基于样条变换的非线性PLS的反应高炉炉温的参数预测

Parameters Prediction on Furnace Temperature on Blast Furnace Based on a Nonlinear Spline Transform-PLS

高炉冶炼过程是一个大时滞、强非线性的系统,现有的高炉炉温预测模型不够准确,因此,建立了基于香农熵的广义相关系数时滞分析模型和基于样条变换的非线性偏最小二乘回归(ST-PLS)的反应炉温的参数预测模型,得出影响高炉炉温的主要参数的滞后时间,预测出能够综合反应高炉炉温的4个参数([Si],[S],铁还原速率及铁水温度)。试验证明,模型具有较高的预测精度,当相对误差分别为0.11和0.18时,模型预测[Si]的命中率分别为0.714 3和0.918 4,[S]的命中率分别为0.734 7和0.918 4,铁还原速率的命中率分别为0.612 2和0.816 3,铁水温度的命中率分别为1.000 0和1.000 0。

The blast furnace smelting process was a system with large delay and non-linear.The existing blast furnace temperature prediction models were not comprehensive.The practice showed that the temperature prediction by using only the silicon content in hot metal was not accurate and sufficient.Therefore,two models were presented to calculate the lag time of the main parameters and predict the parameters related to furnace temperature（,[S],iron reduction rate and temperature in molten iron）.The delay analysis model was based on the generalized correlation coefficient of Shannon entropy,and the parameters prediction model of the blast furnace temperature was based on a nonlinear spline transform-PLS.Experiments show that the presented model has a high forecast precision.When the relative errors are 0.11 and 0.18,the hit rate of ,[S],iron reduction rate and temperature in molten iron are [0.714 3,0.734 7,0.612 2,1.000 0] and [0.918 4,0.918 4,0.816 3,1.000 0].