Neural Based Adaptive Control of a Class of Dynamical Nonlinear Processes

A nonlinear adaptive controller for a class of nonlinear plants with incompletely known and time varying dynamics is presented. It is based on a recurrent neural network used as a dynamical model of the plant. The adaptive controller design is realized by using an input-output feedback linearizing technique. The model parameters, that is the controller parameters are updated on-line such that the behaviour of closed loop system is closely to those of a linear system. A local convergence of the algorithm is provided for the case of constant reference output. Computer simulations are included to illustrate the performances of the proposed controller. Key-Words: Nonlinear systems, Nonlinear control, Adaptive control, Neural networks.