Abstract
Optimizing the control algorithm for distributed drive electric vehicles (DDEVs) is challenging due to the numerous parameters involved and the variability of optimal parameters across different vehicle operating states. Traditional expert-based methods fall short, and the extensive state space of vehicle operations adds complexity to the optimization process. This paper proposes a novel approach using minimum unit encoded neural networks to optimize vehicle control parameters. By reconstructing the data space of typical operating conditions and implementing Monte Carlo minimum unit encoding within the total operating condition space, this method significantly enhances the optimization efficiency of weight coefficients for multi-motor control systems. Our results show that the proposed algorithm outperforms traditional methods, achieving superior optimization across all operating conditions. Simulation results indicate that the vehicle’s longitudinal velocity error is reduced to less than 1.2%, the sideslip angle remains below 8°, and the tire slip ratio is kept under 0.08%.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call