Research on the Forecast Model of the Boron Removal from Metallurgical Grade Silicon by Slag Refining Based on GA-BP Neural Network

摘要

A purification process was developed to removal impurity element boron from metallurgical grade silicon by electromagnetic induction slag melting. Vacuum melting furnace was used to purify boron in both Al_2O_3-MgO-CaO-SiO_2 slag system and Al_2O_3-CaO-SiO_2 slag system. The relationship between different slag chemistry and the removal of boron in silicon were studied using Back Propagation (BP) Neural Network model. The best slag chemistry for the removal of boron was predicted by Genetic Algorithm (GA) contributed by the use of Matlab. The results show that the mass fraction of Boron in silicon is reduced from 11.7496×10~(-6) to 2.3259×10~(-6) after slag melting in 28.96%Al_2O_3-3.43%MgO-36.24%CaO-31.37%SiO_2 slag system. The relative error obtained with GA-BP Neural Network model was below 0.35%.