Prediction of Blast Furnace Temperature Based on Improved Extreme Learning Machine

摘要

In iron-making process of blast furnace, temperature is an important indicator that relates closely to working condition. Silicon content in hot metal is one of the main parameters to reflect the temperature inside the blast furnace. By predicting the silicon content in hot metal, the theoretical basis for subsequent parameters adjustment is provided. Aiming at the non-linear feature of silicon content, a prediction method based on improved extreme learning machine is proposed. The improved extreme learning machine uses flower pollinate algorithm to optimize its parameters, and the prediction model of silicon content is constructed by optimized extreme learning machine. Verified with production data, the simulation results show that compared with the basic extreme learning machine, the improved algorithm can speed up the prediction accuracy and generalization ability, and it also has good stability.