Learning in and performance of the new neural network based adaptive backthrough control structure

Abstract

This paper presents a new neural network (NN) based Adaptive Rackthrough Control (ABC) scheme for both linear and nonlinear dynamic plants. Unlike other feedforward NN based control schemes the ABC proposed here is comprised of one neural network only that simultaneously acts as both - plant model and the controller (plant inverse). For linear noiseless plants, the resulting feedforward controller, for equal orders of the plant and plant model, is a perfect adaptive poles-zeros canceller. In the case of monotonia nonlinear dynamic system, the proposed ABC control represents the nonlinear predictive controller. The ABC scheme is based on the discrete nonlinear (NARMAX) dynamic model. For nionotonic nunliuearities, the desired control signal results from the nonlinear optimization procedure with guaranteed convex search function and consequently with a unique solution.