Optimal Torque Distribution Control of Multi-Axle Electric Vehicles with In-wheel Motors Based on DDPG Algorithm

Abstract

In order to effectively reduce the energy consumption of the vehicle, an optimal torque distribution control for multi-axle electric vehicles (EVs) with in-wheel motors is proposed. By analyzing the steering dynamics, the formulas of additional steering resistance are given. Aiming at the multidimensional continuous system that cannot be solved by traditional optimization methods, the deep deterministic policy gradient (DDPG) algorithm for deep reinforcement learning is adopted. Each wheel speed and deflection angle are selected as the state, the distribution ratio of drive torque is the optimized action and the state of charge (SOC) is the reward. After completing a large number of training for vehicle model, the algorithm is verified under conventional steering and extreme steering conditions. The maximum SOC decline of the vehicle can be reduced by about 5% under conventional steering conditions based on the motor efficiency mapused. The combination of artificial intelligence technology and actual situation provides an innovative solution to the optimization problem of the multidimensional state input and the continuous action output related to vehicles or similar complex systems.

Keywords

• electric vehicles (evs)
• independent-drive technology
• deep reinforcement learning (drl)
• optimal torque distribution