A Method for Filtering Data-Driven Predictive Control Aimed at Vehicle Lateral Stabilization in Compacted-Snow Rutted Road Conditions

Designing a data-driven predictive controller for lateral stability of vehicles on compacted-snow rutted roads, especially when the data is subjected to non-stationary noise disturbances, is a significant challenge. This paper introduces a novel subspace predictive equation construction method that incorporates filtering using the inverse of a first-order disturbance model. Based on this, a filtering data-driven predictive controller is designed. The proposed method is tested through numerical simulations and MATLAB/CarSim co-simulation. Simulation results show that the proposed subspace predictive equation with filtering effectively mitigates the variance and bias introduced by non-stationary noise. The co-simulation results demonstrate that the controller ensures the stability and robustness of the vehicle system.