Research On Effect Of K₂TiF₆ On Fe Reduction In The Process Of Magnesium Smelting

The corrosion resistance of magnesium and its alloy can be deteriorated by theharmful element Fe. In order to improve the corrosion resistance of magnesium andits alloy, the product standard strictly define the content of the impurity Fe. Thethermodynamic feasibility of K₂TiF₆ on Fe reduction was calculated in the process ofmagnesium smelting. The influence factors of Fe reduction, such as, the content, andthe temperature of K₂TiF₆ being added and the holding time were studied anddiscussed. The principal degree of the influence factors was analyzed by the means ofrange analysis. The model was established by the way of regression analysis andartificial neural network. Finally, using the perceptron principle, the experiment datawas optimized. The conclusions are as follows:(1) The results of the thermodynamic calculation show that K₂TiF₆ on Fe reduction in the processof magnesium smelting is feasible.(2)The experiments show that reducing the content of MgO in the melt of magnesium,can improve the effect of reducing Fe.(3)The contribution of the content and temperature of K₂TiF₆ being added and theholding time is markedly. The content of Fe reduced quickly before the holding timeas to 20min, while the content of Fe reduced relatively slowly when the holding timeis longer than 20min. Increasing the temperature, the Fe content reduce. With theaddition of K₂TiF₆ increased, the content of Fe reduce remarkably. But when theaddition reach to 0.3%, the content of Fe is not reduced obviously.(4)The content of Fe in magnesium melt can be reduced from 0.028% to 0.004% when0.3⁰.4% K₂TiF₆ is added, which is counted by pure titanium to magnesium melt, and thetemperature is 780℃, holding time 20min respectively.(5)Analyzing the experimental data by range analysis, the result shows that the sequence of theweightiness of the Fe reducing effect is temperature, addition and holding time.(6)Model was established by the way of multivariant linearity regression and BPalgorithm of artificial neutral network. The model which was established with BPalgorithm of artificial neutral network, is better and its forecasting is more accurate.The perceptron principle of artificial neutral was used for classifying the data. Marking out the line of classification of reducing Fe. And then the accuracy of the line of classification was demonstrated by experiment.