Neural minimal learning backstepping control of stochastic active suspension systems with hydraulic actuator saturation

This study develops a novel radial basis function (RBF) neural network (NN) minimal learning backstepping control law for active suspension systems represented by nonlinear dynamics with additive stochastic terms and practical hydraulic actuator saturation. The suggested approach is equipped with an adaptive mechanism to cope with the mismatched uncertainties and a proper adaptation law is designed through the so-called minimal learning and the command filter techniques. By utilizing the concept of the bounded-in-probability Lyapunov stability, the backstepping controller is resilient dealing with stochastic disturbances arisen form un-modeled dynamics of active suspension systems. Moreover, the mean-value theorem is considered for handling the practical issue of the hydraulic actuation saturation issue. The presented robust adaptive control scheme theoretically assures that the output tracks a time-varying desired reference with a pre-determined small error. Eventually, a closed-loop active suspension plant with the hydraulic actuator is simulated numerically to show the advantages and performance of the developed controller over the state-of-the-art methods.