Abstract: In order to predict more accurately the phosphorus content of molten steel at the end of dephosphorization converter smelting，DC04 steel was selected as the research object by smelting in a steel company. According to the process parameters of hot metal conditions, slagging materials, oxygen blowing amount and the like obtained by industrial tests, the grey correlation degree of each process parameter on the end-point phosphorus content of the dephosphorization converter was obtained by using the grey correlation analysis method, and a prediction model on the end-point phosphorus content of the dephosphorization converter smelting was established by combining the BP neural network algorithm. Through continuous optimization of the model, when the end point w(P) error of dephosphorization converter was ±0.004 %, ±0.006 % and ±0.008 %, the hit rate reached 83.33 %, 90.00 % and 93.33 %, respectively. The application of this model in the field can provide technical reference for iron and steel enterprises to determine the end point phosphorus content more accurately and quickly.

Key words: DC04 steel, grey correlation analysis, phosphorus content, neural network