Model Predictive Control of Vehicle Stability Using Differential Driving Torque

Abstract

Electric vehicles (EVs) with distributed drive configurations demonstrate improved energy storage potential through battery-dominated systems, enabling independent torque allocation across individual wheels. This paper proposes a differential torque control framework for distributed-drive electric vehicles to enhance trajectory tracking accuracy and yaw stability during double-lane change maneuvers. A hierarchical control architecture with three layers are developed, integrating model predictive control with quadratic programming-based torque allocation to coordinate longitudinal velocity tracking and lateral path following. The lateral controller generates real-time differential torque commands (front-rear axle torque variation range: $\pm 282.68$–$\pm 409.42N·m$) through a 3-DOF vehicle dynamic model, while the longitudinal controller maintains speed errors below 0.1 m/s through four-wheel independent torque regulation. Co-simulation on the CarSim-Simulink platform demonstrates the controller's adaptability to road friction coefficients ($\mu =0.5,0.8$) and speed conditions ($u=40,50,60$ km/h). The results achieve maximum yaw rate stabilization at 0.38 rad/s during high-speed maneuvers. Simulation results reveal that despite lateral deviation amplification (80–160 m trajectory segments) and torque oscillation divergence under $\mu =0.5$, $u=60$ km/h, the control system maintains vehicle stability through adaptive yaw moment compensation. The proposed control strategy can adjust the vehicle speed in real time according to road curvature, thereby improving the accuracy of path tracking and driving stability.