Adaptive Optimal Control of Nonlinear Active Suspension Systems with Completely Unknown Dynamics

Ride comfort, tire deflection and suspension stroke are generally conflicting in the active suspension control. To achieve a compromise among the three requirements, a novel online adaptive optimal control method is developed for the nonlinear active suspension systems subject to completely unknown dynamics. Based on the adaptive dynamic programming (ADP) algorithm, this paper proposes an identifier-critic structure with a dual neural network (NN) to realize the active suspension optimal control. To this end, a cost function concerning ride comfort, suspension stroke and control input is constructed. Then, an identifier NN with a novel adaptive law based on the parameter estimation error is used to approximate the unknown nonlinear dynamics. To obtain the optimal solution, a critic NN with the proposed adaptive law is suggested to design the optimal controller. Simulation results are provided to verify the effectiveness of the proposed method.