Based on the characteristics of composition strong coupling with physical properties of smelting slag, the temperature and density of slag system were selected as the secondary variables of soft sensor. By applying the regression algorithm with Taylor series three times expansion, a composition soft-sensing model of the ternary slag system was established based on two-temperature two-density method. The experimental results of slag system of FeO-Fe2O3-SiO2 show that regression calculation enjoys characteristics of high accuracy. The average relative error (absolute value) of density is only 0.412%. Multi-group soft measurements achieve superior predictive hit rate, which shows that the model is suitable for compositions of the slag system with wide range. Analysis of the density-temperature characteristic curves shows that the model reasonably takes account of the inherent law of density, composition and temperature of the ternary slag system. The research serves as a nice soft-sensing model for predicting compositions of the ternary slag system.%针对火法冶炼熔渣成分含量与其物理性质强耦合的特点，以渣系温度和密度为软测量辅助变量，采用泰勒级数三次展开回归算法，建立基于两温两密度法的三元渣系成分软测量数学模型. FeO-Fe2O3-SiO2渣系的实例验证表明，回归计算精度高，平均相对误差（绝对值）仅为0.412%，多组软测量预测命中率高，且渣系组分含量适应范围较宽.依此模型计算绘制的密度-温度特性曲线分析表明，该模型能较好地反映三元渣系密度与组分含量、温度间的内在规律，为实现三元渣系组分含量预测提供了良好的软测量模型.
展开▼
