UKF-Based Adaptive Dynamic Programming for Optimal Control of Uncertain Quarter-Vehicle Suspension Systems With Input Constraints

Research on uncertain suspension systems plays a crucial role in advancing intelligent vehicles. Despite its importance, the field still faces significant challenges, particularly in developing optimal control strategies for uncertain suspension systems. This article proposes a novel optimal control strategy based on the unscented Kalman filter (UKF) for uncertain nonlinear active quarter-vehicle suspension systems with input constraints. The key innovation lies in embedding the UKF into the iterative adaptive dynamic programming framework, where the UKF not only estimates system states, but also enhances the accuracy of future state prediction during the learning process. This integration enables adaptive updates with improved robustness to uncertainties, offering a new approach that combines stochastic filtering with reinforcement learning. The resulting UKF-based optimal control strategy is further validated through hardware-in-the-loop testing, confirming its effectiveness in real-world conditions.