Neural network adaptive robust control of nonlinear systems in semi-strict feedback form (Q5943835)

Neural networks (NN) and an adaptive robust control design philosophy have been integrated to design performance oriented control laws for a class of nonlinear systems in the semi-strict feedback form. In a sense this is the generalization of recently proposed NN adaptive robust control. Both, repeatable unknown nonlinearities and nonrepeatable unknown nonlinearities, such as external disturbances, are considered. All unknown but repeatable nonlinear functions are approximated by outputs of multi-layer NNs. All NN weights are tuned on-line with no prior training needed. Certain robust control terms can then be constructed to achieve a guaranteed output tracking transient performance. The proposed strategy is then applied to the precision motion control of linear motor drive systems with nonnegligible electrical dynamics. The analysis of experimental results shows a good performance of the method. But additional spikes in the position error of the motor are observed caused by Coulomb friction. It is not surprising to see these spikes since NNs can only approximate continuous functions and are not able to handle discontinuities, like Coulomb friction, well. But this problem should require another analysis that is not performed here.