Longitudinal Velocity and Slip Rate Control for Distributed Electric Vehicles Via Learning-based MPC

This paper proposes a learning-based model predictive control (MPC) framework for integrated longitudinal velocity tracking and slip rate control in distributed electric vehicles. A deep neural network (DNN) is embedded to dynamically compensate for unmodeled nonlinear tire dynamics, enhancing prediction accuracy. The proposed learning-based MPC optimizes speed tracking and slip constraints through rolling horizon optimization, with model mismatch errors learned from driving data. Validation via CarSim/Simulink co-simulation demonstrates that the framework achieves a 62.8% reduction in speed tracking error under low-traction conditions, eliminates overshoot, and improves steady-state acceleration.