Converter input-output mixture model based on mechanism and data-driven

To realize the fine control of the energy flow network, a converter input -output model is established based on mechanism and data model. The input and output of the material in the converter process are analyzed. According to the actual production data, the converter smelting relevant parameters, including oxygen utilization rate, slag basicity, magnesium content in slag, converter thermal efficiency, are obtained by mathematical statistics and regression method. By applying the smelting mechanism, the input-output model is established with the conditions of molten iron and scrap of smelting, the target molten steel composition and temperature as the inputs; the information of calculated amount of oxygen and slag as the outputs. The accuracy of the model is improved through revising the mechanism model by comparing the parameters calculated by the mechanism model and the neural network respectively. The model is programmed through the C # language. The results of model calculation show that, in the mixed model, the hit rates of the adding amounts of lime, dolomite and the oxidized pellets increased 11.1 %, 8.3 % and 8.3 % respectively in comparison with the mechanism model in the same error range.