Gaussian Process Regression Mechanical Characterization Model and Interval Type-2 Fuzzy Smith Time-Delay Compensation Control for a Magnetorheological Semi-Active Suspension

The control precision of semi-active suspension systems equipped with magnetorheological (MR) dampers is significantly affected by time-delay effects and the accuracy of mechanical characteristic modeling. To address this, a double-loop control architecture is developed, integrating an outer-loop predictive force control module for time-delay compensation and an inner-loop precise excitation current calculation module. A novel Gaussian process regression (GPR)-based mechanical characterization model is constructed by designing composite covariance functions and optimizing hyperparameters through data-driven methods. For the outer-loop control, an interval type-2 fuzzy Smith time-delay compensation strategy is proposed, which utilizes vehicle body vertical acceleration, suspension deflection, and tire displacement as state inputs to dynamically capture and respond to vehicle motion variations, thereby enhancing system anti-interference capability. The inner-loop control leverages the GPR model for accurate excitation current computation. Simulation results validate that the proposed double-loop controlled MR semi-active suspension system demonstrates superior control efficacy and robustness, effectively adapting to complex and variable driving conditions.