Smelting condition identification for a fused magnesium furnace based on an acoustic signal

To promote energy efficiency during fused magnesium furnace smelting, four smelting states were introduced in the smelting stage: an unmelted state, semi–molten state, molten state, and overheating state. A smelting identification system to distinguish these smelting states was developed through the use of linear predictive coding and a principal component analysis algorithm. A new smelting condition identification system was obtained. Corresponding pilot productions were conducted to compare the differences between employing the method and not employing the method. All of the pilot production data showed that feeding raw materials over time during the overheating state and decreasing current injection in the molten state could reduce energy consumption as well as increase crystal purity.