Koopman-MPC-based energy-efficient integrated control of attitude maneuver and vibration suppression for nonlinear in-wheel motor-active suspension on uneven roads

When in-wheel motor-driven electric vehicles (IWMD-EVs) operate on uneven roads, the ride comfort and stability deteriorate significantly, which also results in increased energy consumption. Model predictive control (MPC) is an effective approach for balancing these performance aspects. However, strong tire-suspension nonlinearities add significant computational burden. To address this, a Koopman-based MPC approach is proposed to improve comfort, ensure stability while reducing energy consumption and meeting real-time requirements. First, an 11-DOF nonlinear vehicle model with dynamic damping is established to improve prediction accuracy. Next, the Koopman characteristic function of the finite-dimensional linear system is approximated using extended dynamic mode decomposition (EDMD) and sparse identification of nonlinear dynamics (SINDy), which use as few characteristic functions as possible to more accurately capture nonlinearities. The advantages of this approach include reduced model complexity, improved accuracy, and a low risk of overfitting. Then, a Koopman-based globally linear MPC controller is designed to achieve multi-objective optimization by adjusting weights and to improve controller’s solving speed. Objectives, including simultaneously reducing vertical vibration, pitch/roll motion, and energy consumption, are balanced by active suspension forces. Constraints on suspension deflection, tire dynamic load, and actuator limits ensure stability. Finally, CarSim-MATLAB/Simulink co-simulation and Hardware-in-the-Loop (HIL) testing validate the controller, confirming improved ride comfort, ensured stability, and real-time feasibility. The energy consumption of the active suspension force can be reduced by up to 69.68% compared to the NMPC method on B-class roads.