Research On Sulfur Content Prediction Model For The End Of LF Refining

In the metallurgical industry,the sulfur elemental on most of the steel is a kind of harmful clements,it will affect the steel for high strength,low temperature toughness and weldability,so we must strictly control the content of sulfur in steel.At present domestic hot metal desulphurization technology is relatively backward.mostly used human experience to control the smelting process.As a result,the desulphurizing powdeer consumption too much,economic benefit is also not ideal.With the development of society,the steel industry need the quality of iron is higher and higher.Relying on experience to control the smelting process has been unable to adapt to the reform of the production process.In order to realize hot metal pretreatment process automation,fast rhythm,the demand for efficient production.in this paper used the mechanism model,the BP neunral.the RBF neural network and the hybrid model with the RBF neural network and the traditional mechanism model to study the LF desulphurizing model,and provide the effective technical for the liquid stcel refining desulturation.Based on practical production data as the basis,and established the model to predict the final sulfur content of LF refiningu,the main work is as follows:Processing the actual data and choose the suitable training sample.Part as the training sample part as the predicted sample data.To understand the mechanism of traditional desulfurization process.Analysis of factors affecting the desulfurization effect The thermodynamic principle and dynamic principle of desulfurization reaction are introduced.Determine some desulfurization related parameters.Mechanism model for prediction of sulfur content in the steel.Aiming at the problems and shortages of the mechanism prediction model,In this paper,a neural network model is used to forecast the sulfur content.By analyzing and comparing the prediction results of RBF and BP neural network prediction model,we can know that the RBF neural network forecasting model is better than the BP neural network prediction model from the prediction speed and effect.So in this paper,a hybrid model based on RBF neural network model and mechanism model is established,which is superior to the single neural network model in theory.Through C#and MATLAB language to simulate various forecasting model.Simulation results show that the hybrid prediction model for LF refining sulfur end-point content prediction accuracy rate is very high,Prediction accuracy rate is over 90%,and other model forecast accuracy rate is below 70%.It is concluded that the mixed model can provide technical support for the production of LF furnace refining desulfurization.