Research On Prediction Of End Phosphorus Content For Converter Refining Ferrochrome

Phosphorus is trace element containing in ferrochrome alloy products, it is one of the harmful impurities and one of the important factors affecting product quality. In recent years, the purity of chrome ore and coke raw material used in medium-low carbon ferrochrome product dropped, phosphorus content was high, as a result, the phosphorus content of ferrochrome product had a tendency to increase obviously. At present, the detection of end phosphorus in ferrochrome smelting is still by artificial means, efficiency lags behind, which is not benefit for real time production guiding. Therefore, the research on prediction method of end phosphorus content for converter refining medium-low carbon ferrochrome has important significance to improve the production process and production efficiency.Converter smelting ferrochrome is a complex physical and chemical process, there are many factors affecting end phosphorus content. Important factors affecting end phosphorus content were analyzed and input variables of neural network were fixed by researching the reaction characteristics of converter smelting low carbon ferrochrome and thermodynamics and kinetics condition of hot metal dephosphorizing pretreatment. According to production process and sample data of JiTie liaoyang steel co.,807#converter smelting medium-low carbon ferrochrome, a prediction model was established based on RBF neural network, which shooting rate was 95% within absolute error ±0.003%, the precision was high, more statistical parameters of measurement and management departments were involved, the model structure still need improve. Therefore, besides mechanism analyses, grey correlation analysis method was used for quantitative analysis of the influence factors. The influencing degree of every factor was quantitatively analyzed by grey correlation analysis method, grey incidence degree sequence was fixed, the most important factors were fixed and network structure was simplified and production cycle was shortened. BP neural network was trained by LM algorithm, prediction model of end phosphorus content was established based on LMBP neural network, the simulation result showed that, the shooting rate was 90% within absolute error±0.003%, production requirement was met. In the actual production process we can choose reasonable forecast system platform according to enterprise production management and product quality requirements. Operation interface of end phosphorus content prediction system was designed by C# and MATLAB programming, the software design of end phosphorus content was established. Online prediction of end phosphorus content for converter smelting ferrochrome was realized, which providing theory basis for improving production process and product quality. Through online prediction, product quality was guaranteed. It provides a theoretical basis to enhance automation of production, and conducive to enterprise integration of production management, promoting energy efficiency, enhancing the competitiveness of the products.