Nonlinear black-box modeling of electric arc furnace: an application of fuzzy logic systems

Abstract

Presents the application of fuzzy logic systems and adaptive fuzzy logic systems to model electric arc furnaces. The main objectives are to provide the rationale and to justify the use of fuzzy modeling for electric furnaces. To this end, the principles of fuzzy logic systems are described briefly, and justifications for application of fuzzy systems for modeling are provided. This is done with reference to three important properties of fuzzy logic systems, namely, nonlinear black-box modeling capability, universal approximation ability and functional equivalence with radial basis function networks. Finally, the detailed investigation regarding the applications of both classic and adaptive fuzzy logic systems to electric arc furnace modeling is presented. It is demonstrated that the application of adaptive fuzzy logic systems as a non-conventional system identification method to model nonlinear systems can be considered as an alternative to artificial neural networks. The proposed modeling methods are described, and their use is illustrated using the actual recorded data.