Data-driven vehicle stability control via co-simulation of digital twin and constrained MPC

With the rapid advancement of intelligent connected vehicles and autonomous driving technologies, vehicle lateral stability under high-speed or emergency steering conditions has become a critical safety concern. Traditional control strategies often exhibit limited performance in complex driving scenarios. This study proposes a novel co-simulation framework integrating digital twin technology with constrained Model Predictive Control (MPC) to enhance lateral stability and torque distribution. A high-fidelity digital twin model was constructed to improve real-time accuracy, while an MPC-based controller was designed to optimize handling under extreme conditions. Experimental results demonstrated improved performance: lateral stability error was reduced by 62.5%, and yaw rate error by 57.1%, compared to traditional methods. The key novelty lies in the dynamic, data-driven integration of the digital twin for real-time MPC optimization. These findings provide a robust theoretical foundation and technical support for intelligent vehicle development.