Research On Prediction Model For End Phosphorus Content During Dephosphorization By Multi-Refining Converter

Phosphorous is harmful to most kinds of steels. The increase of phosphorous content will decrease the ductility and toughness of the steel, especially with the increase of carbon content in steel, this phenomenon is usually called "cold brittleness". Therefore to produce high carbon steels, it must reduce the phosphorous content in steels. Accurate control of end phosphorus content has significant meaning to improve the quality of liquid steel and reduce costs. This paper mainly research on dephosphorization technical process by Multi-Refining Converter, to set up prediction model for end phosphorus content during dephosphorization.Dephosphorization reaction process in converter steelmaking is complex because of the numerous effect factors, so the correct selection of the factors which have important effect on dephosphorization will improve the accuracy of prediction model for end phosphorus content during dephosphorization. The paper which considers mechanism aspect about converter dephosphorization, analyzes influence factors about dephosphorization, adopts Grey Relating Analysis Theory in Grey System Theory to calculate the correlation degree, and finally gets the Grey Relating Serial about phosphorous content, which provides theory foundation for converter dephosphorization end point controlling. To solve the default data and yawp data in measuring process by Multi-Refining Cenverter, The paper not only calculates default data by Grey Relating Analysis Theory, but also testing the yawp data by this method.According smelt techniques and locale data, the paper makes research on forecast method about converter smelt end point phosphorous content, analyzes factors how to influence phosphorous content, makes sure controlling variable, and sets up the forecast GM(0,N) Model based Grey System Theory to predict phosphorous content correctly. The paper also build Multi linear regression model to compare with the Grey forecast model.For enhance calculated precision of Forecast Model, the paper improves it from two sides: First, using Clustering method to pretreatment data for the Model, then using the pretreatment data to build Forecast Model; second, to prevent'data saturation'phenomenon when estimate parameters using the least square estimation arithmetic, this paper adopts identification method combines forget factor with restricted memory length calculate parameters.Using improved Prediction Model for End Phosphorus Content during Dephosphorization by Multi-Functional Refining in Converter to forecast, when windage range is in±0.02%, the Forecast Model's hit rate is 96%, when windage range is in±0.015%, the Forecast Model's hit rate is 84%, that illuminates this Forecast Model's predictive result is satisfying.