Research On Optimal Ore Blending Of Low-Silicon Sintering Driven By Data

After the"Double Carbon"plan was proposed,domestic steel companies have been on the way.Low silicon sintering is the key direction for future sinter development.However,the low w（SiO₂）content can cause problems such as reduced drum strength and deterioration of low temperature reduction.Based on the current ore allocation scheme of a domestic steel plant,a data-driven optimized ore allocation experiment and model study for low-silica sinter was conducted with the aim of exploring the optimal process parameters and ore allocation scheme for improving the quality of low-silica sinter.The combination of various ore powders in the low-silica sintering allocation scheme was explored.The results show that the assimilation,fluidity and bonding phase strength of the bacal powder are low,while the Brazilian blended powder,Mac powder and super special powder have better sintering characteristics.The sintering performance of the mineral powders can be improved by appropriately increasing their proportion of ore mixes in reasonable combination with baka.Through low-Si sintering experiments,the results showed that the SiO2 content of sintered ore decreased from 5.60%to 4.30%with the increase of bacca ratio,the drum index and finished product were rising first and then sharply decreasing,while the RDI+3.15 and RI were as low as below 70%,and the quality of sintered ore under low-Si conditions deteriorated significantly.By adjusting the alkalinity and MgO content to optimize the sintering,the sinter drum index and yield reached the optimum values of 72.30%and 82.64%with RDI+3.15 and RI of 73.53%and 81.46%,respectively,at an alkalinity of 2.2 and MgO content of 2.2%,which greatly improved the quality of the sintered ore.Establishing a data-driven intelligent recommendation model for low-silica sintered ore allocation.,mix composition（TFe,CaO,MgO,SiO₂,Al₂O₃）,alkalinity,mix moisture,ignition temperature,material thickness,negative sintering pressure,sintering time and vertical sintering speed as inputs,and the utilisation factor,yield,return rate,drum index,softening interval,RDI_+3.15 and RI as predictors,The nonlinear linkage was established by XGBoost algorithm with errors between 5%.An intelligent recommendation model for low-Si sintering raw material ratio and sintering process parameters was established based on the prediction model,which led to a significant improvement in sinter performance when comparing the output sinter performance data of the same period.Figure 55;Table 51;Reference 58...