Abstract: A discussion is devoted to designing neur al network adapt ive nonlinear dynamic inversion controllers according to the feedback linearization theor y. Firstly, with two-time-scale, a design of a fast loop controller and aslow loop one is given using the dynamic inversion method. Secondly, a model inversion based neural network non-linear direct adaptive scheme is present ed, in which an online neural network is designed to compensate for the modelinversion er rors. It is demonstrated by the distr ibuted simulation of a local area network that this adaptive controldesign is of better adaptability and robustness.

Key words: model inversion, neural network, adaptive control