neural network. A hierarchical control structure is adopted to coordinate the control effort of each part, using a steady-state single-track model to interpret driver’s intention, a closed-loop controller to stabilize yaw motion, and a control allocator to achieve the minimum power consumption of in-wheel motors. LADRC is selected as the yaw rate controller to generate the desired direct yaw moment for the pursuit of excellent control accuracy and strong anti-disturbance performance, while the optimization method is introduced to solve the control allocation problem for the redundantly actuated system. Considering the real-time operation demand, we choose NN to fit the optimal solution based on the off-line data obtained through stochastic pre-training and Co-simulations of Matlab/Simulink and CarSim using experience-replay and soft-update training tricks. HIL experiment verifies the feasibility of the proposed control strategy.