基于偏最小二乘回归的高炉焦比影响因素分析

Analysis on Impact Factors of Coke Ratio in Blast Furnace Using Partial Least Squares Regressive Model

影响高炉焦比的基本因素之间通常存在显著的相关性,常规最小二乘回归法估计回归系数存在较大的偏差甚至出现反常现象。偏最小二乘法能够克服自变量集合多重相关性对回归建模的不良影响,提高模型可靠性。基于偏最小二乘回归法建立了10成分模型,通过比较标准回归系数分析各因素对焦比的影响程度和自变量与因变量的正、负相关性,调整主要因素,使焦比接近或达到最优值。

Since the marked relativity usually exists among the impact factors of coke ratio in BF, it will induce excessive computational error and so much as abnormality to estimate coefficients with the ordinary least square regression. Partial least squares can overcome the adverse effects of muhicoUinearity to regression modeling and improve modeling effectively. Based on partial least squares, the model for 10 kinds of composition is built. The effect degree of each factor on coke ratio and the positive or negative rclativities are analyzed according to calculated standard regression coefficient. By adjusting main factors, the coke ratio is as close as or even equal to the optimal value.