Neural network control for non-affine nonlinear systems

Recently, adaptive neural control has been attracting an increasing attention for nonlinear unknown dynamic systems [1][2]. This paper is dedicated to the discussions on a few techniques in the design of adaptive neural network control for non-affine systems which are known to be difficult to control. The techniques include implicit function theorem based neural control for classes of the non-affine systems in Brunovsky form, implicit function theorem with backstepping design for classes of the non-affine systems in pure-feedback form, and pseudo inverse control. This paper is aimed to provide an overview of the state of art of stable control design for non-affine systems using neural network parametrization, and to list the advantages and disadvantages of neural network control.