Application of an expert system to arc furnace steel-making

The correlation function, a means to quantitatively describe contradictory problems, was introduced into an arc furnace expert system to distinguish if the composition of molten steel was qualified before tapping. The neural network prediction model was used as an evaluation basis of the arc furnace expert system, which was trained by the hybrid genetic algorithm to a high precision. Simulation results show that many reasonable advices can be given to operators in steel-making process by the expert system.