Supervisory Control for Co-inject Hot Metal Desulfuration Processes

Abstract

Mg-CaO Co-inject Desulphurization Process is a kind of new technics in which the reagent is directly injected into the hot metal which contains excess sulfur for the purpose of desulphurization by combining melt magnesium reagent with sulfur in the hot metal into MgS which will become the scruff floated on the top of the hot metal, and, finally, the final content of sulfur in the hot metal meets the production restriction. Since the performance of Mg reagent is changeful �� and the content of sulfur in the hot metal can not be obtained in the reagent injection process at real time, the result of desulphurization process is quite uncertain. Therefore, based on conventional loop control, this paper present a novel desulphurization process intelligent supervisory control technique by means of the simulation to domain expert's method to deal with the internal relation between the process data from different tanks, in which the self-adapting setup control mechanism combining a self-learning setup model and a feedback setup compensating model is build for restraining the influence of reagent performance change. The technique has been successfully applied in a plant desulphurization station control system design, and the controlled process has achieved optimization.