ABSTRACT
Extensive progress has been made in the past half decade or so in the use of neural networks techniques for control [4,7], Neural networks offer the ability to approximate nonlinear mappings, and thus, to model nonlinear system dynamics, which is the feature to be most readily exploited in the synthesis of nonlinear controllers. Presented in this paper is a practical perspective on the use of neural networks for control, which the authors have obtained from the operating experience gained out of numerous neural network based Intelligent Arc Furnace™ Controller installations worldwide. A detailed discussion of related issues arising out of such real-world installations has also been provided.
