Singularity-Free Adaptive Backstepping Design for Strict-Feedback Systems using Neural Networks

A switching-type neural network (NN) based adaptive backstepping control design is presented for the tracking tasks of strict-feedback systems. It consists of four parts in each virtual control design step: a one-layer NN for approximating the unknown nonlinearity to render the adaptive control applicable; a certainty-equivalence adaptive controller for compensating the resembled nonlinearities; a high-gain controller which takes over temporarily once the former is approaching singularity; last, a nonlinear damping component for counteracting the degradation due to the approximation errors. Among others, it has the distinct features of requiring minimal prior knowledge of the unknown nonlinearities, less control effort, and relatively simple control structure.