Dynamic Neural Adaptive Control Schemes for Linear and Nonlinear Systems

In this paper we describe three adaptive control schemes using the model of a neuron, called the dynamic neural unit (DNU), as the basic functioning element. In the first adaptive control scheme, the DNU is used in inverse-control mode, where the dynamic structure of DNU is made approximately an inverse-model of the plant under control making the overall system transfer function almost unity. In the second adaptive control scheme, called dynamic neuro-controller, a multi-stage neural network is developed using DNU as the basic processing node. This controller is effectively used to control unknown nonlinear systems. The third adaptive control scheme makes use of the combination of a linear feedback controller and the DNU as a feedforward controller. The effectiveness of these adaptive control schemes is demonstrated through computer simulation studies.