Adaptive Neural Command Filtered Backstepping Control of Uncertain Nonlinear Systems Subject to Bouc-Wen Hysteresis Input

An adaptive neural command filtered backstepping (CFB) approach is proposed for uncertain nonlinear systems subject to the Bouc-Wen hysteresis input. First, radial basis function neural networks (RBFNNs) are invoked to model the unknown nonlinear dynamics and then command filtering is employed to eliminate the “explosion of terms” problem that exists in the backstepping approach. Because of using command filtering, challenges in choosing time constant of the filters in the dynamic surface control (DSC) approach is eliminated and filtering errors are compensated. To handle the unknown control direction obtained because of the hysteresis nonlinearity, Nussbaum function is employed. Furthermore, number of adaptive parameters and online computation are reduced considerably. It is proved that all signals in the closed-loop system are bounded and tracking error converges to the vicinity of zero. Simulation results verified the efficiency of the proposed method.