Improved electric arc furnace operation via implementation of a geno-fuzzy control system

The U.S. steel industry is growingly increasingly interested in the process of slag foaming in their electric arc furnace plants. Although slag foaming has been shown to improve plant efficiency, this highly dynamic process can he very difficult to consistently control. This article describes a computer control system developed to effectively manipulate the slag foaming process, and the implementation of the controller in an electric arc furnace plant. The control system is a model-following controller based on fuzzy logic. It uses a neural network to simulate the slag foaming process. Furthermore, the control system uses an evolutionary algorithm to effectively tune its fuzzy rule base in response to the dynamic behavior of the slag foaming process. Results are presented, which demonstrate the effectiveness of the control system in this process characterized by relatively slow process dynamics.