基于多元线性回归的焦炭强度预测研究

Prediction of Coke Strength Based on Multiple Linear Regression Analysis

选取配合煤的主要质量指标,运用多元线性回归法建立预测方程,然后对预测方程的回归系数做t检验,剔除对焦炭强度影响不显著的指标,重新做多元线性回归分析,得出简单、准确的预测方程。应用检验结果表明,所得方程回归效果良好,可以较好地预测焦炭强度,预测值的相对误差在5%以内。

Coke strength was predicted by selecting the indexes of blending coal quality. The indexes were used as independent variables for multiple linear regression analysis and got regression models. Then took regression coefficient to make t test, eliminated the no significant index and renewed to make the linear regression models. The research indicated that the model can be used to effectively predict coke strength and the relative error forecasting results was within 5%.