Simulation Research on Vehicle Active Suspension System Based on Model Predictive Control

Abstract

Vehicle suspension systems are essential for ensuring both handling stability and passenger comfort. Conventional passive suspensions exhibit deficiencies when adapting to intricate and fluctuating road surfaces. While conventional control techniques such as LQR and PID are prevalent, they fall short in managing the mechanical limitations and actuator physical constraints. In this research, a quarter-vehicle active suspension mathematical model with two degrees of freedom (2-DOF) is developed, and a Model Predictive Control (MPC) approach is introduced. By strictly adhering to actuator force and suspension working space (SWS) limitations, the MPC controller is engineered to optimize the acceleration of the vehicle body. MATLAB/Simulink is utilized to perform simulations under both random and bump road excitations. In addition to conventional time-domain indicators, this research integrates a frequency-domain assessment aligned with ISO 2631 standards to thoroughly evaluate ride comfort. Moreover, an evaluation of computational complexity via a software-in-the-loop (SIL) setup on a real-time processor, alongside a weight sensitivity analysis, is performed. The findings reveal that the introduced MPC controller surpasses both the passive suspension and the traditional PID controller, markedly decreasing the peak body acceleration, mitigating resonance within the sensitive 4-8 Hz frequency range, and displaying enhanced robustness. With an average solving time of 1.2 milliseconds per step, its viability for real-time application is validated. This investigation establishes a robust theoretical foundation for the practical engineering implementation of active suspension systems.