Adaptive Neural-Based Backstepping Control of Uncertain MIMO Nonlinear Systems

It is proposed an approach for adaptive neural-based backstepping control for uncertain MIMO nonlinear systems that uses two neural networks in each backstepping design step. This leads to a more straightforward implementation when compared to methodologies that employ just one NN in each design step, as the neural networks inputs here do not depend on derivatives of the virtual control laws. Furthermore, it is verified that the total number of NN´s necessary to obtain an adequate tracking response is significantly reduced. Semiglobal uniform ultimate boundedness of all the signals in the closed loop of the MIMO nonlinear system is achieved and all the outputs converge to small neighborhoods of the desired reference trajectories.