Disturbance-Compensation-Based Multilayer Neuroadaptive Control of MIMO Uncertain Nonlinear Systems

Abstract

Committed to further enhancing the achievable tracking performance, a novel disturbance-compensation-based composite multilayer neural network adaptive control algorithm is developed for a class of multiple input multiple output nonlinear systems with modeling uncertainties. Specially, an extended state observer is utilized to estimate the exogenous disturbance and meanwhile predict the system state. Moreover, the nonlinear function uncertainties are approximated by the multilayer neural networks. Furthermore, the modeling uncertainties can be compensated in a feedforward manner. Notably, the multilayer neural network weights are updated via the composite adaption laws driven by the output tracking error and the prediction errors of the system state and control input, which brings improved function approximation performance. Finally, the application results demonstrate the efficacy of the integrated intelligent controller.